{"meta":{"query_hash":"72fb47ad8ea2","filters":{"category":"bibliometrics"},"cohort_total":923,"direct_labels_cover":923,"predictions_cover":923,"exported":923,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/72fb47ad8ea2","api":"https://metacan.xera.ac/api/v1/cohort?category=bibliometrics"},"results":[{"id":"W100260889","doi":"10.5683/sp2/34sn5w","title":"Dramatic Growth of Open Access March 31, 2019","year":2019,"lang":"en","type":"dataset","venue":"Summit (Simon Fraser University)","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Computer science","score_opus":0.08996723890744256,"score_gpt":0.376945656635178,"score_spread":0.2869784177277354,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W100260889","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002805374,0.00010755216,0.000096106094,0.0010401091,0.0014798145,0.0010331887,0.97730553,0.000028922532,0.016103372],"genre_scores_gemma":[0.003145351,0.00052578055,0.00009865251,0.00079223304,0.0002240322,0.0000028366721,0.9209767,0.000046112455,0.07418828],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99354976,0.0006966318,0.0010238758,0.0014982534,0.0025952284,0.00063625287],"domain_scores_gemma":[0.99136853,0.0029211377,0.0016781067,0.0027278515,0.00093979377,0.00036455737],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.0033877518,0.0005341216,0.0014278793,0.002834717,0.00026836834,0.004738296,0.05463035,0.0009594284,0.003117792],"category_scores_gemma":[0.0042017,0.00047225098,0.0003406408,0.0054737525,0.0003300438,0.010649614,0.016851956,0.0015467983,0.002127799],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012759681,0.00008318803,0.05585783,0.00008800668,0.00009177523,0.00007270437,0.0000041359963,0.00002205951,1.4008006e-7,0.00017196512,0.9431179,0.00036266318],"study_design_scores_gemma":[0.0008726522,0.000056353856,0.0009566121,0.00019514083,0.00014080909,1.6200595e-8,0.0016164411,0.000032927095,0.000035063586,0.0019640005,0.99356914,0.0005608437],"about_ca_topic_score_codex":0.006697081,"about_ca_topic_score_gemma":0.014141517,"teacher_disagreement_score":0.058084905,"about_ca_system_score_codex":0.0002515573,"about_ca_system_score_gemma":0.0012375333,"threshold_uncertainty_score":0.9999174},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W131789328","doi":"","title":"Exploring the Role of Books as a Knowledge Translation Mechanism: Citation Analysis and Author Survey","year":2011,"lang":"en","type":"article","venue":"Journal of the Association for Information Systems","topic":"Knowledge Management and Sharing","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Lakehead University","funders":"","keywords":"Citation; Scholarly communication; Computer science; Citation analysis; Library science; Content analysis; Knowledge management; World Wide Web; Sociology; Social science; Publishing; Literature; Art","score_opus":0.19547752645600486,"score_gpt":0.3120364624736172,"score_spread":0.11655893601761236,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W131789328","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.83712596,0.00036810647,0.018175824,0.0003618836,0.003387072,0.0018289764,0.00004148119,0.00003984103,0.13867085],"genre_scores_gemma":[0.9990328,0.000027704767,0.000062860796,0.0000068142735,0.000080479396,0.000017471773,0.000003697652,0.0000030021458,0.000765151],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.99837565,0.000410555,0.0006306925,0.000036481957,0.00044004325,0.00010655612],"domain_scores_gemma":[0.99692845,0.00044333065,0.0015979643,0.000075249474,0.00092352706,0.000031496573],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008072832,0.000057217836,0.00017534452,0.0003225474,0.00038451087,0.00014443247,0.00020834933,0.000050210703,0.000003078846],"category_scores_gemma":[0.001000721,0.00003802081,0.00016041193,0.0006909442,0.000017781414,0.0015488643,0.000018346902,0.000075258446,0.0000037388977],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000116748735,0.000045353903,0.14400788,0.00009128766,0.0013909047,3.0424456e-8,0.57155734,0.00020719689,0.00008991655,0.26295707,0.00025954173,0.019276738],"study_design_scores_gemma":[0.0016783205,0.0001975973,0.6781697,0.00018621169,0.002110568,0.0000014268201,0.2068964,0.013237835,0.0017422539,0.013394705,0.082012266,0.00037271436],"about_ca_topic_score_codex":0.00060876017,"about_ca_topic_score_gemma":0.00067959726,"teacher_disagreement_score":0.5341618,"about_ca_system_score_codex":0.000156635,"about_ca_system_score_gemma":0.000063396415,"threshold_uncertainty_score":0.29573855},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1474036562","doi":"10.1007/s11367-015-0951-2","title":"I publish, therefore I am. Or am I? A reply to A bibliometric investigation of life cycle assessment research in the web of science databases by Chen et al. (2014) and Mapping the scientific research on life cycle assessment: a bibliometric analysis by Hou et al. (2015)","year":2015,"lang":"en","type":"article","venue":"The International Journal of Life Cycle Assessment","topic":"Environmental Impact and Sustainability","field":"Environmental Science","cited_by":18,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal","funders":"Polytechnique Montréal","keywords":"Ranking (information retrieval); Web of science; Scopus; Citation; Publication; Chen; Computer science; Library science; Bibliometrics; Journal ranking; Database; Data science; Information retrieval; MEDLINE; Political science","score_opus":0.16032396174720037,"score_gpt":0.45762717775225104,"score_spread":0.29730321600505066,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1474036562","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8889464,0.0003234926,0.0005994809,0.10774006,0.00017334893,0.0008029205,0.00014104655,0.000008734057,0.001264477],"genre_scores_gemma":[0.9896521,0.0005694614,0.0014945872,0.0080407765,0.000039206334,0.00006590283,0.000038230428,0.000025251882,0.00007446386],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9833223,0.002914492,0.0014311959,0.0007791449,0.010767179,0.000785672],"domain_scores_gemma":[0.99304795,0.0029707244,0.0010338068,0.0011947344,0.0009866824,0.00076612685],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.07150945,0.00029833114,0.0005315834,0.053000167,0.0005795622,0.0014284046,0.0036377104,0.000058708192,0.0002811173],"category_scores_gemma":[0.0048200446,0.00016523786,0.00016405612,0.17272317,0.0029207328,0.0023898196,0.0020427785,0.0013204876,0.000009911027],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00045430305,0.0023030217,0.6510927,0.000034496108,0.0005412634,0.000018592706,0.002605791,0.016179113,0.009871181,0.0005962504,0.31175408,0.004549194],"study_design_scores_gemma":[0.0012953328,0.0008743328,0.94635713,0.000101345606,0.000052810825,0.000012511453,0.016536912,0.017879745,0.00031829003,0.00082219025,0.015523137,0.00022627493],"about_ca_topic_score_codex":0.004070056,"about_ca_topic_score_gemma":0.00032104063,"teacher_disagreement_score":0.29623094,"about_ca_system_score_codex":0.0018525783,"about_ca_system_score_gemma":0.0023702683,"threshold_uncertainty_score":0.99979275},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1483618109","doi":"10.1111/jebm.12022","title":"Visualization of evidence‐based medicine domain knowledge: production and citation of cochrane systematic reviews","year":2013,"lang":"en","type":"article","venue":"Journal of Evidence-Based Medicine","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"China Medical Board","keywords":"Citation; Knowledge translation; Systematic review; Cochrane Library; Psychological intervention; Production (economics); MEDLINE; Bibliometrics; Medicine; Developing country; Business; Political science; Alternative medicine; Computer science; Library science; Knowledge management; Economic growth; Nursing; Pathology; Economics","score_opus":0.842957060811485,"score_gpt":0.5825738731360253,"score_spread":0.26038318767545976,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1483618109","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16316828,0.48662338,0.2692674,0.06798482,0.0028673229,0.009745757,0.0000025945822,0.000011460553,0.00032899817],"genre_scores_gemma":[0.98448795,0.008023638,0.0053435354,0.00045462672,0.0007536495,0.00007988853,0.0000029650887,0.000020663974,0.0008330784],"study_design_codex":"not_applicable","study_design_gemma":"systematic_review","domain_scores_codex":[0.94851714,0.019810535,0.022867136,0.0006303273,0.0079351,0.0002397445],"domain_scores_gemma":[0.89528346,0.03883656,0.044315193,0.0026883592,0.018417709,0.0004587246],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.24522083,0.00042188086,0.00805038,0.0019801538,0.000092950984,0.00008040948,0.0010313413,0.00012665949,0.0034963603],"category_scores_gemma":[0.5259732,0.00015062978,0.00085439114,0.0035202778,0.00054286694,0.0013830456,0.000035043533,0.00022318194,0.000079903395],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00074196566,0.0009508242,0.05608591,0.33120292,0.0012784749,0.000027577576,0.014033392,0.0016018267,0.21547139,0.002231777,0.3322066,0.044167336],"study_design_scores_gemma":[0.0028789924,0.0056508183,0.020996535,0.9264463,0.005308474,0.00017835115,0.0060002315,0.019107135,0.0038722663,0.005995139,0.0030185909,0.00054717675],"about_ca_topic_score_codex":0.00004119201,"about_ca_topic_score_gemma":0.000014188989,"teacher_disagreement_score":0.8213197,"about_ca_system_score_codex":0.00010743547,"about_ca_system_score_gemma":0.00034676565,"threshold_uncertainty_score":0.9974146},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W1488788462","doi":"","title":"Scholarly Metrics under the Microscope: From Citation Analysis to Academic Auditing","year":2015,"lang":"en","type":"article","venue":"Partnership The Canadian Journal of Library and Information Practice and Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Citation; Reputation; Audit; Documentation; Library science; Dimension (graph theory); Sociology; Computer science; Social science; Mathematics; Accounting; Business","score_opus":0.6638953583953724,"score_gpt":0.5733188140237518,"score_spread":0.09057654437162055,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1488788462","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6304858,0.0066863033,0.0022717626,0.33938354,0.0003021449,0.00033667748,0.000035582034,0.0000051247594,0.020493083],"genre_scores_gemma":[0.99200195,0.00078288134,0.00064055395,0.0058581643,0.00015819813,0.000003040696,0.0000062435242,0.0000039761894,0.0005449857],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9920554,0.0017267272,0.00075246155,0.00014729415,0.004896035,0.00042204245],"domain_scores_gemma":[0.9736465,0.019438736,0.00055864727,0.00033856375,0.0043199505,0.0016975583],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.06336895,0.00008136334,0.0001718208,0.03449224,0.0011577796,0.016623976,0.001469234,0.00011731566,0.00009858706],"category_scores_gemma":[0.106013454,0.00004051841,0.00006061346,0.10541618,0.0003008348,0.04249504,0.0002607655,0.0014402965,0.00008372542],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00071734306,0.000043328586,0.22343339,0.000015706357,0.0009416488,0.000047142807,0.042422075,0.0029424923,0.00009173308,0.12029958,0.39858356,0.21046199],"study_design_scores_gemma":[0.0003565802,0.00021645334,0.09925648,0.000013039944,0.000076437296,0.000038352107,0.06608954,0.0024158098,0.00015620174,0.013987589,0.8172839,0.00010959015],"about_ca_topic_score_codex":0.001517922,"about_ca_topic_score_gemma":0.00023802108,"teacher_disagreement_score":0.41870037,"about_ca_system_score_codex":0.0000675764,"about_ca_system_score_gemma":0.0022345185,"threshold_uncertainty_score":0.9843969},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W1493387337","doi":"10.1111/j.1365-2044.2010.06424.x","title":"National representation in the anaesthesia literature: a bibliometric analysis of highly cited anaesthesia journals <sup>*</sup>","year":2010,"lang":"en","type":"article","venue":"Anaesthesia","topic":"Cardiac, Anesthesia and Surgical Outcomes","field":"Medicine","cited_by":67,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Michael's Hospital; Health Sciences Centre; McMaster University Medical Centre; Children's Hospital of Eastern Ontario","funders":"","keywords":"Medicine; Bibliometrics; Citation; Anesthesiology; MEDLINE; China; Subject (documents); Family medicine; Library science; Anesthesia; Political science; Law","score_opus":0.02573323452045045,"score_gpt":0.32150612655895505,"score_spread":0.2957728920385046,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1493387337","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98644614,0.0017623779,0.00004508545,0.0070722965,0.000039558647,0.00059342384,0.000013741509,0.00006446766,0.0039628954],"genre_scores_gemma":[0.99459267,0.0012441914,0.00093653495,0.002492719,0.00019671873,0.000058778096,0.00017596206,0.000044718334,0.00025769498],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9947191,0.00064317044,0.00123351,0.0006447286,0.0021800909,0.0005794124],"domain_scores_gemma":[0.9967256,0.00072922226,0.0004665597,0.0009918525,0.0008048266,0.0002818917],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0034386336,0.00043849045,0.0016301437,0.058474325,0.00015058697,0.00023666973,0.00056051323,0.00042508077,0.00018848394],"category_scores_gemma":[0.0013422209,0.00030322454,0.0014628745,0.17423236,0.00023491881,0.00043354745,0.000031247237,0.00091260416,0.000031351286],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032359158,0.0005848004,0.9401734,0.000043947515,0.00093887094,0.0011983464,0.001817182,0.0005955116,0.000499629,0.011731639,0.0009543332,0.04113874],"study_design_scores_gemma":[0.0012358556,0.00027394152,0.97816724,0.0000739312,0.0012771548,0.00040531403,0.00026494276,0.0029566928,0.00012770924,0.00030975288,0.01459809,0.00030940192],"about_ca_topic_score_codex":0.0002255991,"about_ca_topic_score_gemma":0.00006621662,"teacher_disagreement_score":0.11575805,"about_ca_system_score_codex":0.000076748416,"about_ca_system_score_gemma":0.0002455827,"threshold_uncertainty_score":0.999942},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W1495039464","doi":"10.7910/dvn/sikkdv","title":"Dramatic Growth of Open Access","year":2010,"lang":"en","type":"dataset","venue":"uO Research (University of Ottawa)","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science","score_opus":0.3336734854130005,"score_gpt":0.5300456706965908,"score_spread":0.1963721852835903,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1495039464","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0028061757,0.000025827649,0.00009219905,0.004044221,0.00037869366,0.000812066,0.9565208,0.000013134564,0.035306897],"genre_scores_gemma":[0.0118156625,0.0007828812,0.0021347217,0.0002453661,0.00042598537,0.000005126705,0.9332267,0.00006135182,0.051302217],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9903596,0.0012921139,0.000662358,0.0009698327,0.0060916566,0.00062443456],"domain_scores_gemma":[0.98886585,0.004274911,0.0011388861,0.0022925665,0.0030095198,0.00041828433],"candidate_categories":["metaresearch","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["open_science"],"category_scores_codex":[0.02206549,0.00023248619,0.0010230504,0.0029294125,0.0005936166,0.0029944465,0.06933033,0.00086969463,0.008761232],"category_scores_gemma":[0.02012994,0.00021526398,0.00021952778,0.0044313893,0.0015264483,0.0075706393,0.023502786,0.0033369705,0.00063357974],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001301601,0.000112688154,0.0029221976,0.00011043335,0.000045024462,0.00003841874,0.00009360436,8.620935e-7,0.000015369105,0.0014587428,0.99317694,0.0020127203],"study_design_scores_gemma":[0.0004523299,0.00007251802,0.005411168,0.00011653565,0.00002484909,0.000003593305,0.0008947082,0.00004078101,0.000060965744,0.013700164,0.9790104,0.00021201302],"about_ca_topic_score_codex":0.015678521,"about_ca_topic_score_gemma":0.007945547,"teacher_disagreement_score":0.04582754,"about_ca_system_score_codex":0.000092800066,"about_ca_system_score_gemma":0.0013680337,"threshold_uncertainty_score":0.99896234},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W1496204915","doi":"10.1038/nm1201-1261b","title":"Canada tops cancer research impact list","year":2001,"lang":"en","type":"article","venue":"Nature Medicine","topic":"Health and Medical Research Impacts","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"TOPS; Cancer; Gerontology; Medicine; Internal medicine; Engineering","score_opus":0.17326723448763734,"score_gpt":0.5781129353655865,"score_spread":0.40484570087794913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1496204915","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.50619215,0.014590604,0.000008940679,0.43667465,0.0007975394,0.00071042945,0.00002625872,0.000049151702,0.04095027],"genre_scores_gemma":[0.92785996,0.0037837778,0.000022428478,0.05060867,0.003957391,0.00003758206,0.00005665975,0.000031422125,0.013642085],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9935443,0.00017688119,0.00034897024,0.0003732209,0.0036168532,0.001939789],"domain_scores_gemma":[0.98890007,0.0007769765,0.000047153797,0.0005286074,0.000986609,0.008760598],"candidate_categories":["metaresearch","research_integrity","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0034750183,0.00020523109,0.0005254374,0.00033987933,0.00019444288,0.000009589892,0.00023253816,0.00040647804,0.0070101228],"category_scores_gemma":[0.0574182,0.00011958606,0.000056323464,0.0015468473,0.0003492608,0.000056584468,0.00005481414,0.0040780655,0.000025782108],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00058930134,0.00006135544,0.017511625,0.00015888101,0.000045093613,0.0016433039,0.000080438534,5.117336e-7,0.0025365944,0.00009107069,0.9394449,0.03783692],"study_design_scores_gemma":[0.002496478,0.00094157865,0.049065076,0.00074695895,0.000026322012,0.0002801058,0.00019217828,0.000053112348,0.00058202277,0.000066580105,0.9454446,0.00010499449],"about_ca_topic_score_codex":0.8980513,"about_ca_topic_score_gemma":0.62995327,"teacher_disagreement_score":0.4216678,"about_ca_system_score_codex":0.002135873,"about_ca_system_score_gemma":0.035482172,"threshold_uncertainty_score":0.99821955},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":true,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"medium"}],"label_agreement":"split"},{"id":"W1516312676","doi":"","title":"Highly Cited Canada Articles in Science Citation Index Expanded: A Bibliometric Analysis","year":2015,"lang":"en","type":"article","venue":"Canadian social science","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Science Citation Index; Library science; Index (typography); Web of science; Bibliometrics; Citation analysis; Institution; Multidisciplinary approach; Political science; Social science; Sociology; Computer science; MEDLINE; World Wide Web; Law","score_opus":0.400457680858481,"score_gpt":0.5049024716013509,"score_spread":0.10444479074286994,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1516312676","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9663247,0.00022598392,0.00041095944,0.0012904062,0.00056822057,0.0001919362,0.00004392327,0.000014719063,0.030929176],"genre_scores_gemma":[0.9989911,0.000011438743,0.00012477794,0.0004967713,0.00006211941,0.0000107166015,0.0000022299234,0.000006033837,0.00029481587],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9761279,0.00016494906,0.00071583904,0.0012112252,0.019898016,0.00188204],"domain_scores_gemma":[0.98648196,0.00089145557,0.00023704038,0.0006218688,0.00813454,0.0036331532],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.04286531,0.00017225067,0.00037241547,0.81387734,0.001205565,0.005289014,0.005154308,0.00009476436,0.000118654796],"category_scores_gemma":[0.06894001,0.00014747972,0.00008727541,0.9800926,0.002038422,0.0023288259,0.00043075345,0.0002764648,0.000058621274],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000004637906,0.000021143534,0.9184352,7.215703e-7,0.0000060973457,0.000045728706,0.00061818806,0.00010210048,0.00028697783,0.00096083444,0.0025475991,0.07697078],"study_design_scores_gemma":[0.00029091546,0.00002890092,0.9761351,0.0000020584848,0.00000784815,0.0000014058211,0.004569602,0.012481585,0.0002782971,0.0017782804,0.0041935234,0.00023252679],"about_ca_topic_score_codex":0.9391933,"about_ca_topic_score_gemma":0.97974277,"teacher_disagreement_score":0.16621524,"about_ca_system_score_codex":0.005950873,"about_ca_system_score_gemma":0.030951861,"threshold_uncertainty_score":0.9978651},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"agree"},{"id":"W1519442902","doi":"","title":"Bibliometric analysis of the American Journal of Veterinary Research to produce a list of core veterinary medicine journals.","year":2006,"lang":"en","type":"article","venue":"PubMed","topic":"Veterinary Practice and Education Studies","field":"Health Professions","cited_by":52,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Veterinary medicine; Bibliometrics; Medicine; Library science; Computer science","score_opus":0.6419000699967484,"score_gpt":0.5869553323203693,"score_spread":0.054944737676379085,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1519442902","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9871603,0.00140751,0.000013349557,0.008287547,0.00065705186,0.0006246632,0.000017589542,0.000006169598,0.0018258181],"genre_scores_gemma":[0.99731505,0.0005271587,0.00015426369,0.00023440566,0.0005800752,0.00025930494,0.0000018363098,0.000015391446,0.00091251556],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99532837,0.0014717149,0.0013423775,0.00020245965,0.0011256696,0.0005294374],"domain_scores_gemma":[0.99291605,0.0021749146,0.0018437429,0.00055578304,0.0023496875,0.00015980235],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.008735168,0.00012688883,0.00079440593,0.04503832,0.0004160959,0.000008691016,0.0005632578,0.000047938764,0.00012559767],"category_scores_gemma":[0.0045678667,0.00007982696,0.00018113192,0.15748481,0.00044100478,0.00018308846,0.00036044177,0.0006580109,0.0000037116467],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022682324,0.00077615236,0.8052382,0.000471654,0.0015041104,0.000035865014,0.0068583223,0.0000567735,0.013269016,0.00004998225,0.07732481,0.0921469],"study_design_scores_gemma":[0.00030137663,0.0011026122,0.96504056,0.00013646783,0.00035511883,0.00003987862,0.006439787,0.0000027646254,0.000045674446,0.00005234463,0.026417766,0.00006565022],"about_ca_topic_score_codex":0.0014360697,"about_ca_topic_score_gemma":0.00003369103,"teacher_disagreement_score":0.15980238,"about_ca_system_score_codex":0.0001547082,"about_ca_system_score_gemma":0.00021662643,"threshold_uncertainty_score":0.9657854},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W1530517025","doi":"10.1002/asi.23027","title":"The knowledge base and research front of information science 2006–2010: An author cocitation and bibliographic coupling analysis","year":2014,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":118,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Webometrics; Bibliographic coupling; Knowledge base; Scientometrics; Citation analysis; Front (military); Web of science; Data science; Computer science; Bibliometrics; Information science; Period (music); Network analysis; Citation; Library science; World Wide Web; Political science; MEDLINE; Physics","score_opus":0.23216091820544282,"score_gpt":0.5205154533318411,"score_spread":0.2883545351263983,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1530517025","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9861547,0.0002085948,0.0069395,0.005485698,0.0003723175,0.0002899472,0.000016734704,0.000007701477,0.0005248531],"genre_scores_gemma":[0.9987474,0.00036140077,0.00077875156,0.000034933728,0.00001584554,0.0000057996085,6.8049945e-7,0.0000012611599,0.000053909174],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99172753,0.00011710211,0.0010565193,0.0001626187,0.00651302,0.00042322828],"domain_scores_gemma":[0.9625207,0.0038041514,0.0019463864,0.0003855405,0.03114583,0.00019737221],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.1563829,0.00007295605,0.00022294626,0.12054029,0.0021830841,0.0031942758,0.0017116831,0.00012036723,0.0000017579562],"category_scores_gemma":[0.13574907,0.000039200524,0.00006032514,0.23789826,0.0018951222,0.010067176,0.0004871218,0.00030500896,0.0000024875164],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006565619,0.00006275449,0.5927094,0.00002207323,0.000074126576,3.99706e-8,0.003422791,0.00042001903,0.0033864866,0.07882194,0.0029771514,0.31803754],"study_design_scores_gemma":[0.0007305701,0.00050586375,0.6841904,0.000014837546,0.000052664476,0.000006887899,0.006948883,0.25581768,0.0021835314,0.03082073,0.018618667,0.000109289715],"about_ca_topic_score_codex":0.000023647923,"about_ca_topic_score_gemma":0.0000313437,"teacher_disagreement_score":0.31792825,"about_ca_system_score_codex":0.00019152842,"about_ca_system_score_gemma":0.0005925051,"threshold_uncertainty_score":0.99911594},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W1625449191","doi":"10.1111/j.1756-5391.2011.01127.x","title":"Visualization studies on evidence‐based medicine domain knowledge (series 2): structural diagrams of author networks","year":2011,"lang":"en","type":"article","venue":"Journal of Evidence-Based Medicine","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"China Medical Board","keywords":"Zhàng; China; Visualization; Citation; Cluster analysis; Graph drawing; Computer science; Domain (mathematical analysis); Library science; Data science; Information retrieval; Geography; Data mining; Artificial intelligence; Mathematics","score_opus":0.9085942578151824,"score_gpt":0.6040802977165222,"score_spread":0.3045139600986603,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1625449191","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.41678205,0.27888423,0.235272,0.05330828,0.011890634,0.0027905842,0.000007784271,0.00003287692,0.0010315503],"genre_scores_gemma":[0.9911826,0.0014852963,0.003692849,0.0010416247,0.001627153,0.000016787626,0.0000025855218,0.000026011481,0.0009250678],"study_design_codex":"not_applicable","study_design_gemma":"systematic_review","domain_scores_codex":[0.9627207,0.012035186,0.016222617,0.0007479131,0.007862512,0.00041107624],"domain_scores_gemma":[0.93909436,0.028417561,0.020545464,0.002320327,0.00909028,0.0005319876],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.15449385,0.00061888737,0.0065466203,0.0015931365,0.00021500493,0.00008201125,0.0020909049,0.00018040203,0.0055659716],"category_scores_gemma":[0.2378163,0.00022957209,0.001342356,0.0032798485,0.0011493532,0.0011326063,0.000083176965,0.00046448005,0.000051824278],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.007976664,0.0009861822,0.22533655,0.004489833,0.0050740233,0.00063049525,0.049717266,0.016271925,0.0062286714,0.03304093,0.52974576,0.12050169],"study_design_scores_gemma":[0.016926404,0.07174381,0.23323873,0.30279425,0.018608226,0.0005903781,0.07009902,0.13864766,0.01043741,0.089099266,0.04435157,0.003463283],"about_ca_topic_score_codex":0.00002353862,"about_ca_topic_score_gemma":0.000052151347,"teacher_disagreement_score":0.57440054,"about_ca_system_score_codex":0.00017011445,"about_ca_system_score_gemma":0.00043702533,"threshold_uncertainty_score":0.9953431},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W1645565896","doi":"","title":"Análisis bibliométrico de la teoría de gestión estratégica basada en recursos y competencias","year":2003,"lang":"es","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Organizational Management and Innovation","field":"Business, Management and Accounting","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Political science","score_opus":0.1164361084106354,"score_gpt":0.4802326337203018,"score_spread":0.3637965253096664,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1645565896","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.86205083,0.008629872,0.002509636,0.001619942,0.0010825754,0.0010327589,0.00003339386,0.00014767723,0.12289333],"genre_scores_gemma":[0.98256624,0.01201249,0.0012325599,0.0021370342,0.001111208,0.000031550688,0.0000615463,0.00014209263,0.000705274],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99554074,0.0005231581,0.0013300333,0.0006788553,0.0011410263,0.000786208],"domain_scores_gemma":[0.9955762,0.0007606404,0.0018096719,0.0005429499,0.001190838,0.00011971358],"candidate_categories":["metaepi_narrow","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0049731936,0.0006258595,0.0008621027,0.0074535306,0.0006699344,0.011205368,0.002948509,0.0003333387,0.020374725],"category_scores_gemma":[0.0030702096,0.0006439375,0.00022485614,0.017429499,0.00021360103,0.00929327,0.0010579323,0.0007100122,0.00015045195],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007923151,0.0004519691,0.8170942,0.00047800544,0.00027688264,0.00007249327,0.00004807987,0.00026544338,0.0141339265,0.037454933,0.12471987,0.004924965],"study_design_scores_gemma":[0.00078695716,0.0000082548795,0.8246617,0.0006944571,0.00035621173,0.000013771516,0.00014741946,0.0003571926,0.0031239453,0.020542393,0.1485651,0.0007426351],"about_ca_topic_score_codex":0.0011169637,"about_ca_topic_score_gemma":0.000019174313,"teacher_disagreement_score":0.12218806,"about_ca_system_score_codex":0.00026382803,"about_ca_system_score_gemma":0.00031819264,"threshold_uncertainty_score":0.9996012},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W164663678","doi":"","title":"Citation analysis to reconstruct the dynamics of Antarctic ozone hole research and formulation of the Montreal Protocol","year":2013,"lang":"en","type":"article","venue":"KTH Publication Database DiVA (KTH Royal Institute of Technology)","topic":"Atmospheric Ozone and Climate","field":"Earth and Planetary Sciences","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Montreal Protocol; Ozone layer; Ozone; Atmosphere (unit); Ozone depletion; Environmental science; Atmospheric sciences; Layer (electronics); Astrobiology; Safeguarding; Meteorology; Chemistry; Geology; Physics","score_opus":0.029860493780894257,"score_gpt":0.2876424118904356,"score_spread":0.2577819181095413,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W164663678","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9798964,0.000019691739,0.0017620764,0.005423352,0.000046614492,0.010177666,0.00041678283,0.000032091033,0.0022253203],"genre_scores_gemma":[0.99222237,0.000008514524,0.0062642787,0.000030420993,0.000009155061,0.00091520266,0.00042002436,0.0000027306937,0.00012727799],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9987142,0.00005412861,0.00044956777,0.00024050471,0.00034473903,0.00019685928],"domain_scores_gemma":[0.9982073,0.00007515475,0.00035222762,0.0006991419,0.0006162943,0.00004991572],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005911036,0.000096510455,0.00019898625,0.00025860587,0.00019546028,0.000035364275,0.00052246894,0.000094628725,0.00028834297],"category_scores_gemma":[0.0004484536,0.00005978811,0.000045827797,0.0030114066,0.0005548504,0.00049920776,0.00012420412,0.00017856955,0.000009532394],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000058937723,0.00009583945,0.8597383,0.000110231435,0.00016899212,2.4854646e-7,0.0001299463,0.002917393,0.00021435448,0.009853975,0.0009256468,0.12578613],"study_design_scores_gemma":[0.00032675877,0.00020182037,0.8438777,0.000042028732,0.00009559221,0.000002194313,0.00067266874,0.14998281,0.0010880892,0.0020845868,0.0015083515,0.000117441734],"about_ca_topic_score_codex":0.010601509,"about_ca_topic_score_gemma":0.00827546,"teacher_disagreement_score":0.14706542,"about_ca_system_score_codex":0.000012839677,"about_ca_system_score_gemma":0.000080318656,"threshold_uncertainty_score":0.995987},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W1671256466","doi":"","title":"Organization of hospital chaplaincy services: a bibliometric study","year":2011,"lang":"en","type":"article","venue":"","topic":"Religion and Society in Latin America","field":"Social Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Accreditation; Commission; Object (grammar); Library science; Work (physics); Index (typography); Health care; Political science; Sociology; Medicine; Medical education; Engineering; Computer science; Law","score_opus":0.02167991972547393,"score_gpt":0.2928166534097556,"score_spread":0.27113673368428165,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1671256466","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7933344,0.000019289526,0.00023518175,0.00007849103,0.00018455702,0.00024256186,5.9204433e-7,0.00008377916,0.20582111],"genre_scores_gemma":[0.99833703,0.0002244553,0.00076097716,0.00010452671,0.000036527075,0.0000029957882,0.0000010594554,0.0000067560477,0.0005256515],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9991625,0.00007510346,0.0001636928,0.00011585246,0.00034933173,0.0001334978],"domain_scores_gemma":[0.9993726,0.00003858291,0.00010349654,0.00012281386,0.00029593255,0.00006660058],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.000322009,0.00005293047,0.00010353883,0.0018561126,0.00017711084,0.000022378497,0.00023457318,0.000047730668,0.0020980102],"category_scores_gemma":[0.000109983215,0.000045746947,0.000032596836,0.045688163,0.00010383481,0.0001509655,0.00003778249,0.000036777954,0.000070892755],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000010850484,0.0007284033,0.6813257,0.000007600001,0.000023885517,8.891289e-7,0.3038476,1.8088993e-7,0.000009768556,0.011543983,0.00060715957,0.0019037242],"study_design_scores_gemma":[0.00047845958,0.0007192707,0.5622533,0.000009798575,0.000029922281,1.426963e-7,0.4323755,0.000018313534,0.00016478081,0.0006581594,0.003052483,0.00023985787],"about_ca_topic_score_codex":0.0083433315,"about_ca_topic_score_gemma":0.000092987735,"teacher_disagreement_score":0.20529546,"about_ca_system_score_codex":0.000018668787,"about_ca_system_score_gemma":0.000043851756,"threshold_uncertainty_score":0.9988142},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W1724869758","doi":"10.1023/a:1014256102041","title":"Collaboration and Cognitive Structures in Social Science Research Fields. Towards Socio-Cognitive Analysis in Information Systems","year":2001,"lang":"en","type":"article","venue":"Scientometrics","topic":"Knowledge Management and Sharing","field":"Social Sciences","cited_by":43,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Mainstream; Consolidation (business); Cognition; Field (mathematics); Sociology; Value (mathematics); Social network analysis; Knowledge management; Epistemology; Social science; Computer science; Psychology; Political science; Business; Social capital","score_opus":0.09573227246461083,"score_gpt":0.44943950812195993,"score_spread":0.3537072356573491,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1724869758","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.88329846,0.0001846134,0.0005126857,0.00019873948,0.000229792,0.00045383457,0.000008257317,0.000018917255,0.11509472],"genre_scores_gemma":[0.9993936,0.000112290916,0.000017639848,0.000029698713,0.00008288825,0.00002116191,0.000010905481,0.0000021323945,0.00032972186],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9969449,0.00026432282,0.00026032238,0.0002679544,0.0017488993,0.0005135717],"domain_scores_gemma":[0.9980844,0.00031250942,0.000087659326,0.000055787084,0.0013671606,0.00009247985],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.009587261,0.00007555395,0.00016276387,0.013945316,0.0011210191,0.0013179001,0.00027218673,0.00010289104,0.000023031142],"category_scores_gemma":[0.005288403,0.00007779912,0.00002486976,0.088501714,0.00080137316,0.0019861222,0.00017664299,0.00022150583,0.000010791246],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000037105674,0.00005849356,0.6752026,0.00003205007,0.00003249327,0.0000073024967,0.18902326,0.00005790975,0.0000058819314,0.04560138,0.00010158243,0.08983993],"study_design_scores_gemma":[0.00058834575,0.000032815424,0.79225373,0.000027577642,0.000038033973,1.2501476e-7,0.20112053,0.0031734267,0.000015249943,0.0014381902,0.0011476729,0.00016428217],"about_ca_topic_score_codex":0.0021985674,"about_ca_topic_score_gemma":0.0035517197,"teacher_disagreement_score":0.11705114,"about_ca_system_score_codex":0.00045410387,"about_ca_system_score_gemma":0.00037472718,"threshold_uncertainty_score":0.99971884},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1766971555","doi":"10.1023/a:1016013912188","title":"Informetric analysis of a music database","year":2002,"lang":"en","type":"article","venue":"Scientometrics","topic":"Scientific Research and Discoveries","field":"Physics and Astronomy","cited_by":10,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Zipf's law; Computer science; Poisson distribution; Information retrieval; Equivalence (formal languages); Informetrics; Database; Data mining; Statistics; Mathematics; Bibliometrics","score_opus":0.08002575222830237,"score_gpt":0.32166463378928656,"score_spread":0.24163888156098418,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1766971555","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8791011,0.00026618483,0.0076649785,0.000025487869,0.0002602391,0.0001228896,0.00048787627,0.000016521615,0.112054735],"genre_scores_gemma":[0.99565965,0.000006578233,0.00041660128,0.0000119877795,0.000028428667,0.000004374548,0.00008341619,0.000003113548,0.003785873],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9981932,0.000017212462,0.0002206262,0.00022307686,0.0010259687,0.0003199233],"domain_scores_gemma":[0.9988667,0.00015245045,0.00010182027,0.00043943856,0.00025826285,0.0001813795],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00066544546,0.00007611402,0.00018290356,0.010690344,0.00009747974,0.00014636696,0.00036930788,0.0000131430415,0.008052825],"category_scores_gemma":[0.00036095895,0.000062784726,0.00014518373,0.106951036,0.0001541643,0.00040686165,0.00014260874,0.000067036264,0.00013254356],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000063912757,0.0006249488,0.8541872,0.00002305439,0.00069139304,0.0000020278746,0.0010058586,0.00037981145,0.0009634753,0.019211112,0.042918857,0.079985894],"study_design_scores_gemma":[0.0037151824,0.000492286,0.5039988,0.00004130023,0.0023049251,0.0000010045399,0.0067263176,0.33950132,0.027062243,0.002156284,0.11224534,0.0017549691],"about_ca_topic_score_codex":0.0001474459,"about_ca_topic_score_gemma":0.000005130658,"teacher_disagreement_score":0.35018837,"about_ca_system_score_codex":0.00001742804,"about_ca_system_score_gemma":0.000032850854,"threshold_uncertainty_score":0.99285394},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W1774927437","doi":"10.1023/a:1012761903614","title":"Investigating the concentration within a research community using joint publications and co-authorship via intermediaries","year":2001,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":11,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Quebec Statistical Institute; Université Laval","funders":"","keywords":"Wonder; Field (mathematics); Period (music); Political science; Sociology; Computer science; Operations research; Epistemology; Mathematics; Philosophy","score_opus":0.9143610746626762,"score_gpt":0.6578600446631518,"score_spread":0.2565010299995244,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1774927437","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9726645,0.00052262517,0.017817054,0.0034463578,0.00062736904,0.00056147017,0.000022608447,0.00005157914,0.004286471],"genre_scores_gemma":[0.99563587,0.00009657643,0.002963396,0.0002551333,0.00011031246,0.000021703223,0.000008446481,0.000014578344,0.0008939884],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9741909,0.004210167,0.0012878188,0.0008727388,0.018108575,0.0013297987],"domain_scores_gemma":[0.9336367,0.042077243,0.0010344805,0.0031642043,0.018022709,0.0020646104],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.20395179,0.00019774341,0.0003197948,0.09063404,0.004677109,0.012246966,0.0036195558,0.00018760307,0.0001540988],"category_scores_gemma":[0.3803259,0.00012482706,0.00009228703,0.5311368,0.0045192726,0.0015484529,0.001782836,0.0019430174,0.00011283209],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011161773,0.0002788132,0.8948023,0.000015592826,0.000021555321,0.0000047153185,0.012732099,0.00019895787,0.0070301443,0.012337531,0.008636611,0.06393056],"study_design_scores_gemma":[0.00077392976,0.00036570337,0.5864945,0.00004516,0.000016353495,0.0001360675,0.014907818,0.25502226,0.0039103036,0.1230517,0.014759676,0.00051652745],"about_ca_topic_score_codex":0.0005938893,"about_ca_topic_score_gemma":0.000114592156,"teacher_disagreement_score":0.4405028,"about_ca_system_score_codex":0.00040056213,"about_ca_system_score_gemma":0.0006439393,"threshold_uncertainty_score":0.99818987},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1828896994","doi":"","title":"Arthritis and rheumatism are neglected health priorities: a bibliometric study.","year":2001,"lang":"en","type":"article","venue":"PubMed","topic":"Musculoskeletal Disorders and Rehabilitation","field":"Medicine","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Medicine; Rheumatism; MEDLINE; Rheumatoid arthritis; Arthritis; Disease; Physical therapy; Family medicine; Intensive care medicine; Internal medicine","score_opus":0.021873177126159563,"score_gpt":0.2710635213917629,"score_spread":0.24919034426560333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1828896994","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96627027,0.026763778,0.00006501088,0.003517159,0.00011640187,0.0023606103,0.000003770863,0.00009213986,0.0008108633],"genre_scores_gemma":[0.97963625,0.018547548,0.00010979725,0.00035838404,0.00004657883,0.0008064276,0.000007459003,0.00001636348,0.0004711735],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9987937,0.00007778888,0.00028228477,0.00024142463,0.00027682673,0.0003279764],"domain_scores_gemma":[0.99932367,0.00008693042,0.000093172115,0.00017467019,0.000083082115,0.00023847139],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00024606814,0.00011142664,0.0003142631,0.005957565,0.000101047124,0.000038168026,0.00003387117,0.000046466,0.000038971346],"category_scores_gemma":[0.0004747867,0.00009398287,0.000048887614,0.01392168,0.00005289698,0.000101337464,0.000022691971,0.00011247207,0.0000050908047],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000027636543,0.0002903867,0.18918914,0.00013784779,0.000028405708,0.00000871141,0.00040367068,1.4585112e-7,0.000001947283,0.000014113961,0.0012615126,0.8086365],"study_design_scores_gemma":[0.0023896408,0.00040300546,0.9937229,0.00014241303,0.000011361339,0.00003626787,0.0012215822,0.000014158432,4.063622e-7,0.00013568022,0.0018327945,0.00008974958],"about_ca_topic_score_codex":0.00037090943,"about_ca_topic_score_gemma":0.00020251017,"teacher_disagreement_score":0.8085467,"about_ca_system_score_codex":0.00005248564,"about_ca_system_score_gemma":0.000035803987,"threshold_uncertainty_score":0.66889054},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W1840044458","doi":"10.1186/s12995-015-0076-3","title":"Telemedicine – a scientometric and density equalizing analysis","year":2015,"lang":"en","type":"article","venue":"Journal of Occupational Medicine and Toxicology","topic":"Telemedicine and Telehealth Implementation","field":"Medicine","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Telemedicine; Citation; Health informatics; Medicine; Informatics; Productivity; Citation analysis; Health care; Subject (documents); Quality (philosophy); Web of science; Library science; Data science; Public health; Computer science; Nursing; Pathology; Political science; Economic growth","score_opus":0.20668197869672694,"score_gpt":0.4726709086051637,"score_spread":0.26598892990843676,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1840044458","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9803091,0.0020199583,0.0046735215,0.01218863,0.0003034104,0.00017865385,0.0000024853657,0.000009640392,0.00031454657],"genre_scores_gemma":[0.99075323,0.00026379307,0.0029027162,0.005181014,0.0008079668,0.0000025672039,0.000021989947,0.0000070021374,0.00005974622],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.997956,0.000095147116,0.00083017774,0.00017701047,0.00072072895,0.00022098032],"domain_scores_gemma":[0.9974152,0.0004493192,0.0005330232,0.00010046168,0.0008495906,0.0006524024],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002145327,0.00013234121,0.00071590464,0.0028852185,0.000074084615,0.0000071951026,0.00005188278,0.00008079487,0.000115934294],"category_scores_gemma":[0.0025394282,0.00008974798,0.0000556057,0.002156717,0.00027396658,0.00013406816,0.000036382517,0.00023530278,0.0000011955384],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009581777,0.00011074897,0.96730477,0.00007296856,0.00053798215,0.00016914717,0.0013328147,0.0000035552337,0.0007946684,0.0010159573,0.009570976,0.018128246],"study_design_scores_gemma":[0.007246577,0.005420441,0.9713911,0.00006553316,0.0019250353,0.0020577188,0.0020016814,0.00030738153,0.00008168478,0.0005072725,0.008911799,0.000083815285],"about_ca_topic_score_codex":0.00006891307,"about_ca_topic_score_gemma":0.000035210385,"teacher_disagreement_score":0.01804443,"about_ca_system_score_codex":0.00008822936,"about_ca_system_score_gemma":0.000355622,"threshold_uncertainty_score":0.36598164},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W1859296017","doi":"10.1186/1471-2458-6-301","title":"A bibliometric analysis in the fields of preventive medicine, occupational and environmental medicine, epidemiology, and public health","year":2006,"lang":"en","type":"article","venue":"BMC Public Health","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":94,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Public health; Biostatistics; Epidemiology; Medicine; Preventive healthcare; Environmental medicine; Environmental epidemiology; Environmental health; Bibliometrics; Occupational medicine; Population; Family medicine; Library science; Pathology","score_opus":0.5681965202413667,"score_gpt":0.575281995445556,"score_spread":0.00708547520418934,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1859296017","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.832119,0.022800308,0.040529143,0.10242229,0.00013630016,0.00066496857,0.0001253394,0.0000117620875,0.001190918],"genre_scores_gemma":[0.99332935,0.0028998505,0.0009440565,0.0024789681,0.00010109695,0.000021871996,0.00008669411,0.0000054548586,0.0001326834],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9849613,0.004650559,0.0025727216,0.0009934698,0.0055896477,0.0012323413],"domain_scores_gemma":[0.9731256,0.02392163,0.0011076037,0.0006741067,0.00040878417,0.00076230825],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.1320133,0.00018461826,0.0009974277,0.505332,0.00031306173,0.00021902959,0.0011563112,0.0001317246,0.0003684885],"category_scores_gemma":[0.055936996,0.00010144147,0.00010516129,0.710571,0.0010441912,0.0004971984,0.00032277554,0.00034479366,0.0000039985252],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004810743,0.0001658453,0.8950847,0.000018410303,0.000020907393,6.2568347e-7,0.00043538518,0.000006507757,8.5792584e-7,0.003511562,0.016887551,0.08386283],"study_design_scores_gemma":[0.0006068487,0.00044526058,0.9790449,0.000007198271,0.0000063394855,0.000008834256,0.00270815,0.003997907,1.261855e-7,0.0044030724,0.008692787,0.00007855061],"about_ca_topic_score_codex":0.012896754,"about_ca_topic_score_gemma":0.002666965,"teacher_disagreement_score":0.20523901,"about_ca_system_score_codex":0.0002059156,"about_ca_system_score_gemma":0.0009180653,"threshold_uncertainty_score":0.9936764},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1886607088","doi":"10.1016/j.respol.2015.03.003","title":"Knowledge stocks, knowledge flows and innovation: Evidence from matched patents and innovation panel data","year":2015,"lang":"en","type":"article","venue":"Research Policy","topic":"Firm Innovation and Growth","field":"Economics, Econometrics and Finance","cited_by":209,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Economic and Social Research Council; Queen's University; Queen's University Belfast","keywords":"Panel data; Business; Industrial organization; Stock (firearms); Knowledge flow; Knowledge management; Knowledge economy; Investment (military); Measure (data warehouse); Marketing; Economics; Econometrics; Computer science; Data mining","score_opus":0.6756483836744561,"score_gpt":0.4639262424223971,"score_spread":0.21172214125205902,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1886607088","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96615326,0.005118477,0.00079736183,0.006890927,0.00025186985,0.00050776376,0.0005106116,0.00006196247,0.019707762],"genre_scores_gemma":[0.9943405,0.00039773676,0.0007742619,0.00025457653,0.0006471721,0.000042853822,0.00031311414,0.00003107115,0.0031987631],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978738,0.00009865796,0.00077237806,0.0006748414,0.00013811045,0.00044223876],"domain_scores_gemma":[0.99759036,0.00025367746,0.00019837297,0.00081083114,0.0010043796,0.00014238864],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043593454,0.00016419576,0.00031313582,0.0030897546,0.00019164369,0.00027984093,0.00053237175,0.00015574838,0.00009235202],"category_scores_gemma":[0.006037281,0.0001810383,0.000011128525,0.008497719,0.00018095349,0.0009881628,0.0009497867,0.00039512818,0.00071060145],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006818076,0.0001964355,0.08947204,0.00013397966,0.000067499044,0.0000040639275,0.0040013,0.0000013248358,0.0005884174,0.82383144,0.054422546,0.027212767],"study_design_scores_gemma":[0.003104143,0.00028826448,0.27025878,0.00028165898,0.000006257429,0.000012650032,0.0007205748,0.02058598,0.00039696213,0.3053985,0.39817524,0.0007710076],"about_ca_topic_score_codex":0.0017434168,"about_ca_topic_score_gemma":0.000101196056,"teacher_disagreement_score":0.518433,"about_ca_system_score_codex":0.00022276556,"about_ca_system_score_gemma":0.00040907058,"threshold_uncertainty_score":0.9133581},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W1941175482","doi":"10.1002/asi.23418","title":"Dimensions and uncertainties of author citation rankings: Lessons learned from frequency‐weighted in‐text citation counting","year":2015,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Citation; Weighting; Computer science; Rank (graph theory); Citation analysis; Citation impact; Field (mathematics); Subject (documents); Information retrieval; Statistics; Scheme (mathematics); Mathematics; Data science; Library science; Combinatorics; Physics","score_opus":0.36826216302652676,"score_gpt":0.492080593318006,"score_spread":0.12381843029147926,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1941175482","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97971344,0.00015134098,0.0017009008,0.017337736,0.000416501,0.00016871505,0.00001987696,0.0000062765516,0.00048521277],"genre_scores_gemma":[0.99776316,0.00008585454,0.0020068532,0.00007488667,0.000012463387,0.000003389838,0.0000012972533,0.0000016572725,0.00005042363],"study_design_codex":"observational","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9944927,0.00007985512,0.0009347575,0.00012956274,0.0041549196,0.00020819307],"domain_scores_gemma":[0.98239565,0.0020309843,0.0021307785,0.00016553087,0.013195981,0.00008105476],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.0318094,0.00006322107,0.00021703416,0.023174819,0.00027664736,0.000629636,0.00074256794,0.0001361866,0.0000025326103],"category_scores_gemma":[0.13454843,0.00004007589,0.000040128063,0.05246556,0.00030477426,0.0034245087,0.00019145884,0.00019883066,0.0000025294662],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007602598,0.0000676298,0.63326037,0.000011942812,0.000038395472,3.7403402e-7,0.009836748,0.00028045572,0.011132689,0.0936509,0.0029411574,0.24870333],"study_design_scores_gemma":[0.0022606482,0.0002984042,0.24591257,0.000060020477,0.000026023768,0.00000800958,0.023995452,0.027251286,0.005107863,0.68702215,0.007903318,0.00015425222],"about_ca_topic_score_codex":0.00006164548,"about_ca_topic_score_gemma":0.000024080437,"teacher_disagreement_score":0.5933713,"about_ca_system_score_codex":0.00032602894,"about_ca_system_score_gemma":0.0005737058,"threshold_uncertainty_score":0.996956},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1941622179","doi":"10.1111/j.1365-2648.2010.05429.x","title":"Research citation analysis of nursing academics in Canada: identifying success indicators","year":2010,"lang":"en","type":"article","venue":"Journal of Advanced Nursing","topic":"Academic Writing and Publishing","field":"Arts and Humanities","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Manitoba; CancerCare Manitoba","funders":"","keywords":"Scopus; Citation; Citation analysis; Nursing research; Bibliometrics; Summative assessment; Medicine; Subject (documents); Library science; Medical education; Nursing; MEDLINE; Psychology; Political science; Computer science; Pedagogy","score_opus":0.08563898176354096,"score_gpt":0.384438687922677,"score_spread":0.29879970615913604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1941622179","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99169374,0.00042406577,0.00015793087,0.0011417323,0.0012713622,0.00004967074,0.0000075347125,0.000005066141,0.0052488823],"genre_scores_gemma":[0.99852145,0.000043866035,0.000649431,0.00003197661,0.00068312354,6.411322e-7,0.000004363534,0.000013911212,0.00005125847],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.99771345,0.00013683311,0.0008196898,0.00013392727,0.00086064124,0.00033546827],"domain_scores_gemma":[0.9979902,0.00054558547,0.0007626659,0.000105532956,0.00048439464,0.000111650894],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023373505,0.00009579013,0.00036536614,0.0025903517,0.00024004986,0.00012727788,0.0003601243,0.00009443347,0.00008501637],"category_scores_gemma":[0.0006717645,0.00009221588,0.00009689774,0.0011549337,0.00024503944,0.001231457,0.000012214095,0.0022304077,2.4322793e-7],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028072682,0.00025522456,0.06440661,0.00011799891,0.0004158328,0.000045703695,0.22488397,0.014020813,0.015094389,0.061014086,0.0042164307,0.6152482],"study_design_scores_gemma":[0.0018161125,0.00016676703,0.2109725,0.0070390543,0.0010601799,0.00006114073,0.73156095,0.005243616,0.007818104,0.022604886,0.010915418,0.0007412795],"about_ca_topic_score_codex":0.033314332,"about_ca_topic_score_gemma":0.1948358,"teacher_disagreement_score":0.6145069,"about_ca_system_score_codex":0.0007168347,"about_ca_system_score_gemma":0.00069028226,"threshold_uncertainty_score":0.9731229},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","scholarly_communication"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W1950766814","doi":"10.1002/asi.23575","title":"The power–law relationship between citation‐based performance and collaboration in articles in management journals: A scale‐independent approach","year":2015,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"Universidad del Norte; Universidad Católica del Norte","keywords":"Citation; Citation analysis; Citation impact; Scale (ratio); Bibliometrics; Computer science; Library science; Geography","score_opus":0.3021124265811342,"score_gpt":0.4858847286269922,"score_spread":0.18377230204585804,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1950766814","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9840862,0.000102711296,0.0014046807,0.011857731,0.00018716363,0.0003404085,0.00000393857,0.0000042383513,0.0020129366],"genre_scores_gemma":[0.9987358,0.000032641703,0.0010791075,0.000074117335,0.000008160502,0.0000110329665,3.1349376e-7,0.0000014243044,0.000057435434],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9940333,0.000113612376,0.0009476248,0.00011575942,0.0045367843,0.00025297122],"domain_scores_gemma":[0.99156094,0.001606213,0.0011751462,0.00018521819,0.0053765927,0.00009586407],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.061516386,0.00005915736,0.00014440257,0.017902134,0.00052335917,0.0017640183,0.000948335,0.00009748955,4.4259815e-7],"category_scores_gemma":[0.038114842,0.000033045573,0.000025459314,0.06636781,0.0002691578,0.0036003564,0.00018658588,0.000261913,0.000002831572],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001774871,0.000016026237,0.96886957,0.0000033384495,0.000002892568,7.700751e-8,0.0006470917,0.00040603313,0.000009324293,0.014455386,0.00033781608,0.0152347125],"study_design_scores_gemma":[0.0011253327,0.00010420992,0.94006413,0.000015701331,0.0000047722533,0.0000047242274,0.008176398,0.00959987,0.00017069066,0.036786642,0.0038945328,0.00005301376],"about_ca_topic_score_codex":0.000004321991,"about_ca_topic_score_gemma":0.00002564278,"teacher_disagreement_score":0.048465677,"about_ca_system_score_codex":0.0005847867,"about_ca_system_score_gemma":0.0003572362,"threshold_uncertainty_score":0.9992722},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W1953706973","doi":"10.1007/s11192-016-2157-1","title":"Neophilia ranking of scientific journals","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":18,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Institute on Aging; University of Illinois at Chicago","keywords":"Ranking (information retrieval); Journal ranking; Information retrieval; Scientific literature; Computer science; Library science; Data science; Political science; Citation; Geology","score_opus":0.7322136034264928,"score_gpt":0.6133548832507849,"score_spread":0.11885872017570787,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1953706973","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95462674,0.0021876637,0.02599094,0.0015964389,0.0042207,0.00028231234,0.000101821825,0.000043757842,0.01094963],"genre_scores_gemma":[0.9864747,0.00018074313,0.0018732687,0.00005303145,0.00011506751,0.000004282975,8.5659644e-7,0.000015671732,0.011282366],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9605085,0.00035390694,0.0017099311,0.0013636119,0.034763012,0.0013010231],"domain_scores_gemma":[0.96909523,0.011056822,0.0010530971,0.0021534045,0.01551424,0.001127233],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.10989885,0.00022311913,0.0005636096,0.3311329,0.00060516875,0.0046070055,0.0061568823,0.00015952627,0.0029112697],"category_scores_gemma":[0.23368925,0.000117548196,0.00038906085,0.6913339,0.00126198,0.0017193832,0.0014450144,0.00022139132,0.0012319804],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016902004,0.00016456137,0.27150992,0.000005237749,0.000013393643,0.000009699492,0.00011183134,0.000008658365,0.04910531,0.0023077675,0.015953306,0.6607934],"study_design_scores_gemma":[0.002352449,0.00039873007,0.68733615,0.000114950424,0.000019723577,0.00003949886,0.00025834888,0.000977283,0.06320393,0.06730294,0.17730926,0.00068673596],"about_ca_topic_score_codex":0.000013861546,"about_ca_topic_score_gemma":0.000004497225,"teacher_disagreement_score":0.66010666,"about_ca_system_score_codex":0.00023739987,"about_ca_system_score_gemma":0.0005451105,"threshold_uncertainty_score":0.9995457},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1955009066","doi":"10.1186/s40064-015-1431-7","title":"Global methaemoglobinaemia research output (1940–2013): a bibliometric analysis","year":2015,"lang":"en","type":"article","venue":"SpringerPlus","topic":"Methemoglobinemia and Tumor Lysis Syndrome","field":"Medicine","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"An-Najah National University","keywords":"Scopus; Bibliometrics; Library science; Medicine; MEDLINE; Political science; Computer science","score_opus":0.14963565944774557,"score_gpt":0.4095609136670474,"score_spread":0.2599252542193018,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1955009066","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93795073,0.0027930546,0.0012259413,0.0016004565,0.0004237173,0.0005038595,0.00002147386,0.00024437852,0.055236366],"genre_scores_gemma":[0.9867684,0.0001150106,0.0057771,0.00016101506,0.00029807485,0.00004953903,0.000032882323,0.000039824372,0.0067581087],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99456745,0.000378066,0.0006415928,0.00088768435,0.0024270131,0.0010981834],"domain_scores_gemma":[0.9956906,0.00020591616,0.00015206505,0.0014580557,0.0012061681,0.001287231],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.004452109,0.00034842113,0.0011625069,0.06560248,0.00016501131,0.00012246224,0.00051543326,0.00025924479,0.00039227415],"category_scores_gemma":[0.002105988,0.00029728512,0.0005658652,0.32173017,0.00024989466,0.00016818523,0.000407489,0.0005567762,0.0017571293],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005959656,0.0007776847,0.9114542,0.00018610277,0.003741152,0.00234718,0.0004150188,0.000320153,0.0004301447,0.0021693637,0.051173426,0.02638959],"study_design_scores_gemma":[0.0047109416,0.000932374,0.9052509,0.000084581596,0.0033130932,0.00080215215,0.00043308377,0.0026274805,0.00086494454,0.002452596,0.07782402,0.00070383796],"about_ca_topic_score_codex":0.002328841,"about_ca_topic_score_gemma":0.00011011603,"teacher_disagreement_score":0.2561277,"about_ca_system_score_codex":0.0011272042,"about_ca_system_score_gemma":0.00064063683,"threshold_uncertainty_score":0.9999479},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W1967988001","doi":"10.1002/asi.10401","title":"A classification of author co‐citations: Definitions and search strategies","year":2004,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":63,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Citation; Computer science; Ranking (information retrieval); Information retrieval; Dialog box; Field (mathematics); Set (abstract data type); Classification scheme; Library science; Mathematics; World Wide Web","score_opus":0.45324412709551765,"score_gpt":0.5484433231515129,"score_spread":0.09519919605599525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1967988001","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96094525,0.00008379877,0.024614703,0.013751134,0.000065252054,0.00015917709,0.000023453485,0.0000060533584,0.00035117418],"genre_scores_gemma":[0.98614454,0.00031252528,0.013363775,0.00015798466,0.000007749282,0.0000047102185,3.4514304e-7,0.0000013302356,0.00000702797],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9963189,0.000021353077,0.0006214915,0.00010629703,0.0027354306,0.00019656702],"domain_scores_gemma":[0.9907926,0.00078761356,0.0010057655,0.00022200921,0.0071059414,0.00008607421],"candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.012270492,0.000053832413,0.00017913293,0.01076668,0.0005112202,0.00067992194,0.0010467438,0.000043637745,0.0000016581114],"category_scores_gemma":[0.012402558,0.000031327076,0.00010440448,0.07291707,0.0038212745,0.0026941649,0.00017467282,0.00017843646,0.0000019433635],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003956971,0.00011054745,0.024547795,0.00002833414,0.000047181053,1.8802581e-7,0.004337339,0.00029357648,0.033727318,0.6107148,0.003863125,0.32229024],"study_design_scores_gemma":[0.0015326502,0.0015163083,0.37943402,0.000041181942,0.000031961285,0.00017372865,0.23968054,0.005978982,0.009196152,0.34677935,0.015424814,0.00021030642],"about_ca_topic_score_codex":0.000011595512,"about_ca_topic_score_gemma":0.0000013388473,"teacher_disagreement_score":0.35488623,"about_ca_system_score_codex":0.00008949387,"about_ca_system_score_gemma":0.0010846596,"threshold_uncertainty_score":0.99888974},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1971715723","doi":"10.1016/j.juro.2009.02.066","title":"Publications by the American Academy of Pediatrics Section on Urology: The Quality of Research Design and Statistical Methodology","year":2009,"lang":"en","type":"article","venue":"The Journal of Urology","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Medicine; Section (typography); Pediatric urology; Pediatrics; Urology","score_opus":0.914132387819782,"score_gpt":0.6758417498767043,"score_spread":0.2382906379430777,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1971715723","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3663084,0.0020775984,0.50263387,0.12802769,0.00010086397,0.0004393321,0.0000127439025,0.0000012862766,0.00039824186],"genre_scores_gemma":[0.99388546,0.0004677318,0.003691602,0.0016835283,0.00011262736,0.000003044516,2.7230666e-7,0.0000032435041,0.00015247377],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.84493786,0.14726461,0.004897461,0.00023826424,0.0024290737,0.00023274365],"domain_scores_gemma":[0.8433903,0.14715041,0.0072456663,0.0009388722,0.0011997038,0.0000750844],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.39098138,0.0001036155,0.0015353347,0.00036362646,0.00020470742,0.000043737356,0.0017338331,0.00008702098,0.00025224674],"category_scores_gemma":[0.096058115,0.000032209737,0.00026163695,0.0014466727,0.0011544002,0.000057513225,0.000073968695,0.0008795196,0.00001322249],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023490894,0.0005150898,0.016708203,0.000031839543,0.00064293685,0.0000017423773,0.0044727647,0.0033017634,0.008215416,0.22427967,0.63923126,0.10025023],"study_design_scores_gemma":[0.00049758045,0.0050471034,0.5033249,0.000002367257,0.001116904,0.0004413449,0.002158062,0.0032773134,0.00015541464,0.4399794,0.043872137,0.0001274036],"about_ca_topic_score_codex":0.000018768711,"about_ca_topic_score_gemma":0.0000035652108,"teacher_disagreement_score":0.6275771,"about_ca_system_score_codex":0.000011280752,"about_ca_system_score_gemma":0.00008177182,"threshold_uncertainty_score":0.9115562},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W1977960824","doi":"10.1016/j.evalprogplan.2014.04.003","title":"Taking stock of four decades of quantitative research on stakeholder participation and evaluation use: A systematic map","year":2014,"lang":"en","type":"review","venue":"Evaluation and Program Planning","topic":"Evaluation and Performance Assessment","field":"Decision Sciences","cited_by":33,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ministry of Labour, Employment and Social Solidarity","funders":"","keywords":"Stakeholder; Stock (firearms); Stakeholder engagement; Stakeholder analysis; Environmental planning; Business; Environmental resource management; Geography; Political science; Environmental science; Public relations; Archaeology","score_opus":0.9237604363882755,"score_gpt":0.7285006439353664,"score_spread":0.19525979245290914,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1977960824","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023071304,0.96526396,0.00080074405,0.00006319011,0.00026833685,0.009801521,0.00002095147,0.000029457215,0.000680513],"genre_scores_gemma":[0.6417393,0.34407473,0.007533412,0.00004231188,0.000150695,0.005765997,0.00032544858,0.00008691047,0.0002812028],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9795019,0.010037118,0.0024337198,0.0007033616,0.007007329,0.00031656332],"domain_scores_gemma":[0.9822555,0.01027158,0.0030675025,0.00054332876,0.0037525916,0.0001094722],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.0759058,0.0003095739,0.0015842564,0.0013606582,0.00028108776,0.00042127052,0.00028033357,0.00024635863,0.00021592421],"category_scores_gemma":[0.015474351,0.00021506031,0.00014398819,0.0012187039,0.00019229874,0.00046487772,0.000090378,0.00038501495,0.00002382339],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000031918087,0.0000818664,0.0007044342,0.028520126,0.00011076365,2.9086834e-7,0.0023368758,0.0008881291,0.0000011554752,0.001386885,0.00012057815,0.965817],"study_design_scores_gemma":[0.0019669274,0.003030404,0.010977639,0.22636166,0.0030038722,0.000010988403,0.0046628374,0.7026354,0.000011465294,0.0033876577,0.04319855,0.0007526116],"about_ca_topic_score_codex":0.000016676016,"about_ca_topic_score_gemma":0.000018185847,"teacher_disagreement_score":0.96506435,"about_ca_system_score_codex":0.00013199577,"about_ca_system_score_gemma":0.00067683065,"threshold_uncertainty_score":0.9928187},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W1979812464","doi":"10.1007/s11192-010-0249-x","title":"Mathematics 1868–2008: a bibliometric analysis","year":2010,"lang":"en","type":"article","venue":"Scientometrics","topic":"Complex Systems and Time Series Analysis","field":"Economics, Econometrics and Finance","cited_by":43,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Productivity; Distribution (mathematics); Range (aeronautics); Mathematics; Bibliometrics; Quarter (Canadian coin); Statistics; Mathematical economics; Econometrics; History; Library science; Economics; Computer science; Economic growth","score_opus":0.040057207972988534,"score_gpt":0.25840684375512885,"score_spread":0.2183496357821403,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1979812464","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9140246,0.0018326128,0.027156983,0.00018858047,0.0013135353,0.00021397544,0.00016953796,0.00013198027,0.054968186],"genre_scores_gemma":[0.9838277,0.000079920406,0.010160224,0.00007116372,0.0001453048,0.000015267347,0.000021576303,0.000027737155,0.0056510884],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9972323,0.000010455689,0.0010696267,0.0007236893,0.00034513907,0.0006187649],"domain_scores_gemma":[0.99739295,0.00019287682,0.00065319816,0.0011489186,0.00026616923,0.00034587234],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.002787555,0.00023873821,0.00084277894,0.35937294,0.00027297597,0.000619211,0.0008242433,0.00015718208,0.0049384646],"category_scores_gemma":[0.0014642386,0.00024934788,0.0006385334,0.7178043,0.0001283555,0.0003612175,0.0002095707,0.0003133886,0.0020299936],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000022488691,0.0004435472,0.81135297,0.00004496218,0.0010807603,0.000013981279,0.0005758913,0.00019851445,0.00013491839,0.17545484,0.0026541925,0.008043173],"study_design_scores_gemma":[0.0007733971,0.00011396879,0.5448301,0.000006138839,0.00043726133,0.000029346105,0.0004539177,0.059397835,0.00008568605,0.012847179,0.379807,0.00121818],"about_ca_topic_score_codex":0.0012071561,"about_ca_topic_score_gemma":0.00030495,"teacher_disagreement_score":0.3771528,"about_ca_system_score_codex":0.00007385346,"about_ca_system_score_gemma":0.000029264069,"threshold_uncertainty_score":0.9999959},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W1980121455","doi":"10.1108/14691930910922860","title":"A scientometric analysis of the Proceedings of the McMaster World Congress on the Management of Intellectual Capital and Innovation for the 1996‐2008 period","year":2009,"lang":"en","type":"article","venue":"Journal of Intellectual Capital","topic":"Intellectual Capital and Performance Analysis","field":"Business, Management and Accounting","cited_by":70,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McMaster University; Lakehead University","funders":"","keywords":"Originality; Intellectual capital; Productivity; Scientometrics; Delegate; Sociology; Identity (music); Discipline; Management; Social science; Library science; Political science; Qualitative research; Economics; Economic growth; Law; Computer science","score_opus":0.02514335089884866,"score_gpt":0.2437797157616595,"score_spread":0.21863636486281082,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1980121455","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9939413,0.00048399225,0.00009238772,0.0016656711,0.0004222798,0.0005029438,0.000010333609,0.0000053457143,0.0028757325],"genre_scores_gemma":[0.9973821,0.00009089664,0.000031482497,0.0006783759,0.00024480667,0.000008567539,0.0000020776413,0.0000153562,0.0015463438],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9972329,0.000020907684,0.0011682373,0.00022988899,0.00104121,0.0003068276],"domain_scores_gemma":[0.9955868,0.0007388997,0.0015422635,0.0002508588,0.0018648312,0.000016328202],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015197318,0.00028320507,0.000573695,0.0032267633,0.00040406338,0.00021767171,0.0011015955,0.000068028436,0.000624358],"category_scores_gemma":[0.0014113771,0.00012228142,0.0007037119,0.016669208,0.00055745395,0.00051130133,0.00027608883,0.0004036151,0.0000052057794],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.015002136,0.0051210253,0.062166605,0.0034105883,0.043737546,0.000013035271,0.1815277,0.007344487,0.031888645,0.31611136,0.20705953,0.12661733],"study_design_scores_gemma":[0.009141067,0.0066099614,0.35068938,0.003965783,0.038352985,0.00019205901,0.2495855,0.1425439,0.1325788,0.014653534,0.048444416,0.0032426307],"about_ca_topic_score_codex":0.00003242604,"about_ca_topic_score_gemma":0.00003937726,"teacher_disagreement_score":0.30145782,"about_ca_system_score_codex":0.000062413776,"about_ca_system_score_gemma":0.000043968863,"threshold_uncertainty_score":0.80090016},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1980831228","doi":"10.3152/147154302781780822","title":"Do US Congressional earmarks increase research output at universities?","year":2002,"lang":"en","type":"article","venue":"Science and Public Policy","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Appropriation; Competition (biology); Instrumental variable; Quality (philosophy); Estimation; Political science; Public economics; Economics; Public administration; Business; Management; Econometrics","score_opus":0.6548266178634141,"score_gpt":0.590624394806003,"score_spread":0.06420222305741108,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1980831228","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.77392673,0.0006565427,0.000010981187,0.011295636,0.00021962292,0.00013449599,0.00004445988,0.00002046044,0.21369104],"genre_scores_gemma":[0.9613104,0.00041835805,0.000085126434,0.0007190213,0.00020773809,0.0000048204183,0.0000012388548,0.0000061988026,0.037247106],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9757745,0.00046985075,0.00039780253,0.0011135534,0.02043226,0.0018119998],"domain_scores_gemma":[0.98363155,0.0039607356,0.00014836044,0.0010905565,0.008699827,0.0024689394],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","sts","insufficient_payload"],"category_scores_codex":[0.052281838,0.00015079412,0.00022732113,0.09847669,0.0029662116,0.007179978,0.004241098,0.00011630768,0.0034045302],"category_scores_gemma":[0.06596924,0.00010391794,0.00006928962,0.26478153,0.005034984,0.002681775,0.00423573,0.00045655677,0.0013254825],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000028435481,0.00024439764,0.16653013,0.000005760507,0.0000091357615,0.00006726114,0.0008231889,0.00000314416,0.00087201747,0.06137651,0.43347433,0.33656567],"study_design_scores_gemma":[0.0007052226,0.00018778208,0.17054522,0.000010150861,0.0000016015248,0.00007378275,0.0023383277,0.008825029,0.00018110436,0.008144483,0.80869436,0.00029291603],"about_ca_topic_score_codex":0.001931156,"about_ca_topic_score_gemma":0.000035126486,"teacher_disagreement_score":0.37522003,"about_ca_system_score_codex":0.000911843,"about_ca_system_score_gemma":0.002710685,"threshold_uncertainty_score":0.9994521},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W1981886524","doi":"10.1002/asi.22688","title":"<scp>SciMAT</scp>: A new science mapping analysis software tool","year":2012,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":1187,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Universidad de Granada; McGill University","keywords":"Wizard; Workflow; Computer science; Software; Key (lock); Preprocessor; Software engineering; Data science; World Wide Web; Database; Programming language; Operating system","score_opus":0.05196923175334746,"score_gpt":0.34842610769681986,"score_spread":0.2964568759434724,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1981886524","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7817136,0.00007199423,0.21380046,0.0031523574,0.0007336466,0.00016806807,0.00000909911,0.00003898368,0.00031176512],"genre_scores_gemma":[0.93046427,0.000021294381,0.06772479,0.0014787924,0.00007412121,0.0000028777247,4.5951168e-7,0.000002543762,0.00023086915],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9956865,0.00001783388,0.0008506853,0.00023135736,0.002671872,0.00054178364],"domain_scores_gemma":[0.994259,0.0005955732,0.0020256003,0.00080682564,0.002126514,0.00018647901],"candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.019403324,0.00010858652,0.00032370916,0.0022618074,0.0012539855,0.0009835976,0.0031309794,0.00003275443,0.0000032909659],"category_scores_gemma":[0.020198619,0.00006406442,0.00032521924,0.030768927,0.0038255525,0.0058855163,0.0010900729,0.00018135575,0.000014684745],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000048354023,0.0000452368,0.088010944,0.000008388066,0.00015002846,1.4245477e-7,0.0069285035,0.00047346667,0.0013672131,0.008810039,0.13039921,0.763802],"study_design_scores_gemma":[0.00058431167,0.00019371112,0.14266223,0.000027441036,0.00024823574,0.00011806476,0.074628584,0.022503618,0.0016434115,0.011095625,0.7461097,0.00018503143],"about_ca_topic_score_codex":0.000011367391,"about_ca_topic_score_gemma":7.47178e-7,"teacher_disagreement_score":0.763617,"about_ca_system_score_codex":0.0001522741,"about_ca_system_score_gemma":0.0007120204,"threshold_uncertainty_score":0.99888545},"labels":[{"model":"gemma","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"simulation_or_modeling","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1986314829","doi":"10.1136/oemed-2013-101717.180","title":"180 Occupational epidemiology: A bibliometric analysis by country and era","year":2013,"lang":"en","type":"article","venue":"Occupational and Environmental Medicine","topic":"Occupational Health and Safety Research","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Epidemiology; Publishing; Population; Demography; Geography; Medicine; History; Library science; Political science; Sociology; Law; Pathology; Computer science","score_opus":0.07153965856317186,"score_gpt":0.4395871186156648,"score_spread":0.36804746005249295,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1986314829","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9855475,0.003393421,0.00087171164,0.0063348445,0.00017178798,0.00089315913,0.0006696998,0.000030016403,0.002087847],"genre_scores_gemma":[0.98614466,0.0022218858,0.00057514064,0.0053086476,0.0003678374,0.00038450092,0.0037687323,0.000014150856,0.0012144467],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9972097,0.0004218347,0.0007669564,0.000453169,0.00060729653,0.00054101105],"domain_scores_gemma":[0.99504864,0.0039250343,0.00022525934,0.00018860865,0.00005660673,0.0005558444],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0012817122,0.0002078424,0.00047882728,0.0065138307,0.00080203085,0.000005815448,0.000100824014,0.00020273717,0.0092116995],"category_scores_gemma":[0.00078158965,0.00015565798,0.000040455056,0.008897945,0.00050175795,0.0002369683,0.00011753499,0.0004782253,0.00027997018],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015055829,0.000056152832,0.9569162,0.0000654277,0.00010243641,6.676573e-7,0.00007622142,0.000004709814,0.000112898,0.0007316709,0.026303302,0.015479731],"study_design_scores_gemma":[0.001067714,0.00017505555,0.98094976,0.000028747576,0.00008390373,0.0000033413457,0.00023732234,0.0026951462,7.54042e-7,0.00075350975,0.013865132,0.000139642],"about_ca_topic_score_codex":0.002315112,"about_ca_topic_score_gemma":0.00004343593,"teacher_disagreement_score":0.024033507,"about_ca_system_score_codex":0.00014498991,"about_ca_system_score_gemma":0.0000946973,"threshold_uncertainty_score":0.99169403},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1986726592","doi":"10.3152/147154300781782147","title":"Impact of collaborative research on academic science","year":2000,"lang":"en","type":"article","venue":"Science and Public Policy","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":119,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Institut National de la Recherche Scientifique; Université du Québec à Montréal","funders":"","keywords":"Research policy; Political science; Public relations; Sociology; Public administration","score_opus":0.7446807243754998,"score_gpt":0.7166696462433838,"score_spread":0.028011078132116007,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1986726592","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8832105,0.0001906876,0.000004783341,0.0034694844,0.00007329052,0.00021595467,0.00004289124,0.000011686706,0.11278072],"genre_scores_gemma":[0.99737036,0.000579678,0.00007398598,0.0001658549,0.0001161199,0.0000082525285,3.0186686e-7,0.0000048949482,0.0016805523],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9689531,0.00030602855,0.00057451957,0.0012191444,0.027030293,0.0019169546],"domain_scores_gemma":[0.9721659,0.0037367318,0.0001906603,0.0013432,0.020586161,0.0019773922],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.11105411,0.00014526506,0.0002770204,0.14495985,0.0018360169,0.0040666913,0.006555762,0.00009848775,0.00070725393],"category_scores_gemma":[0.13985837,0.00008727179,0.000068768444,0.649469,0.0147226155,0.0034221602,0.00095330365,0.0006162042,0.00041343816],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000039880197,0.00015290863,0.06106233,0.0000017169568,0.0000040735035,0.0000023048633,0.0007125039,0.000025440495,0.010551439,0.071870156,0.008114133,0.84746313],"study_design_scores_gemma":[0.00048204113,0.0015766177,0.9268859,0.000014586676,9.619396e-7,0.000014014526,0.0010894873,0.0032450343,0.0034062257,0.04104366,0.021991756,0.0002497009],"about_ca_topic_score_codex":0.00123317,"about_ca_topic_score_gemma":0.0000079212205,"teacher_disagreement_score":0.86582357,"about_ca_system_score_codex":0.00070396485,"about_ca_system_score_gemma":0.021440495,"threshold_uncertainty_score":0.99946344},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W1993126323","doi":"10.1556/scient.67.2006.3.7","title":"Indirect patent citations","year":2006,"lang":"en","type":"article","venue":"Scientometrics","topic":"Intellectual Property and Patents","field":"Business, Management and Accounting","cited_by":50,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Center for Interuniversity Research and Analysis on Organizations; University of Ottawa; Statistics Canada","funders":"","keywords":"Index (typography); Quality (philosophy); Citation; Measure (data warehouse); Variable (mathematics); Citation analysis; Computer science; Information retrieval; Mathematics; Data mining; Library science; Epistemology; World Wide Web; Philosophy","score_opus":0.2538094299036781,"score_gpt":0.25225322252204485,"score_spread":0.0015562073816332567,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1993126323","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.65749043,0.00017323902,0.0016621964,0.00035312562,0.0015614459,0.00020071633,0.0000039486076,0.00019504427,0.33835986],"genre_scores_gemma":[0.9935892,0.0000023164127,0.00018866209,0.00086963543,0.00051426445,0.000006574587,0.000037794587,0.000010852067,0.004780687],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99885416,0.0000049536993,0.00016593169,0.00019761099,0.0004992402,0.00027811292],"domain_scores_gemma":[0.99945766,0.000043348642,0.000077597215,0.00013030213,0.00028147266,0.000009628807],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0004973737,0.000090743066,0.00008620462,0.0027732493,0.0002672363,0.00052003236,0.00022334601,0.00003914983,0.0005256935],"category_scores_gemma":[0.0004841019,0.000072895076,0.000053705193,0.011634976,0.00004641608,0.0005555399,0.00009036941,0.00007825599,0.0030629076],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000051666084,0.0010064172,0.35918784,0.00021875136,0.000044850167,0.000039355607,0.00023108744,0.0016751662,0.0021916542,0.098125115,0.47813192,0.059096165],"study_design_scores_gemma":[0.00073713966,0.000026030088,0.2533346,0.000029691262,0.000048262602,0.0000025965946,0.00011084218,0.011467942,0.00078436657,0.011244012,0.72166157,0.000552959],"about_ca_topic_score_codex":0.0005490694,"about_ca_topic_score_gemma":0.000031676766,"teacher_disagreement_score":0.3360988,"about_ca_system_score_codex":0.00003257796,"about_ca_system_score_gemma":0.000012616178,"threshold_uncertainty_score":0.9977133},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W1993240968","doi":"10.1097/01.prs.0000304600.23129.d3","title":"Levels of Evidence in Plastic Surgery Research over 20 Years","year":2008,"lang":"en","type":"review","venue":"Plastic & Reconstructive Surgery","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":102,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Randomized controlled trial; Evidence-based medicine; Medicine; Placebo; Plastic surgery; Surgery; Alternative medicine","score_opus":0.9182824885556872,"score_gpt":0.572964135707008,"score_spread":0.3453183528486792,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1993240968","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":"methods","model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007730691,0.9852973,0.00030223632,0.0000068464733,0.004557755,0.0011610679,0.00043144263,0.000009133867,0.0005035118],"genre_scores_gemma":[0.0036801894,0.99454963,0.00028422408,0.0000051806924,0.00029682173,0.0002038075,0.000021548347,0.0000778796,0.0008807336],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9114874,0.044659525,0.027042875,0.003155317,0.012364309,0.0012905537],"domain_scores_gemma":[0.4535389,0.5319249,0.009204293,0.0032468848,0.0017755664,0.00030947474],"candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","insufficient_payload"],"consensus_categories":["metaresearch","insufficient_payload"],"category_scores_codex":[0.14799675,0.0010893827,0.024831012,0.008924003,0.00015846736,0.00043009766,0.0024332926,0.0006775048,0.013962706],"category_scores_gemma":[0.6369861,0.0005937038,0.008649891,0.0102240695,0.0011460674,0.0006672803,0.0005037706,0.0015624046,0.0042933044],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016938955,0.00006773239,0.027394768,0.0044187023,0.0006659739,0.00024170514,0.00009181951,0.000008670196,4.6835808e-7,0.00006081455,0.036450397,0.930582],"study_design_scores_gemma":[0.00011913212,0.000037549835,0.04601054,0.069767505,0.0009972451,0.00086909696,0.0002457093,0.00017622387,0.0000014093378,0.0011699075,0.87918997,0.0014157292],"about_ca_topic_score_codex":0.000053140262,"about_ca_topic_score_gemma":0.00007081865,"teacher_disagreement_score":0.92916626,"about_ca_system_score_codex":0.00039693105,"about_ca_system_score_gemma":0.004037193,"threshold_uncertainty_score":0.99965143},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"methods","study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W1999584965","doi":"10.1007/s11192-014-1258-y","title":"On the quality evaluation of scientific entities in Poland supported by consistency-driven pairwise comparisons method","year":2014,"lang":"en","type":"article","venue":"Scientometrics","topic":"Multi-Criteria Decision Making","field":"Decision Sciences","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Laurentian University","funders":"Dalhousie University; Akademia Górniczo-Hutnicza im. Stanislawa Staszica; Politechnika Wrocławska; Fayetteville State University; Rensselaer Polytechnic Institute","keywords":"Pairwise comparison; Ranking (information retrieval); Consistency (knowledge bases); Focus (optics); Computer science; Domain (mathematical analysis); Parametric statistics; Quality (philosophy); Event (particle physics); Information retrieval; Operations research; Management science; Data mining; Mathematics; Statistics; Artificial intelligence; Epistemology; Economics","score_opus":0.49126161277195085,"score_gpt":0.5458934524210635,"score_spread":0.05463183964911261,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1999584965","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95749176,0.00012342274,0.034427453,0.00052468263,0.0016090367,0.00054599316,0.00013416712,0.000023594186,0.0051199035],"genre_scores_gemma":[0.9937315,0.000001399746,0.005281521,0.0001506174,0.00001633525,0.000016219843,0.000015941285,0.000009464148,0.0007769921],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.97865915,0.0051880335,0.0017861543,0.0010412255,0.012817711,0.00050770584],"domain_scores_gemma":[0.96494836,0.02702789,0.001296949,0.0023101703,0.0041741105,0.00024252429],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.1432846,0.0001960841,0.00054223277,0.006494879,0.00043796634,0.0011395134,0.0017743977,0.00010333249,0.0012427135],"category_scores_gemma":[0.13953464,0.00012542662,0.00017351376,0.023135124,0.000710582,0.00034869972,0.00033962663,0.00023490204,0.00017739531],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013816619,0.0014017577,0.53961676,0.000030997217,0.000042153915,0.0000020205246,0.005575716,0.00575669,0.028588714,0.058735475,0.21237168,0.14773987],"study_design_scores_gemma":[0.0019602699,0.00012991039,0.40392196,0.00006194839,0.000044872235,0.0000023212715,0.002456371,0.52433676,0.003973226,0.04380644,0.018918071,0.00038786387],"about_ca_topic_score_codex":0.0001281927,"about_ca_topic_score_gemma":0.0002149756,"teacher_disagreement_score":0.51858,"about_ca_system_score_codex":0.00017371922,"about_ca_system_score_gemma":0.00031577362,"threshold_uncertainty_score":0.9998974},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2001213194","doi":"10.1007/s11192-010-0295-4","title":"Reasons for and developments in international scientific collaboration: does an Asia–Pacific research area exist from a bibliometric point of view?","year":2010,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":35,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Forschungszentrum Jülich; Bundesministerium für Bildung und Forschung","keywords":"China; Regional science; Strengths and weaknesses; Citation; Library science; Citation impact; Citation analysis; Quarter (Canadian coin); Bibliometrics; Scientometrics; Political science; Point (geometry); Geography; Computer science; Law; Psychology; Mathematics","score_opus":0.5650881049249804,"score_gpt":0.5968205699851685,"score_spread":0.03173246506018812,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2001213194","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9833241,0.0005808174,0.0014479689,0.0011715897,0.0051320046,0.0010336011,0.00081985816,0.000027412581,0.006462665],"genre_scores_gemma":[0.96881086,0.00022279513,0.028839646,0.000019878415,0.00011991809,0.00008564748,0.00009371714,0.000022370878,0.0017851877],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9738749,0.00045442168,0.0015780524,0.002153784,0.020705216,0.0012335865],"domain_scores_gemma":[0.959282,0.013308792,0.0006086323,0.0017252322,0.023922266,0.0011530975],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.10445518,0.00027000444,0.0005653087,0.75205237,0.0008316996,0.010614284,0.004762477,0.0002945154,0.00038835325],"category_scores_gemma":[0.19600609,0.00019275086,0.00011686228,0.90742224,0.0014585081,0.0025958137,0.0014709235,0.0007297994,0.000058443784],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000104961255,0.0009815342,0.77835584,0.000017766657,0.000024058038,0.000011200485,0.0013997176,0.0000057024963,0.022391524,0.0071590734,0.008865267,0.18068333],"study_design_scores_gemma":[0.0018494697,0.0003344311,0.82489455,0.000033791544,0.0000071375925,0.0000067617993,0.0050660954,0.00395097,0.0074052797,0.043177497,0.112805925,0.00046809696],"about_ca_topic_score_codex":0.00020445864,"about_ca_topic_score_gemma":0.0010398943,"teacher_disagreement_score":0.18021524,"about_ca_system_score_codex":0.00029817995,"about_ca_system_score_gemma":0.0012464792,"threshold_uncertainty_score":0.9904128},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2004098455","doi":"10.1007/s11192-013-1069-6","title":"Effect of collaboration network structure on knowledge creation and technological performance: the case of biotechnology in Canada","year":2013,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":84,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Concordia University","funders":"","keywords":"Network structure; Productivity; Quality (philosophy); Production (economics); Knowledge management; Field (mathematics); Network analysis; Biotechnology; Business; Computer science; Data science; Engineering; Biology; Economics","score_opus":0.11364746930551028,"score_gpt":0.46945445948011494,"score_spread":0.35580699017460465,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2004098455","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9970747,0.00081252377,0.00008144907,0.00050734694,0.00030354943,0.0005588941,0.000025069447,0.000007662398,0.0006287817],"genre_scores_gemma":[0.9995803,0.00018631513,0.00012709483,0.000014592612,0.000013313213,0.000012615712,0.000001521015,0.0000040527066,0.000060164373],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9953133,0.00035616598,0.0006361364,0.00045993508,0.0027985687,0.00043588757],"domain_scores_gemma":[0.9895739,0.0077330503,0.00034464564,0.00060070213,0.0016248319,0.00012286937],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.01540162,0.00013040802,0.00034576523,0.050631985,0.00017534694,0.000285279,0.0011286217,0.00019247769,0.00009330339],"category_scores_gemma":[0.034254946,0.00006740615,0.000030724907,0.43146506,0.00055149745,0.00024350305,0.00044861136,0.00030889813,0.000007501409],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000022861559,0.00003863651,0.2841277,0.00002394552,0.0000051092957,0.000011504054,0.00006510794,0.0009922805,0.0009183518,0.0016643254,0.0018649116,0.7102652],"study_design_scores_gemma":[0.001534187,0.0029500036,0.74033695,0.000049916896,0.000016602158,0.0001227433,0.0017516516,0.15079904,0.09668067,0.0025941497,0.0028054256,0.0003586424],"about_ca_topic_score_codex":0.053702876,"about_ca_topic_score_gemma":0.07347152,"teacher_disagreement_score":0.7099066,"about_ca_system_score_codex":0.00032988153,"about_ca_system_score_gemma":0.0005209459,"threshold_uncertainty_score":0.97387993},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2012047041","doi":"10.1016/j.healthpol.2010.03.001","title":"The evolution of smoke-free spaces policy literature: A bibliometric analysis","year":2010,"lang":"en","type":"review","venue":"Health Policy","topic":"Smoking Behavior and Cessation","field":"Medicine","cited_by":18,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo; University of Alberta","funders":"","keywords":"Sophistication; Bibliometrics; Publishing; Citation analysis; Public policy; Citation; Tobacco control; Social science; Regional science; Library science; Political science; Sociology; Public health; Medicine; Computer science; Law","score_opus":0.07791316023220482,"score_gpt":0.46787137769055653,"score_spread":0.3899582174583517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2012047041","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0002603261,0.9918293,0.000028860368,0.0057228734,0.00025007207,0.0009183802,0.00022095513,0.00006174268,0.00070754235],"genre_scores_gemma":[0.0020211106,0.994414,0.0000448607,0.00035488827,0.001890465,0.00010085974,0.00030146103,0.000051180323,0.00082117895],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9967611,0.00035088987,0.0010855666,0.00039260913,0.00079427543,0.0006155981],"domain_scores_gemma":[0.9960326,0.0003646561,0.0012845071,0.0016296863,0.00033745912,0.00035107904],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0013648019,0.00037918196,0.0019581993,0.13617733,0.00028756447,0.0001027043,0.00044944268,0.0006466219,0.000018498225],"category_scores_gemma":[0.0020684223,0.00023334032,0.0009826397,0.32269442,0.00014579634,0.00007069811,0.00010446599,0.0010819832,0.000005777109],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000008037341,0.000076034084,0.0018162583,0.0100579085,0.00031389712,0.0000019886609,0.00017923408,9.688301e-8,1.334882e-7,0.0054009827,0.0019939584,0.9801515],"study_design_scores_gemma":[0.0003092044,0.00016270175,0.04400366,0.0038752283,0.0028180892,0.000057781832,0.000019331612,0.0000017362867,3.2865557e-7,0.00020541772,0.9483849,0.00016162814],"about_ca_topic_score_codex":0.021195307,"about_ca_topic_score_gemma":0.00026927935,"teacher_disagreement_score":0.9799898,"about_ca_system_score_codex":0.0014055222,"about_ca_system_score_gemma":0.0066181077,"threshold_uncertainty_score":0.9990134},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2017043167","doi":"10.1007/s11192-009-0069-z","title":"Applying the h-index in exploring bibliometric properties of elite marketing scholars","year":2009,"lang":"en","type":"article","venue":"Scientometrics","topic":"Management and Marketing Education","field":"Business, Management and Accounting","cited_by":36,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Elite; Prestige; Index (typography); Metric (unit); Sociology; Bibliometrics; Library science; Marketing; Social science; Political science; Computer science; Business; Philosophy; Law","score_opus":0.089740442885697,"score_gpt":0.2748107175528747,"score_spread":0.18507027466717768,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2017043167","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9644911,0.0010463112,0.00013664857,0.00053705263,0.0007469446,0.0007487284,2.1126796e-7,0.00007856502,0.0322144],"genre_scores_gemma":[0.99844146,0.00019140731,0.0001460383,0.0005211003,0.00027937113,0.00012443725,0.0000021683652,0.000011816406,0.00028221676],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99783224,0.000037653925,0.00040260333,0.0003033496,0.0010184761,0.0004056692],"domain_scores_gemma":[0.9989232,0.00014573598,0.00029892666,0.00031993046,0.00029740375,0.000014801503],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.009333398,0.00014236358,0.0001654866,0.08039055,0.00021748125,0.00070736546,0.00059127045,0.00003669304,0.00003633398],"category_scores_gemma":[0.0058032214,0.00010600907,0.000062379135,0.24279997,0.000048829228,0.0019953228,0.00020633907,0.00018711983,0.000029478788],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005180074,0.00016839894,0.31747058,0.00033166705,0.0000083913155,0.0000015131877,0.00010828139,0.00019541575,0.0015806842,0.001387331,0.0011261087,0.6775698],"study_design_scores_gemma":[0.00033738677,0.000005506839,0.94795185,0.00023489776,0.000018586681,3.2205605e-7,0.0007611414,0.0042936383,0.00015601012,0.00028640658,0.04574656,0.0002077005],"about_ca_topic_score_codex":0.00011609996,"about_ca_topic_score_gemma":0.00000674531,"teacher_disagreement_score":0.67736214,"about_ca_system_score_codex":0.000059841397,"about_ca_system_score_gemma":0.000021209413,"threshold_uncertainty_score":0.93003243},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2017301591","doi":"10.1007/s11192-014-1274-y","title":"Extending citer analysis to journal impact evaluation","year":2014,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Informetrics; Scientometrics; Impact factor; Ranking (information retrieval); Citation; Measure (data warehouse); Library science; Bibliometrics; Psychology; Information retrieval; Computer science; Sociology; Political science; Data mining","score_opus":0.7601598960034236,"score_gpt":0.7056378674837388,"score_spread":0.054522028519684795,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2017301591","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.767471,0.00037595988,0.2202187,0.0005489162,0.0016291613,0.00028688824,0.000020132225,0.000026743823,0.009422479],"genre_scores_gemma":[0.9943737,0.000032312993,0.0034286114,0.00029727924,0.000285818,0.000008771955,0.0000058856936,0.00001376486,0.0015538674],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9526649,0.00096714275,0.0012019225,0.0011959639,0.04262906,0.0013410222],"domain_scores_gemma":[0.96919125,0.0067912093,0.00063861394,0.0016940971,0.019049495,0.0026353227],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.22129287,0.0002464309,0.00057242514,0.5435105,0.00078729034,0.012955809,0.004299265,0.0001318257,0.004085271],"category_scores_gemma":[0.3109278,0.00015952822,0.00074695185,0.85212326,0.00011818832,0.0013021724,0.0008898209,0.00042614908,0.001649343],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012722144,0.00010420107,0.3140946,8.1168747e-7,0.00012539465,0.0000033650667,0.00017929112,0.0137634305,0.000345959,0.00022168357,0.019117672,0.6520309],"study_design_scores_gemma":[0.0004173702,0.00030360947,0.76743525,0.0000027060487,0.00012616227,0.000014166719,0.00010484914,0.19967523,0.00015971845,0.004045449,0.027452677,0.00026282275],"about_ca_topic_score_codex":0.00004039085,"about_ca_topic_score_gemma":0.000011384209,"teacher_disagreement_score":0.651768,"about_ca_system_score_codex":0.0009269925,"about_ca_system_score_gemma":0.0003741275,"threshold_uncertainty_score":0.999128},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2018907126","doi":"10.1371/journal.pone.0119892","title":"Is Quality and Completeness of Reporting of Systematic Reviews and Meta-Analyses Published in High Impact Radiology Journals Associated with Citation Rates?","year":2015,"lang":"en","type":"article","venue":"PLoS ONE","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":58,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital; University of Ottawa","funders":"University of Ottawa","keywords":"Interquartile range; Citation; Medicine; Checklist; Systematic review; Citation analysis; MEDLINE; Surgery; Psychology; Library science; Biology; Computer science","score_opus":0.9743603511797535,"score_gpt":0.6281372628802134,"score_spread":0.34622308829954007,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2018907126","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9818847,0.015567212,0.00062868925,0.00041630623,0.00000974234,0.0013757382,0.000023796241,0.0000017875635,0.00009201958],"genre_scores_gemma":[0.99599653,0.00009483737,0.00362295,0.00006703596,0.0000057985353,0.000058182366,0.000008212165,0.000006950381,0.0001394767],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9250674,0.035559118,0.034100875,0.00070619537,0.004344512,0.00022187075],"domain_scores_gemma":[0.8790994,0.012936044,0.0988076,0.0024373163,0.0064656152,0.00025397437],"candidate_categories":["metaresearch","metaepi_broad"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.3437126,0.00025938085,0.01724006,0.0006243742,0.000036683203,0.00032111842,0.0004364803,0.000084369196,0.0007024787],"category_scores_gemma":[0.46097627,0.00009998089,0.0008580052,0.0019051975,0.000094913245,0.00044776127,0.0000616313,0.00012449015,0.0000061474652],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000044019445,0.00053032377,0.95105815,0.013704949,0.025362924,0.0000067912924,0.0023832426,0.000055790762,0.0054526622,0.00013088186,0.0012273309,0.00004295733],"study_design_scores_gemma":[0.001406422,0.00050615333,0.91077805,0.005463922,0.061243773,0.00004321594,0.0022342473,0.006804638,0.0013096835,0.0097393105,0.0000041242115,0.00046643923],"about_ca_topic_score_codex":0.0001847883,"about_ca_topic_score_gemma":0.00009969438,"teacher_disagreement_score":0.117263645,"about_ca_system_score_codex":0.000039819937,"about_ca_system_score_gemma":0.00007747138,"threshold_uncertainty_score":0.99735004},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W2023672490","doi":"10.3138/jsp.33.1.11","title":"Are the Best Higher Education Journals Really the Best? A Meta-Analysis of Writing Quality and Readability","year":2001,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"Higher Education Governance and Development","field":"Social Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Readability; Publication; Mandate; Quality (philosophy); Computer science; Meta-analysis; Function (biology); Work (physics); Political science; Advertising; Medicine; Business; Engineering; Biology","score_opus":0.21832808093807748,"score_gpt":0.42565056612358093,"score_spread":0.20732248518550345,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2023672490","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7465953,0.011949666,0.000055623517,0.23392549,0.00047999131,0.00013726343,0.0000113746155,0.000005583357,0.0068396945],"genre_scores_gemma":[0.9916312,0.0009727298,0.0007936958,0.00344801,0.00067019684,0.000007751556,9.5199096e-7,0.0000070458664,0.0024684013],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99503297,0.0013789475,0.0012509628,0.00018026933,0.0019146163,0.00024221097],"domain_scores_gemma":[0.9904449,0.0011012021,0.0035434563,0.00032962408,0.004377541,0.00020330668],"candidate_categories":["scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.02500457,0.00012225287,0.0005850765,0.00028501474,0.0009601726,0.009246406,0.0008954842,0.00012861915,0.0004124764],"category_scores_gemma":[0.007185697,0.00006741017,0.00049512886,0.0017125888,0.00019836532,0.01897829,0.00007429551,0.0010051021,0.0000021078183],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000032019616,0.00041535118,0.94336,0.000014255891,0.013050763,0.000003569555,0.01734961,0.0000459127,0.00006270518,0.009201233,0.0072979107,0.0091666905],"study_design_scores_gemma":[0.000083486084,0.000016764154,0.8466869,0.00003627313,0.010914134,0.000008259962,0.05456581,7.134497e-7,0.000005370087,0.00478661,0.08279184,0.00010381882],"about_ca_topic_score_codex":0.0043344423,"about_ca_topic_score_gemma":0.0012501305,"teacher_disagreement_score":0.2450359,"about_ca_system_score_codex":0.0002392263,"about_ca_system_score_gemma":0.0011564313,"threshold_uncertainty_score":0.99474275},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"evaluation","study_design":"meta_analysis","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"meta_analysis","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2035630854","doi":"10.1353/scp.0.0079","title":"Judging Journal Prices: A Cost Index for Academic Journals","year":2010,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"Publishing and Scholarly Communication","field":"Arts and Humanities","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Index (typography); Economics; Business; Computer science; World Wide Web","score_opus":0.09925764170531497,"score_gpt":0.3165935695281464,"score_spread":0.2173359278228314,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2035630854","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8626355,0.0057054088,0.009075487,0.0760611,0.016177526,0.00062480825,0.00005845643,0.00013202416,0.02952972],"genre_scores_gemma":[0.97926784,0.00032458737,0.0051009716,0.0021824923,0.011949104,0.000011381814,0.000008523763,0.00006875534,0.0010863265],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99525976,0.00026822367,0.0019182609,0.00022384424,0.0016731942,0.0006567387],"domain_scores_gemma":[0.9859827,0.00090717967,0.0033669712,0.00048358028,0.008584995,0.0006745564],"candidate_categories":["metaresearch","sts","scholarly_communication","research_integrity"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.023327963,0.00028891934,0.0005372028,0.0017065853,0.0021319936,0.19869557,0.003855405,0.00047878767,0.00049583404],"category_scores_gemma":[0.022398263,0.00023366735,0.0004766016,0.00028341054,0.00017909716,0.26277012,0.00024102845,0.017240096,0.0000086996915],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006359891,0.0005785178,0.049433548,0.00014291043,0.0009297849,0.00011867738,0.080666795,0.00013312316,0.021263812,0.22367291,0.38286442,0.23955952],"study_design_scores_gemma":[0.0020599782,0.00017597624,0.0051128846,0.0004163967,0.000114316994,0.0014469487,0.006562672,0.00023122833,0.000092197115,0.03150918,0.9519124,0.00036581358],"about_ca_topic_score_codex":0.000041541698,"about_ca_topic_score_gemma":0.00014070333,"teacher_disagreement_score":0.569048,"about_ca_system_score_codex":0.00021292135,"about_ca_system_score_gemma":0.0008359457,"threshold_uncertainty_score":0.9991671},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2043976883","doi":"10.1002/meet.14504701284","title":"Mapping the highly collaborative stem cell research field: Adding last‐author‐based analysis to the author co‐citation analysis family","year":2010,"lang":"en","type":"article","venue":"Proceedings of the American Society for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Scopus; Compromise; Field (mathematics); Citation; Computer science; Operations research; Library science; Data science; Information retrieval; Sociology; Political science; MEDLINE; Mathematics; Social science; Law","score_opus":0.24700646458077738,"score_gpt":0.4998953364659076,"score_spread":0.2528888718851302,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2043976883","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9604208,0.000025573489,0.00689701,0.030409923,0.00009857047,0.00076264987,0.0000590368,0.000029048855,0.0012974148],"genre_scores_gemma":[0.9930104,0.000019573696,0.0057778875,0.00091308757,0.000017758488,0.00012444932,0.0000014841437,0.0000032718096,0.00013210303],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.99214625,0.000033971522,0.0006945371,0.00044400585,0.0060497327,0.0006315123],"domain_scores_gemma":[0.9759975,0.0035643252,0.0010307399,0.00055161613,0.01869819,0.00015762109],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.04533602,0.00012854526,0.0003529325,0.033723235,0.0023637537,0.002246439,0.004290093,0.00010563655,0.0000048924953],"category_scores_gemma":[0.016023926,0.000062433515,0.00030864138,0.5591926,0.0029793058,0.0014435858,0.0008520325,0.00056149706,0.000009730927],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000094230876,0.0001103958,0.19684324,0.00004141822,0.00076263834,7.0306555e-8,0.026460275,0.00064831175,0.13253434,0.034240298,0.09119098,0.5170738],"study_design_scores_gemma":[0.00043179345,0.00066882453,0.13601664,0.000009821808,0.0003014823,0.0000012538468,0.4298768,0.17566682,0.062019404,0.0029233133,0.1917099,0.0003739356],"about_ca_topic_score_codex":0.0000649875,"about_ca_topic_score_gemma":0.000020170366,"teacher_disagreement_score":0.52546936,"about_ca_system_score_codex":0.000105626204,"about_ca_system_score_gemma":0.00048243842,"threshold_uncertainty_score":0.999734},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2044326206","doi":"10.1007/s11192-008-0220-2","title":"A new look at evidence of scholarly citation in citation indexes and from web sources","year":2007,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":85,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Citation; Web of science; Library science; Citation analysis; Webometrics; Citation impact; Bibliometrics; Scholarly communication; Computer science; Sample (material); Impact factor; World Wide Web; Political science; MEDLINE; Publishing","score_opus":0.5726813817650799,"score_gpt":0.5567807477511177,"score_spread":0.015900634013962134,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2044326206","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9728765,0.00912236,0.01575318,0.0004166511,0.00052818324,0.00027749315,0.000022548435,0.000018153753,0.0009849267],"genre_scores_gemma":[0.9885912,0.0006787465,0.009100274,0.000081218095,0.000061686624,0.0000022913212,0.000003677136,0.000009933621,0.001471007],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9797834,0.00024062127,0.0012943669,0.0010849801,0.016842943,0.0007536688],"domain_scores_gemma":[0.97244275,0.020947376,0.00067869056,0.00073543336,0.004372299,0.00082347664],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.061583053,0.0001795445,0.0003808583,0.24832967,0.00020909487,0.0036119826,0.0018837327,0.00020708622,0.0002956545],"category_scores_gemma":[0.19217694,0.00014401932,0.000095451316,0.5391781,0.00028726715,0.0039991075,0.0009399208,0.00033359253,0.00013273461],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005946088,0.00004929365,0.82608587,0.000006706121,0.000004549861,0.000005882124,0.0008868973,0.00003081033,0.012182006,0.0002786964,0.0011107539,0.15929906],"study_design_scores_gemma":[0.0007069465,0.00019801552,0.9824387,0.0000535971,0.0000056810386,0.0000019660008,0.00080319255,0.0018490063,0.005668396,0.0061341,0.0019612918,0.00017911794],"about_ca_topic_score_codex":0.0008616144,"about_ca_topic_score_gemma":0.00068535877,"teacher_disagreement_score":0.2908484,"about_ca_system_score_codex":0.0003777984,"about_ca_system_score_gemma":0.00043073774,"threshold_uncertainty_score":0.99742234},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2049465862","doi":"10.1186/1472-6963-11-268","title":"Paucity of qualitative research in general medical and health services and policy research journals: analysis of publication rates","year":2011,"lang":"en","type":"article","venue":"BMC Health Services Research","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":74,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Cancer Care Ontario; University of Toronto","funders":"","keywords":"Qualitative research; Citation; Nursing research; Health informatics; Health services research; Medicine; Health policy; Health administration; Health care; Bibliometrics; MEDLINE; Public health; Ranking (information retrieval); Qualitative analysis; Library science; Nursing; Social science; Political science; Sociology; Computer science","score_opus":0.9556517327164997,"score_gpt":0.775290194182347,"score_spread":0.1803615385341527,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2049465862","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97050303,0.015705736,0.00022278202,0.010112421,0.00002638004,0.0018063999,0.00007647666,0.0000035011794,0.001543299],"genre_scores_gemma":[0.98961914,0.005526877,0.0036828401,0.00043164424,0.00010444791,0.000116693845,0.000026421187,0.000017262859,0.00047465332],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.6852011,0.25469095,0.017890286,0.0025299513,0.03725348,0.002434235],"domain_scores_gemma":[0.9145306,0.042476393,0.006662241,0.005610083,0.027898194,0.002822514],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.76039135,0.00026516782,0.004306586,0.0154901445,0.00073934195,0.0006502793,0.0033030142,0.00026140708,0.0026745202],"category_scores_gemma":[0.013744738,0.00015640813,0.00041744384,0.034441806,0.0009398926,0.00067056634,0.0013193884,0.00144384,0.00009174114],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014728824,0.0005117475,0.78547764,0.008146445,0.0004968208,0.0000036190022,0.16925377,0.000007783848,0.000038628878,0.013158856,0.00080423086,0.021953152],"study_design_scores_gemma":[0.00037448067,0.000427282,0.87130433,0.0007982214,0.000023044162,0.000004368301,0.07939598,0.02915843,0.000027582775,0.017374083,0.0010007978,0.00011136937],"about_ca_topic_score_codex":0.09955664,"about_ca_topic_score_gemma":0.08406115,"teacher_disagreement_score":0.74664664,"about_ca_system_score_codex":0.00027584413,"about_ca_system_score_gemma":0.0036033106,"threshold_uncertainty_score":0.9982372},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W2050885411","doi":"10.4028/www.scientific.net/amm.163.79","title":"Research Trends of Mechanism and Machine Theory Using Bibliometrics","year":2012,"lang":"en","type":"article","venue":"Applied Mechanics and Materials","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"China; Bibliometrics; Mechanism (biology); Field (mathematics); Perspective (graphical); Social science; Political science; Regional science; History; Library science; Data science; Geography; Sociology; Computer science; Artificial intelligence; Law; Mathematics; Epistemology; Philosophy","score_opus":0.6368217407589188,"score_gpt":0.5625886931724329,"score_spread":0.07423304758648597,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2050885411","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9016365,0.0025864118,0.091971576,0.00007460565,0.00074533833,0.00031122836,0.00017397595,0.000020342668,0.0024800282],"genre_scores_gemma":[0.99034727,0.001214543,0.008021094,0.000040896568,0.00010236446,0.000012966815,0.0000047386297,0.000022232904,0.00023389765],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9927632,0.00042862305,0.0007485805,0.00054706674,0.004655252,0.0008572601],"domain_scores_gemma":[0.9947096,0.0029718091,0.0002799116,0.00057790574,0.0009538063,0.0005069567],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.07586362,0.00018200788,0.00052191823,0.14396073,0.00031191536,0.0010311251,0.00078499713,0.0002036431,0.00067022484],"category_scores_gemma":[0.0039368966,0.00012825837,0.0000436738,0.19417116,0.000107717446,0.00036081806,0.0015497745,0.00018621137,0.000022756147],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004231165,0.000068871705,0.00005798357,0.000015318044,0.000012880072,8.1463685e-7,0.00009766526,1.9419598e-7,0.19349624,0.7587561,0.00014919841,0.04730242],"study_design_scores_gemma":[0.0005300266,0.00014507391,0.001028914,0.000008074666,0.000018960405,0.00001816199,0.00078421173,0.00034379985,0.21716975,0.77760345,0.0021319902,0.00021759982],"about_ca_topic_score_codex":0.00007480814,"about_ca_topic_score_gemma":0.0000012182074,"teacher_disagreement_score":0.08871077,"about_ca_system_score_codex":0.00003580588,"about_ca_system_score_gemma":0.000050810737,"threshold_uncertainty_score":0.9943163},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2051771537","doi":"10.1002/bult.2013.1720390405","title":"The power of altmetrics on a CV","year":2013,"lang":"en","type":"article","venue":"Bulletin of the American Society for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":86,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Impact; OpenAlex","funders":"","keywords":"Altmetrics; Citation; Social media; Computer science; Data science; World Wide Web","score_opus":0.1260122823245402,"score_gpt":0.4456347829134141,"score_spread":0.31962250058887387,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2051771537","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9379225,0.000120460696,0.0010406155,0.054844752,0.00021382906,0.00073501427,0.000024545105,0.000020842186,0.0050774403],"genre_scores_gemma":[0.9962651,0.0001995271,0.0021638367,0.0010406371,0.0000044917983,0.00003525451,1.7580545e-7,0.0000021842036,0.00028876594],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99474645,0.000022736873,0.00057954766,0.00019086586,0.0041037216,0.00035666252],"domain_scores_gemma":[0.98836017,0.0029462965,0.00099154,0.00076599076,0.0068588313,0.00007717804],"candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.012656623,0.00007769094,0.00020161895,0.0071620354,0.0006609843,0.00047698166,0.0030485592,0.000051269333,0.000028219029],"category_scores_gemma":[0.042929877,0.00003601714,0.0001650493,0.11851637,0.0060110386,0.00031097565,0.00078053435,0.00014545002,0.000051267372],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016305103,0.000046577643,0.006585426,0.0000064647134,0.000017455754,1.2176167e-8,0.00036596708,0.000011439217,0.0014495503,0.037731472,0.19965075,0.75411856],"study_design_scores_gemma":[0.00040729568,0.0008349729,0.06514032,0.000008471907,0.0000053274202,0.000004289449,0.015883628,0.002124711,0.008306456,0.013301253,0.89384586,0.00013742212],"about_ca_topic_score_codex":0.000074338925,"about_ca_topic_score_gemma":3.235553e-7,"teacher_disagreement_score":0.7539812,"about_ca_system_score_codex":0.000048968348,"about_ca_system_score_gemma":0.00020770433,"threshold_uncertainty_score":0.996694},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2057264385","doi":"10.1007/s11192-013-1171-9","title":"An overview of forestry journals in the period 2006–2010 as basis for ascertaining research trends","year":2013,"lang":"en","type":"article","venue":"Scientometrics","topic":"Forest ecology and management","field":"Environmental Science","cited_by":41,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institute for Biological Sciences","funders":"","keywords":"Ordination; Ecophysiology; Forest management; Ecology; Vegetation (pathology); Plant ecology; Plant science; Forest ecology; Biology; Forestry; Geography; Botany; Ecosystem","score_opus":0.1892962841462507,"score_gpt":0.43934751497824204,"score_spread":0.2500512308319913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2057264385","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98084027,0.00037947696,0.00013038839,0.0022480655,0.00022353571,0.0004699254,0.000007304086,0.000008467544,0.015692558],"genre_scores_gemma":[0.99607277,0.00010560306,0.0015807314,0.00029182082,0.000024115017,0.00011346375,0.000002603821,0.00000699401,0.0018018959],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9977782,0.00018124041,0.0002562912,0.00027359737,0.000992652,0.00051804073],"domain_scores_gemma":[0.9988206,0.0004116063,0.000090868736,0.0005119308,0.000057274283,0.00010771343],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0071288855,0.00008328147,0.00013267278,0.0018761532,0.0002667429,0.00013145003,0.0011978161,0.00007546146,0.0048529864],"category_scores_gemma":[0.0018740577,0.000057613957,0.00006265669,0.011256471,0.00034206692,0.00041282346,0.00028090712,0.00021826703,0.00020815684],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000033035987,0.00094228366,0.7695544,0.00006412301,0.000022975444,0.000012254167,0.004849581,0.003571514,0.00064368447,0.009671887,0.099251196,0.11138309],"study_design_scores_gemma":[0.00026083284,0.00039354578,0.97617733,0.000012770969,0.0000064625,0.0000036238423,0.0012509822,0.00192033,0.00007296771,0.011268121,0.008549364,0.0000836925],"about_ca_topic_score_codex":0.00088221167,"about_ca_topic_score_gemma":0.00040920175,"teacher_disagreement_score":0.20662293,"about_ca_system_score_codex":0.00013471834,"about_ca_system_score_gemma":0.00002148494,"threshold_uncertainty_score":0.99605674},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2058472726","doi":"10.1007/s11192-007-1707-y","title":"Why are Websites co-linked? The case of Canadian universities","year":2007,"lang":"en","type":"article","venue":"Scientometrics","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":35,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Western University","funders":"","keywords":"Similarity (geometry); Context (archaeology); World Wide Web; Sample (material); Set (abstract data type); Co-occurrence; Web page; Computer science; Information retrieval; Geography; Chemistry","score_opus":0.04234576219966546,"score_gpt":0.288850666366687,"score_spread":0.24650490416702153,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2058472726","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7869862,0.00068021246,0.10875255,0.0023023593,0.001424595,0.00033730906,0.000087303415,0.000117494434,0.09931198],"genre_scores_gemma":[0.9930051,0.000027814118,0.004704607,0.0015312947,0.00003544358,3.7379195e-7,0.0000020205684,0.000003946421,0.000689387],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984185,0.00003175212,0.0002517288,0.00024819595,0.0005538357,0.0004959546],"domain_scores_gemma":[0.99766546,0.0006517378,0.00015543758,0.00069879944,0.00049686205,0.00033171108],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0033388846,0.000102859995,0.00013101706,0.013330006,0.00042087075,0.00022139038,0.0013958534,0.0000794125,0.00004305231],"category_scores_gemma":[0.0006811267,0.000078637895,0.00007690924,0.05604397,0.00026016575,0.0007331033,0.0001917669,0.00014553229,0.000041662275],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021071968,0.00034142198,0.10320693,0.0002015042,0.00010266006,0.0038050143,0.015541338,0.00071445684,0.00018498156,0.5209118,0.25738752,0.09758128],"study_design_scores_gemma":[0.00065411086,0.00030942142,0.09771025,0.000029289371,0.00002407533,0.0006521512,0.015409845,0.014519621,0.0045740693,0.0012904075,0.86421937,0.0006074183],"about_ca_topic_score_codex":0.050752215,"about_ca_topic_score_gemma":0.08462959,"teacher_disagreement_score":0.6068318,"about_ca_system_score_codex":0.00030049478,"about_ca_system_score_gemma":0.00043640437,"threshold_uncertainty_score":0.99785304},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W2058498906","doi":"10.1007/s11192-011-0369-y","title":"Sex differences in research funding, productivity and impact: an analysis of Québec university professors","year":2011,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":204,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université du Québec; Université du Québec à Montréal","funders":"McGill University","keywords":"Productivity; Scientometrics; Political science; Regional science; Sociology; Economics; Economic growth; Social science","score_opus":0.872966831153213,"score_gpt":0.6254438142378042,"score_spread":0.2475230169154089,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2058498906","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9955975,0.0003410447,0.00036256813,0.000091500915,0.00015208228,0.00030880512,0.00005928618,0.000015664875,0.0030715687],"genre_scores_gemma":[0.9969999,0.00009431289,0.0003836482,0.0000030843282,0.000013581109,0.0000019112956,0.0000023901964,0.000006587633,0.0024945633],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98272353,0.0013595909,0.0006469616,0.0013847505,0.012784578,0.0011005736],"domain_scores_gemma":[0.9875354,0.00497048,0.00034513656,0.0012781391,0.0049561406,0.00091468595],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.09664137,0.00017537712,0.00069418584,0.4405308,0.0003897616,0.0007019419,0.0031819332,0.0001542178,0.00048667061],"category_scores_gemma":[0.059531037,0.00012562706,0.00017343384,0.766765,0.0012252252,0.0014253677,0.0014101858,0.0005116217,0.00001835959],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000039398612,0.0005245767,0.96795774,0.000005277978,0.00005193958,0.000009488515,0.003533811,0.000011560521,0.0001515115,0.0005357205,0.00010656422,0.027072396],"study_design_scores_gemma":[0.0002270595,0.0003935306,0.9861859,0.0000036472331,0.000029198762,8.675518e-7,0.004411916,0.0073114973,0.00036163637,0.00073248474,0.0001932674,0.00014899974],"about_ca_topic_score_codex":0.046965227,"about_ca_topic_score_gemma":0.018731663,"teacher_disagreement_score":0.32623416,"about_ca_system_score_codex":0.0005574236,"about_ca_system_score_gemma":0.0008836918,"threshold_uncertainty_score":0.99917394},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W2059600429","doi":"10.1177/016224390202700101","title":"Outline for a History of Science Measurement","year":2002,"lang":"en","type":"article","venue":"Science Technology & Human Values","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institut National de la Recherche Scientifique","funders":"","keywords":"Centrality; Work (physics); Foundation (evidence); Control (management); Sociology; Political science; Social science; Statistics; Law; Economics; Mathematics; Management; Engineering","score_opus":0.7704066506741,"score_gpt":0.5689810652915762,"score_spread":0.20142558538252375,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2059600429","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9506014,0.004558901,0.004175382,0.003037751,0.0021714377,0.0009404613,0.000014265508,0.00024630985,0.03425407],"genre_scores_gemma":[0.9909755,0.000033149576,0.004385201,0.00006266729,0.000033852517,0.00003607131,1.3210297e-7,0.000008418328,0.004464999],"study_design_codex":"bench_or_experimental","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9806763,0.000045227876,0.0008874415,0.0014198794,0.015809821,0.0011613184],"domain_scores_gemma":[0.98456734,0.00044943095,0.00046860747,0.001725967,0.012467568,0.00032109802],"candidate_categories":["metaresearch","bibliometrics","sts","open_science"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.0610751,0.00017592325,0.00040501947,0.14704888,0.0014662212,0.00044510316,0.0097591765,0.00012647068,0.00046291234],"category_scores_gemma":[0.07029786,0.00012974364,0.00013678487,0.2222201,0.0350021,0.0011063734,0.0013369556,0.0002460835,0.00011984939],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000008838147,0.0005923008,0.016159885,0.000013224176,0.0000100608595,0.0000039543975,0.0015140683,0.000022615544,0.6945354,0.089829706,0.033707503,0.16360244],"study_design_scores_gemma":[0.002150512,0.0030705326,0.026465517,0.000073112074,0.00003435514,0.000029283534,0.0049992097,0.039690964,0.25951755,0.5087103,0.15420005,0.0010586076],"about_ca_topic_score_codex":0.000018449966,"about_ca_topic_score_gemma":0.000008621544,"teacher_disagreement_score":0.43501785,"about_ca_system_score_codex":0.001600116,"about_ca_system_score_gemma":0.0010711302,"threshold_uncertainty_score":0.99983376},"labels":[{"model":"gemma","categories":["bibliometrics","sts"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2060451679","doi":"10.4330/wjc.v3.i12.383","title":"Cardiovascular disease research in Latin America: A comparative bibliometric analysis","year":2011,"lang":"en","type":"article","venue":"World Journal of Cardiology","topic":"Scientific Research and Technology","field":"Computer Science","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Institute of Allergy and Infectious Diseases; National Institute of Mental Health; Fogarty International Center; National Heart, Lung, and Blood Institute; Office of Women's Health; National Institutes of Health; Vanderbilt University","keywords":"Latin Americans; China; Medicine; Bibliometrics; Library science; Political science","score_opus":0.2099172115518915,"score_gpt":0.37803018980875447,"score_spread":0.16811297825686297,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2060451679","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.43813765,0.020081185,0.5110974,0.0010572942,0.0009770833,0.00039734357,0.000006803986,0.00006453967,0.028180707],"genre_scores_gemma":[0.9925809,0.00034441822,0.0068504056,0.00001977126,0.00009231479,0.000010831673,7.363634e-7,0.0000040335713,0.00009661134],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.995873,0.0015023464,0.0005317468,0.00041030202,0.0010477097,0.00063493586],"domain_scores_gemma":[0.99729306,0.00048128483,0.00016485118,0.0009939153,0.0007521982,0.00031472274],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0053963033,0.00011811419,0.00088534044,0.16626719,0.00009246841,0.00006748341,0.0019036303,0.000055946766,0.000028501952],"category_scores_gemma":[0.0003646605,0.000093892675,0.0007143238,0.27804822,0.000516493,0.0003217144,0.00045105707,0.00077031215,0.00004366767],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00051592075,0.00037171776,0.591881,0.000033370343,0.014318082,0.011885033,0.0020465702,0.0153530985,0.00035546737,0.032718994,0.015302018,0.31521875],"study_design_scores_gemma":[0.0008345859,0.00035830148,0.9319256,0.000019228646,0.00018161786,0.00006994252,0.00021455095,0.004155271,0.0002779621,0.016200963,0.045540858,0.000221107],"about_ca_topic_score_codex":0.000086140746,"about_ca_topic_score_gemma":0.00002572201,"teacher_disagreement_score":0.55444324,"about_ca_system_score_codex":0.00010320863,"about_ca_system_score_gemma":0.00023095467,"threshold_uncertainty_score":0.84318244},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2065923934","doi":"10.1007/s11192-014-1489-y","title":"Prospective evaluation of the accessibility of Internet references in leading general medical journals","year":2014,"lang":"en","type":"article","venue":"Scientometrics","topic":"Health Literacy and Information Accessibility","field":"Health Professions","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University; Institute for Clinical Evaluative Sciences; St Mary's Hospital Centre; Women's College Hospital; University of Toronto","funders":"","keywords":"The Internet; Sample (material); Library science; Medicine; Computer science; World Wide Web","score_opus":0.2332333098912759,"score_gpt":0.5662370323276832,"score_spread":0.3330037224364073,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2065923934","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9793142,0.00009300073,0.0011215981,0.00023949832,0.00095785764,0.0006567484,0.000007857774,0.000007770272,0.017601486],"genre_scores_gemma":[0.9990654,0.000011678083,0.0003058064,0.00035207404,0.0000633989,0.000029039054,0.000002351836,0.0000027284505,0.0001675102],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99298316,0.0018457604,0.001953492,0.00021028248,0.0027089724,0.0002983458],"domain_scores_gemma":[0.9951205,0.0009210624,0.0013293562,0.00038935276,0.0021017145,0.00013796774],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.061824113,0.000074473835,0.00027336736,0.0011199183,0.00016233721,0.000016785361,0.0006420128,0.00015925328,0.0012068248],"category_scores_gemma":[0.047559045,0.000044021795,0.00005150836,0.007250172,0.0001677337,0.0006650752,0.00023390519,0.00059266575,0.000013564804],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012804698,0.00005728179,0.96787196,0.00018705014,0.000002079021,8.607219e-9,0.0036269904,0.000019922714,0.000019656625,0.0010773606,0.00020488976,0.026919967],"study_design_scores_gemma":[0.00063758594,0.00002812608,0.9483506,0.00031832917,0.0000049997525,1.478333e-7,0.0003424511,0.046791274,0.00040206866,0.0027068683,0.00037624416,0.000041316936],"about_ca_topic_score_codex":0.0002396154,"about_ca_topic_score_gemma":0.00022714637,"teacher_disagreement_score":0.04677135,"about_ca_system_score_codex":0.0003812645,"about_ca_system_score_gemma":0.0012156672,"threshold_uncertainty_score":0.9997062},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2066590496","doi":"10.1111/hir.12097","title":"Evaluating the effectiveness of knowledge brokering in health research: a systematised review with some bibliometric information","year":2015,"lang":"en","type":"review","venue":"Health Information & Libraries Journal","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Knowledge translation; Dialog box; Scopus; Knowledge management; MEDLINE; Empirical research; Computer science; Medicine; Medical education; World Wide Web; Political science","score_opus":0.8867331037637374,"score_gpt":0.7375122507614661,"score_spread":0.14922085300227128,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2066590496","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000013348568,0.97810256,0.0008846534,0.003718953,0.00087284244,0.014963255,0.00019328241,0.00006637787,0.0011847084],"genre_scores_gemma":[0.000023957731,0.98991436,0.0015036264,0.0061201183,0.00026271716,0.0019250781,0.0001826992,0.000039141036,0.000028302693],"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","domain_scores_codex":[0.9187445,0.06074525,0.013793157,0.00029437727,0.0041014007,0.0023212936],"domain_scores_gemma":[0.95969874,0.020396883,0.014330959,0.0008982031,0.0035852983,0.0010898999],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","research_integrity"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.21031672,0.0005212372,0.0034810444,0.02905756,0.0027244412,0.00032970746,0.0012682688,0.00031323536,0.00009689052],"category_scores_gemma":[0.021312233,0.00031651693,0.00019277984,0.05919939,0.00026674537,0.012918704,0.00044304295,0.0043551945,0.00049162266],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000050778108,0.000011499576,0.000021132866,0.54264086,0.000022275715,2.1361328e-7,0.008653757,0.0000130216395,2.9168934e-9,0.003322083,0.016739523,0.42852485],"study_design_scores_gemma":[0.0008752336,0.0007377594,0.00022419059,0.2981394,0.00002319229,0.00007757315,0.0020842056,0.000072783565,2.7024162e-8,0.00014998276,0.69744533,0.0001703302],"about_ca_topic_score_codex":0.00038265245,"about_ca_topic_score_gemma":0.00002539027,"teacher_disagreement_score":0.68070585,"about_ca_system_score_codex":0.0039964593,"about_ca_system_score_gemma":0.072450325,"threshold_uncertainty_score":0.9999287},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2069122309","doi":"10.1016/j.respol.2009.11.004","title":"Tracing the links between science and technology: An exploratory analysis of scientists’ and inventors’ networks","year":2009,"lang":"en","type":"article","venue":"Research Policy","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":229,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Università Bocconi; European Commission","keywords":"Position (finance); Social connectedness; Exploratory analysis; Process (computing); Network science; Order (exchange); Data science; Sociology; Political science; Knowledge management; Business; Computer science; Complex network; Psychology; World Wide Web","score_opus":0.6759460167148517,"score_gpt":0.6428787603658095,"score_spread":0.03306725634904217,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2069122309","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9895003,0.0011457456,0.00057562924,0.007412505,0.000040701823,0.00026017925,0.000015157514,0.000019400692,0.0010303812],"genre_scores_gemma":[0.9991848,0.00026471858,0.00015561287,0.000111995105,0.0001203875,0.0000065436825,0.0000010336037,0.000005074159,0.00014981782],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.98425,0.000626305,0.00058483944,0.0009559014,0.012384639,0.0011983004],"domain_scores_gemma":[0.98560613,0.003685956,0.00018015651,0.0013527418,0.008299977,0.0008750276],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.1105867,0.0001252791,0.00039753414,0.19610167,0.001480091,0.0030226018,0.003578217,0.00021077016,0.00001788467],"category_scores_gemma":[0.07944057,0.000075935495,0.000061259205,0.6007964,0.0055342847,0.001059885,0.0014833657,0.000989258,0.0000062164513],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009533296,0.00010330945,0.29888457,0.000004157993,0.000057530127,0.0000062261975,0.0011272082,0.00009716838,0.0031465974,0.017023856,0.0012595592,0.6782803],"study_design_scores_gemma":[0.00030779748,0.0006630869,0.88529927,0.000017460401,0.000038655642,0.0000042264624,0.002682596,0.06876666,0.0022698238,0.036641497,0.0031048635,0.000204091],"about_ca_topic_score_codex":0.00021495993,"about_ca_topic_score_gemma":0.000060737813,"teacher_disagreement_score":0.6780762,"about_ca_system_score_codex":0.00019082753,"about_ca_system_score_gemma":0.0012130687,"threshold_uncertainty_score":0.9998199},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2070232274","doi":"10.14740/jocmr2115w","title":"Clinical Research Quo Vadis? Trends in Reporting of Clinical Trials and Observational Study Designs Over Two Decades","year":2015,"lang":"en","type":"article","venue":"Journal of Clinical Medicine Research","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Observational study; Medicine; Randomized controlled trial; Clinical study design; Research design; Medical literature; Clinical trial; Evidence-based medicine; Family medicine; Alternative medicine; Surgery; Internal medicine; Pathology; Statistics","score_opus":0.9984930451457487,"score_gpt":0.882353450358009,"score_spread":0.11613959478773961,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2070232274","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":"methods","domain_gemma":"methods","model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":"methods","prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9814129,0.0022086836,0.0005280216,0.010043642,0.0013158858,0.0010319103,0.0000029852863,0.0000017884311,0.0034541942],"genre_scores_gemma":[0.9881671,0.0005674246,0.0034996828,0.00015658136,0.0046322998,0.000016087828,0.0000017924682,0.000021804357,0.0029372326],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.084261335,0.58498734,0.27305886,0.0025399681,0.053848345,0.0013041522],"domain_scores_gemma":[0.04421355,0.81085575,0.099227905,0.005246213,0.036941245,0.0035153371],"candidate_categories":["metaresearch","metaepi_broad","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.98297626,0.00033129327,0.019874178,0.0036714245,0.0001638706,0.00031924975,0.0031441848,0.0004851049,0.0034037814],"category_scores_gemma":[0.97718084,0.00014486746,0.0035225928,0.006741063,0.0020737508,0.0005007812,0.0007333483,0.005859734,0.00009629512],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00077437906,0.0011970511,0.7351842,0.000020202318,0.00039481762,0.00019827813,0.0004157564,0.000010548127,0.0000056606314,0.0002859241,0.106729954,0.15478326],"study_design_scores_gemma":[0.007989861,0.006519939,0.94187474,0.0004861193,0.00025887496,0.00004929565,0.0065058502,0.0023879556,0.0000016489893,0.015580728,0.018220482,0.00012453034],"about_ca_topic_score_codex":0.00016951858,"about_ca_topic_score_gemma":0.00023562026,"teacher_disagreement_score":0.22586842,"about_ca_system_score_codex":0.0000737726,"about_ca_system_score_gemma":0.0025838683,"threshold_uncertainty_score":0.9975072},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2072176804","doi":"10.1136/bmj.d1028","title":"Research output on primary care in Australia, Canada, Germany, the Netherlands, the United Kingdom, and the United States: bibliometric analysis","year":2011,"lang":"en","type":"article","venue":"BMJ","topic":"Primary Care and Health Outcomes","field":"Health Professions","cited_by":137,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"NIHR School for Primary Care Research; National Institute for Health and Care Research","keywords":"Citation; Bibliometrics; Primary care; Publishing; Library science; Family medicine; Kingdom; Medicine; MEDLINE; Sample (material); Political science; Law","score_opus":0.26812950252117274,"score_gpt":0.4824654141450026,"score_spread":0.21433591162382987,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2072176804","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94764555,0.00063050224,0.000006711599,0.0381513,0.0003066118,0.0020594895,0.000100407895,0.000032254473,0.011067152],"genre_scores_gemma":[0.95054483,0.0006256992,0.000017799117,0.043081366,0.00022407476,0.00035572035,0.00028426404,0.000028671391,0.004837554],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99240875,0.004689184,0.000735611,0.00031288288,0.0009176559,0.0009359236],"domain_scores_gemma":[0.989983,0.008052072,0.00025438835,0.0010149623,0.00053038634,0.00016514315],"candidate_categories":["bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0070101703,0.00020434298,0.00043085188,0.021269547,0.0015173786,0.0000369934,0.00078433484,0.00018212425,0.00025432865],"category_scores_gemma":[0.0005057551,0.00009055939,0.00008530278,0.09628533,0.0003224619,0.00005803564,0.00032740433,0.0019543068,0.00004150143],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009850351,0.000045815294,0.4336362,0.000330573,0.00037680994,0.00005350217,0.030410107,0.00013734057,4.4046618e-7,0.00215015,0.53025687,0.0016171564],"study_design_scores_gemma":[0.0011100812,0.000059541948,0.7430903,0.00004721414,0.00010373155,9.1004034e-7,0.009044307,0.0001278886,7.820084e-7,0.0002117196,0.24609587,0.00010766055],"about_ca_topic_score_codex":0.95611006,"about_ca_topic_score_gemma":0.8448243,"teacher_disagreement_score":0.3094541,"about_ca_system_score_codex":0.0008887538,"about_ca_system_score_gemma":0.0021166485,"threshold_uncertainty_score":0.9997825},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2074237094","doi":"10.1007/s11069-014-1260-y","title":"An overview of climate change vulnerability: a bibliometric analysis based on Web of Science database","year":2014,"lang":"en","type":"article","venue":"Natural Hazards","topic":"Climate change impacts on agriculture","field":"Agricultural and Biological Sciences","cited_by":302,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Vulnerability (computing); Climate change; Environmental resource management; Vulnerability assessment; Bibliometrics; Natural hazard; Science Citation Index; Library science; Political science; Geography; Environmental science; Computer science; Citation; Ecology; Meteorology","score_opus":0.08917746577153815,"score_gpt":0.35233080931713184,"score_spread":0.2631533435455937,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2074237094","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.996933,0.0006825763,8.825533e-7,0.0005332876,0.00016056685,0.00025796678,0.00090324815,0.000057368317,0.00047110752],"genre_scores_gemma":[0.99836755,0.0005307064,0.00015191353,0.00048680266,0.00014941355,0.000012873193,0.0002950955,0.0000014195234,0.0000042528345],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99727166,0.00019537738,0.00038764143,0.00054167904,0.0011149811,0.00048866874],"domain_scores_gemma":[0.99820435,0.0003849267,0.0003256952,0.0003150523,0.00052885554,0.00024111953],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0019415176,0.00022910137,0.00053022295,0.005435857,0.0001811995,0.00006778599,0.00075948087,0.000112836504,0.00040433693],"category_scores_gemma":[0.00067675265,0.00007949929,0.00025406643,0.14085247,0.00035576176,0.00052011834,0.00012903578,0.00021419165,0.000007233739],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000120921475,0.0006414018,0.06354304,0.00013813128,0.0000358012,0.0000017577107,0.00006194856,0.000034213386,0.55098885,0.0004205109,0.00009326562,0.38392013],"study_design_scores_gemma":[0.00023234238,0.0007888868,0.9583662,0.00009354153,0.00014301224,0.0000011507135,0.00007121229,0.02743992,0.012000821,0.000021616872,0.00060016615,0.00024113253],"about_ca_topic_score_codex":0.00034756676,"about_ca_topic_score_gemma":0.0007659862,"teacher_disagreement_score":0.89482313,"about_ca_system_score_codex":0.00006309939,"about_ca_system_score_gemma":0.000023490205,"threshold_uncertainty_score":0.8774091},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2074772630","doi":"10.3152/030234211x12924093660192","title":"Scientific publications and patenting by companies: a study of the whole population of Canadian firms over 25 years","year":2011,"lang":"en","type":"article","venue":"Science and Public Policy","topic":"Innovation Policy and R&D","field":"Economics, Econometrics and Finance","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Publication; Publishing; Frontier; Scientific publishing; Scientific literature; Population; Business; Economics; Political science; Sociology; Advertising; Law","score_opus":0.08880438945652705,"score_gpt":0.25503797739600176,"score_spread":0.1662335879394747,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2074772630","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9903121,0.000043658503,0.0000021526835,0.0011106385,0.00009182834,0.00013648356,0.000118918666,0.0000043840782,0.008179862],"genre_scores_gemma":[0.99937457,0.000004117892,0.000019605022,0.00017570691,0.000013872278,0.0000048420693,0.000004166786,0.0000029324162,0.00040016067],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9992289,0.000008642059,0.00028375542,0.00019343077,0.000072790775,0.0002124967],"domain_scores_gemma":[0.99931747,0.000007930771,0.00022944971,0.00023294678,0.0001271218,0.00008507663],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00092789694,0.000046512127,0.00010090482,0.001483712,0.00040066917,0.00015267395,0.00026018976,0.00002572446,0.00002698121],"category_scores_gemma":[0.00034465815,0.00004093504,0.000012873717,0.0038741566,0.0004882661,0.0005705755,0.00007544568,0.000046553247,0.0000028992627],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[3.0755842e-7,0.00005498092,0.59474796,0.0000044116514,0.000004017534,1.2530941e-8,0.004243974,2.8335612e-7,0.00006947777,0.39952874,0.0008333492,0.0005124936],"study_design_scores_gemma":[0.00012991446,0.000019125682,0.9821371,0.000002785199,0.0000010339772,5.0538205e-7,0.0005351447,0.00019834441,0.00001885326,0.006890907,0.010016467,0.000049820927],"about_ca_topic_score_codex":0.36056975,"about_ca_topic_score_gemma":0.021398818,"teacher_disagreement_score":0.39263785,"about_ca_system_score_codex":0.00003865266,"about_ca_system_score_gemma":0.00020429655,"threshold_uncertainty_score":0.9964581},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W2075464897","doi":"10.1007/s11192-012-0859-6","title":"Characterizing a scientific elite (B): publication and citation patterns of the most highly cited scientists in environmental science and ecology","year":2012,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":62,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Citation; Publishing; Publication; Elite; Productivity; Citation analysis; Library science; Ecology; Political science; Sociology; Biology; Computer science; Law","score_opus":0.26447491485703406,"score_gpt":0.4612345109392629,"score_spread":0.19675959608222882,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2075464897","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9956776,0.0006994798,0.00013463551,0.0007187071,0.0016700912,0.00045266986,0.00010116226,0.000010109165,0.0005355787],"genre_scores_gemma":[0.99883133,0.00013830276,0.00034268596,0.00016298507,0.00003054688,0.000013478271,0.000008483442,0.000008122991,0.0004640881],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98372483,0.00026619187,0.0008648708,0.0010733446,0.012976815,0.0010939189],"domain_scores_gemma":[0.9935562,0.0018035665,0.00059426547,0.0009216093,0.0024443201,0.0006800312],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.068968765,0.00016955637,0.0002728926,0.14825515,0.0009451085,0.0040509864,0.0022124788,0.000104637824,0.000064110296],"category_scores_gemma":[0.06760544,0.00011683832,0.000048373186,0.45360926,0.0026510481,0.0032238746,0.0021370817,0.0002516458,0.000028697012],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000458263,0.00014713492,0.8878684,0.000007167814,0.00000163685,3.3263163e-7,0.00082012836,0.0000011938992,0.041489635,0.0006697598,0.000107689186,0.06888234],"study_design_scores_gemma":[0.0003440464,0.000037609912,0.9871478,0.000010713667,0.000003806649,0.000010180237,0.00036640657,0.0025412806,0.004473042,0.00019655743,0.004734856,0.00013367478],"about_ca_topic_score_codex":0.000055574124,"about_ca_topic_score_gemma":0.000022160319,"teacher_disagreement_score":0.3053541,"about_ca_system_score_codex":0.000433768,"about_ca_system_score_gemma":0.0003471444,"threshold_uncertainty_score":0.99698293},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2079245179","doi":"10.1002/asi.22737","title":"Modeling the relationship between an emerging infectious disease epidemic and the body of scientific literature associated with it: The case of <scp>HIV</scp>/<scp>AIDS</scp> in the <scp>U</scp>nited <scp>S</scp>tates","year":2012,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"Data-Driven Disease Surveillance","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"U.S. National Library of Medicine; Centers for Disease Control and Prevention; Public Health Agency of Canada","keywords":"Scientific literature; Human immunodeficiency virus (HIV); Public health; Disease; Medicine; Demography; Gerontology; Family medicine; Internal medicine; Sociology; Pathology; Biology","score_opus":0.022365670023787018,"score_gpt":0.30461583486005356,"score_spread":0.2822501648362665,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2079245179","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99479353,0.0008589243,0.0009863918,0.0019894259,0.00010648779,0.00088234514,0.00023034887,0.00004087258,0.000111659414],"genre_scores_gemma":[0.998305,0.00017939409,0.000313181,0.0009945543,0.00007110569,0.000041050116,0.00004112598,0.000015484657,0.000039083494],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99679226,0.00035067878,0.001065162,0.00023150326,0.0009406587,0.0006197206],"domain_scores_gemma":[0.98680246,0.008120449,0.0025041644,0.00085338217,0.0015168631,0.00020266583],"candidate_categories":["metaresearch","sts"],"consensus_categories":["sts"],"category_scores_codex":[0.010421384,0.0002707424,0.0006025591,0.0005716925,0.001414741,0.00026519407,0.0010453179,0.00013048289,2.1468172e-7],"category_scores_gemma":[0.034844838,0.00012054916,0.000282938,0.006164107,0.005499744,0.0021336316,0.00026632182,0.0010291265,9.534772e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024023237,0.0002002189,0.9253983,0.00017252924,0.00039124987,0.000006485267,0.059453044,0.0017228699,0.00053736795,0.0033409658,0.0063120625,0.0024408952],"study_design_scores_gemma":[0.0053745066,0.0008349079,0.526879,0.00085809594,0.0014863623,0.0011897332,0.3563793,0.092460014,0.0005130995,0.0066646785,0.0072485446,0.00011175876],"about_ca_topic_score_codex":0.000042725664,"about_ca_topic_score_gemma":0.000018880633,"teacher_disagreement_score":0.39851928,"about_ca_system_score_codex":0.00011867682,"about_ca_system_score_gemma":0.0004674469,"threshold_uncertainty_score":0.99988526},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2081441000","doi":"10.1371/journal.pone.0057176","title":"Sino-Canadian Collaborations in Stem Cell Research: A Scientometric Analysis","year":2013,"lang":"en","type":"article","venue":"PLoS ONE","topic":"Biomedical Ethics and Regulation","field":"Medicine","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Global Affairs Canada; University of Toronto; Public Health Ontario","funders":"Canadian Institutes of Health Research; University of Toronto; Chinese Academy of Sciences; University Health Network","keywords":"China; Stem cell; Context (archaeology); Scopus; Proxy (statistics); Political science; Regional science; Biology; Geography; MEDLINE; Computer science; Genetics","score_opus":0.31855502969712385,"score_gpt":0.34892403691759016,"score_spread":0.030369007220466315,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2081441000","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96620953,0.00027712283,0.000018710056,0.024288498,0.000015250186,0.00049681694,0.000009466351,0.000013835112,0.008670758],"genre_scores_gemma":[0.99430496,0.000054144853,0.00073081546,0.00023468747,0.00004414033,0.000038187012,0.000038263795,0.0000064059564,0.004548422],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99836653,0.00008119523,0.00018168298,0.00018412908,0.00089267775,0.00029378486],"domain_scores_gemma":[0.99846756,0.00010807232,0.00002520445,0.00022531532,0.00072769745,0.00044617456],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00085909996,0.00004976329,0.00018136158,0.0057661114,0.00011483301,0.000056675908,0.000074539465,0.00014430017,0.0007280206],"category_scores_gemma":[0.000098190976,0.000041451727,0.00003155383,0.024592103,0.0002598689,0.00006670823,0.000020561418,0.0003092271,0.0003188932],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007899571,0.017556489,0.62950605,0.0015232813,0.0031714253,0.00014617917,0.009390732,0.00011639936,0.27085873,0.011555903,0.014122225,0.041973606],"study_design_scores_gemma":[0.0012681363,0.00033326112,0.94418174,0.00019882878,0.00062256184,6.056692e-7,0.0011577996,0.04257165,0.006762675,0.0010302183,0.00167488,0.00019764899],"about_ca_topic_score_codex":0.019647056,"about_ca_topic_score_gemma":0.033011086,"teacher_disagreement_score":0.3146757,"about_ca_system_score_codex":0.00025195128,"about_ca_system_score_gemma":0.0006017285,"threshold_uncertainty_score":0.99614066},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W2082172830","doi":"10.12927/hcpap..16834","title":"Bibliometric Analysis of Research on Mental Health in the Workplace in Canada, 1991-2002","year":2004,"lang":"en","type":"article","venue":"A Nudge Too Far? A Nudge at All? On Paying People to Be Healthy","topic":"Occupational Health and Safety Research","field":"Health Professions","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"Université de Montréal; Université Laval; McMaster University; Université du Québec à Montréal; University of British Columbia; University of Toronto; Espace pour la vie","funders":"University of Toronto; York University; Centre for Addiction and Mental Health; Université du Québec à Montréal; Université de Montréal; Acadia University; Université Laval","keywords":"Public health; Population health; Health care; Equity (law); Mental health; Health policy; Library science; Health equity; Sociology; Political science; Medicine; Nursing; Psychiatry","score_opus":0.18315363909460605,"score_gpt":0.5055748664173002,"score_spread":0.3224212273226942,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2082172830","genre_codex":"empirical","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5256368,0.00044381162,0.0000063677335,0.46898168,0.0002480903,0.0027506214,0.00034158645,0.000023236227,0.0015678353],"genre_scores_gemma":[0.4610572,0.001144122,0.00011252977,0.5361849,0.000027336313,0.00076210813,0.00029591646,0.00004507912,0.00037080862],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.98555136,0.0045592864,0.0022806162,0.0010867636,0.0032030889,0.0033188937],"domain_scores_gemma":[0.98746824,0.009195423,0.0004820053,0.0011988541,0.00041654447,0.0012389083],"candidate_categories":["metaepi_narrow","bibliometrics","sts","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.014074375,0.0004309301,0.001296516,0.052441128,0.0017327953,0.000026884301,0.0010569656,0.00029571896,0.00052724546],"category_scores_gemma":[0.0015106425,0.000344415,0.00018655862,0.19057623,0.00007607155,0.00013883723,0.00033969895,0.0029272616,0.00040420698],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.0056432174,0.00123925,0.30492774,0.00062517467,0.00012026748,0.000045953104,0.020294629,0.0016718904,0.000008267832,0.002678449,0.6535808,0.009164402],"study_design_scores_gemma":[0.0030465026,0.001672446,0.88862705,0.00002025895,0.000014279629,0.0000024651556,0.005524685,0.00062773825,0.0000022317747,0.000055068376,0.10012903,0.00027826003],"about_ca_topic_score_codex":0.9200066,"about_ca_topic_score_gemma":0.9840148,"teacher_disagreement_score":0.5836993,"about_ca_system_score_codex":0.016271379,"about_ca_system_score_gemma":0.013475612,"threshold_uncertainty_score":0.99990076},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"design_other","genre":"other","about_ca_system":false,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W2090960911","doi":"10.3152/095820210x492495","title":"Which scientific elites? On the concentration of research funds, publications and citations","year":2010,"lang":"en","type":"article","venue":"Research Evaluation","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":100,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Library science; Political science; Sociology; Computer science","score_opus":0.9310062741247978,"score_gpt":0.7398197713321439,"score_spread":0.1911865027926538,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2090960911","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93226904,0.00021742264,0.00021727904,0.039126877,0.00039734074,0.0016017536,0.00004117568,0.000015729473,0.026113413],"genre_scores_gemma":[0.99744844,0.000069896916,0.0002783833,0.000025784844,0.000116332136,0.00024580472,0.000034445435,0.000009335355,0.001771596],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9422621,0.0059069796,0.0009780321,0.0012535495,0.048438225,0.0011610787],"domain_scores_gemma":[0.8053127,0.071029775,0.0003242964,0.0029536593,0.119701736,0.0006778389],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.3787386,0.000099321485,0.0001594555,0.038906608,0.0026283404,0.006547365,0.0021530583,0.00014845052,0.001876795],"category_scores_gemma":[0.48540914,0.00006166334,0.000054751967,0.23699751,0.0021143337,0.0009025843,0.0005463376,0.0014405551,0.000449815],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000052261053,0.00055435323,0.052162718,0.000015776837,0.00002269403,3.1813167e-7,0.0016435173,0.000084688574,0.1486854,0.49824652,0.15359618,0.14493558],"study_design_scores_gemma":[0.0005898558,0.00037971154,0.5472916,0.00002646997,0.0000068312456,0.0000028312909,0.0030066273,0.14016834,0.010703401,0.25717947,0.04049361,0.00015125159],"about_ca_topic_score_codex":0.00009542593,"about_ca_topic_score_gemma":0.0004928461,"teacher_disagreement_score":0.49512887,"about_ca_system_score_codex":0.00015567149,"about_ca_system_score_gemma":0.001654821,"threshold_uncertainty_score":0.9990356},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2091428252","doi":"10.1093/reseval/rvt017","title":"The New Zealand performance-based research fund and its impact on publication activity in economics","year":2013,"lang":"en","type":"article","venue":"Research Evaluation","topic":"New Zealand Economic and Social Studies","field":"Economics, Econometrics and Finance","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"Pennington Biomedical Research Foundation","keywords":"Queen (butterfly); Library science; Management; Sociology; Media studies; Economics; Computer science","score_opus":0.41028521807876034,"score_gpt":0.4383793823756601,"score_spread":0.02809416429689976,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2091428252","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96934795,0.00060747366,0.0000027330816,0.013139114,0.00007246811,0.001028092,0.000010664357,0.000008211368,0.015783297],"genre_scores_gemma":[0.9930647,0.002704796,0.0000053427775,0.000031572934,0.0001652153,0.00030998237,0.000012534736,0.000013172233,0.0036926593],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99830174,0.00025034245,0.0003260877,0.0003664165,0.00015949883,0.00059589144],"domain_scores_gemma":[0.9983696,0.0008107208,0.000110340996,0.00025517013,0.00031478115,0.000139428],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.013400372,0.00010174062,0.00019813067,0.0004547358,0.0006886644,0.00041520895,0.00020168051,0.00009824517,0.00033612753],"category_scores_gemma":[0.0009101593,0.00008338323,0.000040443578,0.00038909618,0.000108705266,0.0005116423,0.00007446357,0.00043331095,0.00092556543],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00050508545,0.00030035694,0.41221863,0.000039512553,0.00012841263,2.760394e-7,0.0024704544,0.0024728447,0.00007020968,0.0440835,0.08263074,0.45508],"study_design_scores_gemma":[0.0009744192,0.00025277032,0.7686061,0.000011383573,0.0000011412575,1.7991674e-7,0.00011432947,0.18036819,0.000079354875,0.03457054,0.014901553,0.00012002102],"about_ca_topic_score_codex":0.002679993,"about_ca_topic_score_gemma":0.00045730226,"teacher_disagreement_score":0.45495996,"about_ca_system_score_codex":0.0009961287,"about_ca_system_score_gemma":0.00026811048,"threshold_uncertainty_score":0.9998523},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W2093566717","doi":"10.3163/1536-5050.103.1.010","title":"Into the gray: a modified approach to citation analysis to better understand research impact","year":2015,"lang":"en","type":"article","venue":"Journal of the Medical Library Association JMLA","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Computer science; Gray (unit); Citation; Information retrieval; Data science; Citation analysis; World Wide Web; Library science; Medicine","score_opus":0.6302172774255392,"score_gpt":0.583660032386797,"score_spread":0.046557245038742257,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2093566717","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.74966776,0.00018832326,0.027308712,0.21408892,0.0007140872,0.0003369646,0.00001358709,0.000008773129,0.0076728933],"genre_scores_gemma":[0.9920156,0.00002151603,0.0013340836,0.002725591,0.00048085928,0.000004707658,0.0000021607868,0.0000093458875,0.0034061135],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9631272,0.0030154127,0.0009994006,0.00029716658,0.03206186,0.0004989923],"domain_scores_gemma":[0.982666,0.0105882855,0.0006973409,0.00060971023,0.0037697612,0.0016689078],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.105182365,0.00011137134,0.00039296123,0.027981184,0.0003727406,0.0029022617,0.004137531,0.00017699825,0.00022972804],"category_scores_gemma":[0.18858373,0.000046000976,0.0004765926,0.2183856,0.000094205156,0.001086068,0.0010137786,0.0009809359,0.000089166024],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016194161,0.00022984997,0.20223299,0.0000015670886,0.00048681884,0.000007631123,0.0054718256,0.0041449447,0.000025465077,0.0009054181,0.7681853,0.018146241],"study_design_scores_gemma":[0.0018080993,0.0009031492,0.7202528,0.000032124546,0.00017365026,0.00001602303,0.009996706,0.044896863,0.00017868498,0.1499992,0.07142688,0.00031579513],"about_ca_topic_score_codex":0.000057464484,"about_ca_topic_score_gemma":0.000009886589,"teacher_disagreement_score":0.69675845,"about_ca_system_score_codex":0.0007689582,"about_ca_system_score_gemma":0.001176594,"threshold_uncertainty_score":0.9981328},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2096273440","doi":"10.1186/1471-2458-13-541","title":"Traffic medicine–related research: a scientometric analysis","year":2013,"lang":"en","type":"article","venue":"BMC Public Health","topic":"Healthcare Systems and Public Health","field":"Medicine","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Public health; Biostatistics; Medicine; Productivity; Regional science; Bibliometrics; Web of science; Environmental health; Library science; Economic growth; Geography; Economics; Meta-analysis","score_opus":0.3371280988953731,"score_gpt":0.48572741007071285,"score_spread":0.14859931117533975,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2096273440","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6506792,0.0032388724,0.0011885813,0.31495535,0.00081052165,0.0038620653,0.000017539576,0.00056090235,0.02468697],"genre_scores_gemma":[0.98677087,0.00026718876,0.0009946283,0.005342573,0.0005719404,0.0002509891,0.00017732412,0.00005595972,0.005568546],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9880914,0.002940032,0.0020667817,0.0010770464,0.0028687175,0.0029559885],"domain_scores_gemma":[0.98901963,0.0010305303,0.00040871045,0.0014568336,0.0022381733,0.005846104],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0312199,0.00032733835,0.0014859696,0.015227851,0.0008074744,0.0002000777,0.00046317684,0.00036574734,0.0047413195],"category_scores_gemma":[0.0064507793,0.00024208704,0.0003045674,0.0549935,0.00046252384,0.00045574817,0.00009951075,0.0013166388,0.0015140509],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000028065997,0.0012678304,0.19237581,0.0035305873,0.00093128684,0.00003842018,0.0115832575,0.000026522324,0.0000192974,0.018420065,0.29982516,0.4719537],"study_design_scores_gemma":[0.002089375,0.0015779132,0.64168864,0.00018151978,0.000054321292,0.0001605203,0.0040349374,0.009329032,2.9588867e-7,0.00015499524,0.3404475,0.00028094032],"about_ca_topic_score_codex":0.035507217,"about_ca_topic_score_gemma":0.0013135456,"teacher_disagreement_score":0.47167274,"about_ca_system_score_codex":0.0018054815,"about_ca_system_score_gemma":0.016499363,"threshold_uncertainty_score":0.9992634},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2097082745","doi":"10.1109/picmet.1999.808098","title":"Knowledge network of contemporary interdisciplinary study of organization and technology: from bibliometrics to epistemology","year":2003,"lang":"en","type":"article","venue":"","topic":"University-Industry-Government Innovation Models","field":"Business, Management and Accounting","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Bibliometrics; Context (archaeology); Field (mathematics); Knowledge management; Subject (documents); Domain (mathematical analysis); Sociology; Outcome (game theory); Engineering ethics; Epistemology; Management science; Computer science; Engineering; Mathematics; Philosophy","score_opus":0.035073731801365006,"score_gpt":0.260684868587617,"score_spread":0.22561113678625203,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2097082745","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9685039,0.000040943756,0.0051872237,0.00033371442,0.00014480138,0.00031705046,0.0000021863973,0.0000503664,0.02541984],"genre_scores_gemma":[0.9990058,9.676369e-7,0.00043882974,0.00010037719,0.00005551318,0.0000014296345,0.00000969944,0.000012377672,0.00037503356],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9992377,0.000015404374,0.00030477718,0.00020325513,0.00013726162,0.00010159445],"domain_scores_gemma":[0.99892974,0.000054018157,0.000262213,0.0001954874,0.0005505305,0.000007992451],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0002390378,0.00009891418,0.00022486462,0.0030537,0.00007392625,0.000023518556,0.00016907915,0.00014973978,0.00014512983],"category_scores_gemma":[0.0001840178,0.000099319215,0.000009498656,0.02323867,0.000044504417,0.00040294247,0.0006312966,0.00011045174,0.000011226962],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017220673,0.00043635952,0.9493232,0.000029715937,0.00005660238,0.000003532957,0.00013818603,0.00007972441,0.00045908353,0.040576596,0.008467877,0.0004119493],"study_design_scores_gemma":[0.01631482,0.0017219504,0.75699234,0.00047500996,0.0006441697,0.000011183742,0.10053038,0.004713144,0.0079429895,0.06553913,0.042732462,0.0023824165],"about_ca_topic_score_codex":0.000057813108,"about_ca_topic_score_gemma":0.000024448665,"teacher_disagreement_score":0.1923308,"about_ca_system_score_codex":0.000027214202,"about_ca_system_score_gemma":0.00002409947,"threshold_uncertainty_score":0.9975229},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2097269731","doi":"10.1023/b:scie.0000027677.79173.b8","title":"The classification of biomedical journals by research level","year":2004,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":89,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Science North","funders":"","keywords":"Computer science; Citation; Science Citation Index; Information retrieval; Basic research; Data science; Library science","score_opus":0.9272708597111224,"score_gpt":0.7093757539334359,"score_spread":0.21789510577768656,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2097269731","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8835268,0.01285592,0.06419225,0.020682685,0.0030844128,0.0008100205,0.00031111806,0.00004413114,0.0144926775],"genre_scores_gemma":[0.99374163,0.0012607563,0.0012231361,0.000047540336,0.000102174934,0.000012813608,0.0000053089684,0.000012666737,0.0035939931],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9392418,0.0007403781,0.0016692668,0.0009770918,0.05584848,0.0015229603],"domain_scores_gemma":[0.95609325,0.0199717,0.00067840895,0.0018405326,0.020194031,0.0012220936],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.20571366,0.00015457203,0.00032768692,0.1777989,0.0015367926,0.004603872,0.007405369,0.00021015077,0.00020755702],"category_scores_gemma":[0.30833465,0.000084796695,0.00020159756,0.7157363,0.002311401,0.0006741668,0.0011950918,0.000835082,0.0006112434],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003475098,0.00084697886,0.02419321,0.0000072707385,0.000025773312,0.0000072678627,0.00036233422,0.000033834884,0.053912126,0.019815782,0.26467213,0.63608855],"study_design_scores_gemma":[0.0013741108,0.0006347309,0.40077558,0.000026228985,0.00000609612,0.00001834893,0.0022291394,0.0013206109,0.01171471,0.08020875,0.50139576,0.00029593564],"about_ca_topic_score_codex":0.0001161045,"about_ca_topic_score_gemma":0.0000130394565,"teacher_disagreement_score":0.6357926,"about_ca_system_score_codex":0.0005187356,"about_ca_system_score_gemma":0.0010557031,"threshold_uncertainty_score":0.9997631},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2097363517","doi":"10.1186/1472-6947-14-71","title":"Publication trends of shared decision making in 15 high impact medical journals: a full-text review with bibliometric analysis","year":2014,"lang":"en","type":"review","venue":"BMC Medical Informatics and Decision Making","topic":"Patient-Provider Communication in Healthcare","field":"Health Professions","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre hospitalier universitaire de Québec; Université Laval","funders":"Université de Lausanne; Canadian Medical Association; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Science Foundation","keywords":"Health informatics; Bibliometrics; Computer science; Data science; Library science; Medicine; Public health; Nursing","score_opus":0.3016296756737849,"score_gpt":0.5506934911576157,"score_spread":0.24906381548383077,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2097363517","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00015404093,0.8554952,0.14216752,0.00009531714,0.00024633625,0.0012358912,0.00008406202,0.0000491578,0.0004724375],"genre_scores_gemma":[0.0013990246,0.9181416,0.078498736,0.0010824583,0.00013322243,0.00033158853,0.0003456233,0.00005988924,0.000007857983],"study_design_codex":"design_other","study_design_gemma":"systematic_review","domain_scores_codex":[0.97974235,0.0023307486,0.010546868,0.0005336094,0.005918692,0.0009277338],"domain_scores_gemma":[0.9534036,0.032854762,0.008712329,0.0022500497,0.0016696936,0.0011095442],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["bibliometrics","research_integrity"],"category_scores_codex":[0.012818912,0.0007271379,0.0051876735,0.07440489,0.00052018533,0.00015163014,0.002155026,0.0016637127,0.007004231],"category_scores_gemma":[0.034766767,0.0004482322,0.00081773946,0.120653875,0.00017892345,0.0006571285,0.001444458,0.0031714726,0.000095127354],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000075275566,0.000063040316,0.0018207935,0.045063436,0.00019472669,0.000004191807,0.0003187271,0.000013705265,5.907229e-10,0.000067841414,0.0040967967,0.94828147],"study_design_scores_gemma":[0.0018120169,0.00027956328,0.004459455,0.7773143,0.0018936575,0.00012503444,0.00045691733,0.06933034,1.7069148e-9,0.0003831374,0.14320314,0.00074244896],"about_ca_topic_score_codex":0.00008013262,"about_ca_topic_score_gemma":0.0006775955,"teacher_disagreement_score":0.94753903,"about_ca_system_score_codex":0.00056633656,"about_ca_system_score_gemma":0.004119029,"threshold_uncertainty_score":0.9997969},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2098698373","doi":"10.1002/meet.14504201210","title":"Going beyond counting first authors in author co‐citation analysis","year":2005,"lang":"en","type":"article","venue":"Proceedings of the American Society for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Citation; Field (mathematics); Citation analysis; Computer science; Statistics; Information retrieval; Data science; Mathematics; Library science; Pure mathematics","score_opus":0.14995783970350976,"score_gpt":0.4770494582581099,"score_spread":0.3270916185546001,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2098698373","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98652,0.000028647804,0.0011058132,0.010460489,0.000042348176,0.00028518084,0.000014535169,0.000027320819,0.0015156817],"genre_scores_gemma":[0.9893425,0.000049000715,0.010179128,0.00031621137,0.000010977989,0.000026912821,7.670372e-7,0.0000023800624,0.00007213395],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9949317,0.000003869618,0.0007239047,0.0003104339,0.0035594136,0.0004706891],"domain_scores_gemma":[0.9924393,0.0006072313,0.0011067387,0.00018754322,0.0055662426,0.00009292149],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0227502,0.0000947189,0.00029196343,0.029348483,0.0006357014,0.0009612847,0.0018426833,0.00007400681,0.0000044121443],"category_scores_gemma":[0.024711829,0.00006355551,0.00017286676,0.306117,0.0021970787,0.0038559188,0.00045951997,0.00018750357,0.000006640835],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000015899888,0.00003961,0.63791406,0.000024208446,0.00004468492,1.2936837e-8,0.0032391013,0.00014571012,0.0044953725,0.0265592,0.0061969236,0.32132518],"study_design_scores_gemma":[0.00084159547,0.00030771262,0.3516653,0.000023559052,0.000080945625,0.0000067632222,0.049463943,0.43674913,0.017814472,0.014295969,0.12832542,0.00042520394],"about_ca_topic_score_codex":0.000025716832,"about_ca_topic_score_gemma":0.00000952717,"teacher_disagreement_score":0.43660343,"about_ca_system_score_codex":0.00020469986,"about_ca_system_score_gemma":0.00017865808,"threshold_uncertainty_score":0.98350346},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2103013209","doi":"10.1007/s11192-009-0020-3","title":"Matrices of science and technology interactions and patterns of structured growth: implications for development","year":2009,"lang":"en","type":"article","venue":"Scientometrics","topic":"Economic and Technological Innovation","field":"Economics, Econometrics and Finance","cited_by":41,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Conselho Nacional de Desenvolvimento Científico e Tecnológico; International Development Research Centre","keywords":"Dimension (graph theory); Articulation (sociology); Computer science; Component (thermodynamics); Key (lock); Scientific literature; Regional science; Data science; Sociology; Political science; Mathematics","score_opus":0.059368761363235184,"score_gpt":0.2929687513041478,"score_spread":0.2335999899409126,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2103013209","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9810764,0.00040747953,0.016514406,0.001149251,0.00008134004,0.0001674228,0.00007876622,0.00001988226,0.0005050513],"genre_scores_gemma":[0.9869646,0.00007008525,0.012886064,0.00003253387,0.0000044557446,0.000008915634,0.0000031113946,0.0000022790662,0.000027960978],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9991687,9.019705e-7,0.0004069901,0.00025299116,0.00003301316,0.00013741784],"domain_scores_gemma":[0.9990955,0.00005042479,0.0003480884,0.00013639881,0.00033798648,0.000031574968],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006003584,0.000057030506,0.00017059423,0.0045008603,0.00012032038,0.000028164734,0.00020979176,0.000054154545,0.0000054227935],"category_scores_gemma":[0.0009311485,0.000058965335,0.0000118492335,0.006246259,0.0003083748,0.0001916394,0.00006845313,0.000049302747,7.559952e-7],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000011125779,0.000026925389,0.14452596,0.000014609753,0.0000037826615,1.0353388e-8,0.0000502418,4.1116283e-7,0.0014468271,0.8344076,0.00001548974,0.019507024],"study_design_scores_gemma":[0.00023735041,0.00010027926,0.6831429,0.000008368225,0.000002848495,0.0000026379776,0.00008451492,0.00019471004,0.012804368,0.3013046,0.0020168035,0.000100617195],"about_ca_topic_score_codex":0.0000062163804,"about_ca_topic_score_gemma":0.0000023183431,"teacher_disagreement_score":0.53861696,"about_ca_system_score_codex":0.00005920239,"about_ca_system_score_gemma":0.000029424438,"threshold_uncertainty_score":0.4016071},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2106923054","doi":"10.1002/asi.20076","title":"Webometrics: An introduction to the special issue","year":2004,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Webometrics; Bibliometrics; Field (mathematics); Data science; Computer science; Web of science; Range (aeronautics); Information retrieval; World Wide Web; Political science; Engineering; Mathematics; MEDLINE","score_opus":0.01046146985079431,"score_gpt":0.27128509353874175,"score_spread":0.2608236236879474,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2106923054","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.26172873,0.000022918044,0.3859853,0.34949937,0.0019304935,0.00045762124,0.000004537912,0.000056278237,0.00031474145],"genre_scores_gemma":[0.7828587,0.00016136545,0.19481218,0.018440109,0.0036709353,0.000014034176,4.958213e-7,0.000005254761,0.00003689513],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99873734,0.0000070283054,0.0003583836,0.00009922883,0.0005923691,0.00020567462],"domain_scores_gemma":[0.9977257,0.000039629063,0.0005887867,0.00041991367,0.0011577343,0.00006818203],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025824618,0.000066329085,0.00012965809,0.0006907399,0.0005754149,0.00029165266,0.0018774413,0.00003402517,0.0000010100804],"category_scores_gemma":[0.0015060608,0.000036527967,0.00009731596,0.01198717,0.0009048963,0.004464386,0.0002957023,0.00020088883,0.000006663853],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013631868,0.00003601709,0.00018187957,0.0000069393295,0.00001554641,6.502392e-8,0.0043190112,0.00090072694,0.00046416352,0.11720789,0.03289913,0.843955],"study_design_scores_gemma":[0.00031748725,0.0009912282,0.0020298825,0.000003680774,0.00000914758,0.00011882021,0.0054615163,0.001830986,0.0037196425,0.007392651,0.9780267,0.00009828901],"about_ca_topic_score_codex":0.0000061410697,"about_ca_topic_score_gemma":0.0000016832175,"teacher_disagreement_score":0.94512755,"about_ca_system_score_codex":0.00019111489,"about_ca_system_score_gemma":0.00040788957,"threshold_uncertainty_score":0.5759438},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2108226858","doi":"10.5597/lajam00174","title":"Aquatic mammal science in Latin America: a bibliometric analysis for the first eight years of the Latin American Journal of Aquatic Mammals (2002-2010)","year":2014,"lang":"en","type":"article","venue":"Latin American Journal of Aquatic Mammals","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Bycatch; Geography; Latin Americans; Fishery; Marine mammal; Library science; Biology; Fishing; Political science","score_opus":0.022301160088785642,"score_gpt":0.25980178962013145,"score_spread":0.23750062953134582,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2108226858","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9846815,0.00011786904,0.011603762,0.0020908297,0.00036170724,0.000457355,0.000035550638,0.000007687235,0.0006437664],"genre_scores_gemma":[0.99384844,0.00018646577,0.005061396,0.00049043674,0.00014042002,0.000017105267,0.0000024925275,0.000034230336,0.00021901581],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9934618,0.0005192416,0.0023747284,0.00040352036,0.0024036493,0.00083704933],"domain_scores_gemma":[0.9894044,0.0033656731,0.0059175007,0.00080549915,0.00017006595,0.00033686688],"candidate_categories":["metaepi_narrow","bibliometrics","sts","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0048938827,0.0003838511,0.0014814357,0.0073079765,0.00030407112,0.000169892,0.0021458887,0.000060397044,0.004036989],"category_scores_gemma":[0.004576823,0.00024919832,0.0007499118,0.078155756,0.0037237853,0.00035731355,0.00030240655,0.0005222298,0.00006629591],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030187634,0.00044550153,0.8370033,0.000033080603,0.00038619855,0.0000110964065,0.002561926,0.0059586926,0.0022730052,0.0001417778,0.0012239661,0.14965957],"study_design_scores_gemma":[0.0010709168,0.0014553558,0.970055,0.00011355733,0.00063045556,0.000026535003,0.005378087,0.017964786,0.00044671475,0.00018393502,0.0023548657,0.0003198172],"about_ca_topic_score_codex":0.0018287366,"about_ca_topic_score_gemma":0.0010677269,"teacher_disagreement_score":0.14933975,"about_ca_system_score_codex":0.00065300503,"about_ca_system_score_gemma":0.0002264627,"threshold_uncertainty_score":0.999996},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2109757953","doi":"10.1007/s11192-011-0515-6","title":"Exploring the web visibility of world-class universities","year":2011,"lang":"en","type":"article","venue":"Scientometrics","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":44,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"University of British Columbia","keywords":"Visibility; Proxy (statistics); The Internet; Computer science; World Wide Web; Test (biology); Corporate governance; Information retrieval; Geography; Business","score_opus":0.22676155982887738,"score_gpt":0.28013886868423954,"score_spread":0.053377308855362154,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2109757953","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.86196464,0.00012351816,0.04310448,0.00018955719,0.002349965,0.00020496685,0.000009140083,0.00015244978,0.091901295],"genre_scores_gemma":[0.98916465,0.0000423004,0.009839698,0.00009632475,0.000024994924,0.0000027278224,3.985235e-7,0.0000036649403,0.00082526426],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99794966,0.00006681293,0.00032867794,0.00033900104,0.0009580336,0.00035778582],"domain_scores_gemma":[0.99757284,0.0004061844,0.00016220628,0.0011359314,0.00059703,0.00012579268],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0028797481,0.00011450955,0.00016174895,0.0043100584,0.0002076769,0.00011462221,0.00232329,0.000027123171,0.000041612726],"category_scores_gemma":[0.00083309424,0.00008257484,0.000113291855,0.042302538,0.00030691255,0.002138501,0.0007919016,0.00017252592,0.000046537873],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000033422264,0.00060147897,0.054225236,0.00013973001,0.000057889207,0.00000983531,0.021356717,0.000113446906,0.0002436468,0.7705525,0.0027188426,0.14994726],"study_design_scores_gemma":[0.0011136519,0.0007071398,0.824467,0.00005757432,0.000041561987,0.000007844661,0.0051981555,0.048310816,0.035971504,0.011144331,0.072054066,0.00092633045],"about_ca_topic_score_codex":0.00008843798,"about_ca_topic_score_gemma":0.00002615241,"teacher_disagreement_score":0.7702418,"about_ca_system_score_codex":0.00016704066,"about_ca_system_score_gemma":0.00029357226,"threshold_uncertainty_score":0.9780538},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2110458718","doi":"10.1136/ebm.13.1.3","title":"Less is more: where do the abstracts in the EBM journal come from?","year":2008,"lang":"en","type":"article","venue":"Evidence-Based Medicine","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Information retrieval; Psychology; History; Medicine; Computer science","score_opus":0.8225130300919945,"score_gpt":0.533793031563296,"score_spread":0.28871999852869845,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2110458718","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.65031856,0.058130402,0.002230668,0.28481272,0.00078734383,0.0008294083,0.0000073312385,0.000005227729,0.0028783032],"genre_scores_gemma":[0.9817999,0.0016942931,0.00035230714,0.012962475,0.000993172,0.00002851397,0.0000015766552,0.000010380094,0.0021573545],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9712953,0.0095728515,0.007614337,0.00070353266,0.01044236,0.00037162603],"domain_scores_gemma":[0.9627073,0.027984455,0.004138571,0.0039983485,0.00094540225,0.0002259632],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch","insufficient_payload"],"category_scores_codex":[0.13924359,0.0003237892,0.002343377,0.0003136965,0.00045133935,0.00046779105,0.004147125,0.000092950526,0.018370153],"category_scores_gemma":[0.054599598,0.00009501126,0.00086430233,0.001911529,0.0005044565,0.00039933447,0.00005952537,0.00072174415,0.0012595712],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003135746,0.00006315833,0.09349565,0.000032616288,0.000089673886,0.00031302607,0.009807767,0.00032752214,0.00008148225,0.0000988805,0.8745691,0.021089742],"study_design_scores_gemma":[0.0012848966,0.0002181178,0.5526252,0.0029927364,0.00046587252,0.0003653626,0.02321186,0.0023067824,0.000057225887,0.005009971,0.41114086,0.0003211353],"about_ca_topic_score_codex":0.00035890625,"about_ca_topic_score_gemma":0.0001461563,"teacher_disagreement_score":0.46342826,"about_ca_system_score_codex":0.000056124594,"about_ca_system_score_gemma":0.00027543269,"threshold_uncertainty_score":0.99951804},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2112801688","doi":"10.2106/jbjs.e.00370","title":"Level of Evidence in Orthopaedic Journals","year":2005,"lang":"en","type":"article","venue":"Journal of Bone and Joint Surgery","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":324,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Hamilton Health Sciences; Hamilton General Hospital","funders":"","keywords":"Medicine; Grading (engineering); Evidence-based medicine; Kappa; MEDLINE; Rating system; Family medicine; Alternative medicine; Pathology","score_opus":0.9531000786402971,"score_gpt":0.5564152615563063,"score_spread":0.39668481708399084,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112801688","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9631046,0.027914798,0.0029505857,0.005438004,0.0002356293,0.000083938154,0.000002650342,4.472583e-7,0.00026932135],"genre_scores_gemma":[0.99258816,0.0033126245,0.0023139915,0.0002623676,0.00020793531,7.107236e-7,6.935122e-8,0.0000047220756,0.001309394],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9722267,0.0036192245,0.019503092,0.00023443815,0.0042332807,0.00018330694],"domain_scores_gemma":[0.9737484,0.008518032,0.015420737,0.0006922198,0.0014268486,0.00019376741],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.2668508,0.00014650385,0.0048884894,0.0018208473,0.00003723837,0.0002045148,0.0001654507,0.0000581372,0.003179322],"category_scores_gemma":[0.07878583,0.00006177694,0.0025143446,0.0010111011,0.00005143153,0.0007728412,0.000053544893,0.00024643567,0.00007655667],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000045698212,0.00013375246,0.57936686,0.00012255664,0.00015458468,0.00014996824,0.00053375436,0.00016472749,0.0021097788,0.000057621353,0.14458895,0.27257177],"study_design_scores_gemma":[0.00027027066,0.00006278293,0.9459771,0.002440015,0.00016786439,0.0011542978,0.0005938788,0.00059701095,0.00037849892,0.002357909,0.045823283,0.00017710452],"about_ca_topic_score_codex":0.000007342512,"about_ca_topic_score_gemma":0.000028879589,"teacher_disagreement_score":0.36661026,"about_ca_system_score_codex":0.000021510294,"about_ca_system_score_gemma":0.00018204318,"threshold_uncertainty_score":0.9977319},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2114152365","doi":"10.1097/acm.0b013e3182820b5c","title":"A Bibliometric Analysis of Evaluative Medical Education Studies","year":2013,"lang":"en","type":"article","venue":"Academic Medicine","topic":"Innovations in Medical Education","field":"Medicine","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Children's Hospital of Eastern Ontario","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"Bibliometrics; Higher education; MEDLINE; Medical education; Psychology; Medicine; Library science; Computer science; Political science","score_opus":0.08303712664523774,"score_gpt":0.5026266621435435,"score_spread":0.4195895354983057,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2114152365","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94808567,0.008289491,0.0006697622,0.038446948,0.0009102919,0.00060838676,7.806179e-7,0.00004716677,0.0029414908],"genre_scores_gemma":[0.9813728,0.0033326205,0.00082179595,0.011538998,0.00070277695,0.00023382898,0.00013140743,0.000016542808,0.0018492716],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9960619,0.0001151162,0.0011457814,0.00032636902,0.0020929324,0.00025790287],"domain_scores_gemma":[0.99514395,0.0004992812,0.00043066495,0.00043732207,0.0032456138,0.00024319545],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0026286794,0.0001763806,0.0008696986,0.08438297,0.000052376137,0.0000021282435,0.00024117492,0.0003202356,0.00814369],"category_scores_gemma":[0.048838545,0.000121254096,0.0000848098,0.23140942,0.00069665274,0.00015188563,0.000059879487,0.00076172414,0.000092646274],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003194968,0.0001882523,0.12290528,0.0003497313,0.0035757374,0.0000018791584,0.007391312,0.000002995839,0.0013415591,0.004195453,0.4290497,0.43096614],"study_design_scores_gemma":[0.0026240402,0.00077228376,0.9272122,0.0020660271,0.009536675,0.00006528927,0.030317701,0.010711804,0.00051952456,0.0054351604,0.010417745,0.00032159546],"about_ca_topic_score_codex":0.0001916235,"about_ca_topic_score_gemma":0.0000017600562,"teacher_disagreement_score":0.80430686,"about_ca_system_score_codex":0.00020347422,"about_ca_system_score_gemma":0.000979021,"threshold_uncertainty_score":0.992763},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2122612712","doi":"10.1371/journal.pone.0016937","title":"Is Qualitative Research Second Class Science? A Quantitative Longitudinal Examination of Qualitative Research in Medical Journals","year":2011,"lang":"en","type":"article","venue":"PLoS ONE","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":112,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Qualitative research; Impact factor; Qualitative property; Qualitative analysis; Longitudinal study; Psychology; Medicine; Social science; Sociology; Statistics; Mathematics; Pathology; Biology","score_opus":0.9881890202619653,"score_gpt":0.7469333192546113,"score_spread":0.24125570100735405,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2122612712","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94245565,0.0008785365,0.00068370294,0.0019448369,0.000043956796,0.0010359505,0.00004102412,0.000002928739,0.052913398],"genre_scores_gemma":[0.9849542,0.000057457648,0.010611386,0.00005977202,0.000039285223,0.00011711236,0.000001992283,0.000013865777,0.0041449605],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.5424928,0.31582248,0.021427786,0.0037219245,0.114401825,0.0021331455],"domain_scores_gemma":[0.7953251,0.13460629,0.007049821,0.0054892194,0.05630734,0.0012222616],"candidate_categories":["metaresearch","sts","insufficient_payload"],"consensus_categories":["metaresearch","insufficient_payload"],"category_scores_codex":[0.8078911,0.00025942648,0.0033400846,0.0059859203,0.00046850677,0.0005160672,0.0037956068,0.00019446142,0.044981696],"category_scores_gemma":[0.32237133,0.00014978564,0.0005041219,0.015541931,0.0034338476,0.0011215091,0.00056941115,0.0013756694,0.002436257],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009480126,0.0025279217,0.0061666723,0.00029472285,0.0005432023,0.000036893445,0.91394323,2.8903818e-7,0.0035207577,0.06650325,0.0044330293,0.0019352413],"study_design_scores_gemma":[0.0006814597,0.0011068296,0.07232341,0.0015449036,0.00008883067,0.0000045113393,0.64539284,0.0040640347,0.010963294,0.2633777,0.0001351412,0.00031704886],"about_ca_topic_score_codex":0.00018079713,"about_ca_topic_score_gemma":0.000599915,"teacher_disagreement_score":0.48551977,"about_ca_system_score_codex":0.00025649098,"about_ca_system_score_gemma":0.0010421086,"threshold_uncertainty_score":0.99927825},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W2128702800","doi":"10.1186/s12887-015-0324-9","title":"Levels of evidence: a comparison between top medical journals and general pediatric journals","year":2015,"lang":"en","type":"article","venue":"BMC Pediatrics","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Medicine; Subspecialty; Consolidated Standards of Reporting Trials; Pediatrics; Randomized controlled trial; MEDLINE; Clinical trial; Family medicine; Evidence-based medicine; Medical journal; Alternative medicine; Internal medicine","score_opus":0.9443336505726487,"score_gpt":0.6247682883888389,"score_spread":0.3195653621838098,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2128702800","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9098965,0.052146867,0.035842378,0.00082434656,0.0004915111,0.00034304708,0.0000359087,0.000005856983,0.0004136203],"genre_scores_gemma":[0.97988635,0.001711279,0.013043319,0.000115589326,0.0035741455,0.0000066051844,0.0000011414357,0.000015459938,0.0016461167],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.95902634,0.009149987,0.013294515,0.0007971254,0.017291654,0.00044035647],"domain_scores_gemma":[0.97039866,0.013651309,0.009155054,0.0018945313,0.0035091669,0.0013912672],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.22209191,0.0003291461,0.0041841925,0.0011579164,0.0001232105,0.00087048305,0.0021791633,0.00028778284,0.0045782602],"category_scores_gemma":[0.23085575,0.00015452877,0.0010802074,0.00331999,0.0000771309,0.00053151476,0.0003951081,0.00046479396,0.00050562574],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000045318598,0.000029314417,0.9138138,0.00006993635,0.00002935071,0.0000046780547,0.00033415365,0.00007195775,0.0000027183955,0.000034987017,0.077034675,0.008569879],"study_design_scores_gemma":[0.000778124,0.00016870357,0.9755319,0.00007012267,0.0015445249,0.00007623374,0.0007186201,0.005032858,0.000017668368,0.0053191427,0.010315428,0.00042671838],"about_ca_topic_score_codex":0.000015642327,"about_ca_topic_score_gemma":0.000013415102,"teacher_disagreement_score":0.06998988,"about_ca_system_score_codex":0.000043063355,"about_ca_system_score_gemma":0.0009364402,"threshold_uncertainty_score":0.9963317},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2131615921","doi":"10.1093/sw/46.3.229","title":"Representation of Ethnic Identity in North American Social Work Literature: A Dossier of the Chinese People","year":2001,"lang":"en","type":"review","venue":"Social Work","topic":"Social Work Education and Practice","field":"Social Sciences","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Ethnic group; Sociology; Identity (music); Construct (python library); Gender studies; Social representation; Representation (politics); Social work; Politics; White (mutation); Social psychology; Identification (biology); Social identity theory; Cultural identity; Social group; Psychology; Social science; Anthropology; Political science; Negotiation; Aesthetics; Law","score_opus":0.05731588102700253,"score_gpt":0.45853758480240653,"score_spread":0.401221703775404,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2131615921","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012297195,0.9708194,0.0000043243394,0.004531475,0.0017360307,0.0011953699,0.000035382636,0.000037901467,0.009342932],"genre_scores_gemma":[0.029260295,0.9644393,0.00003686306,0.00009187972,0.002366306,0.00012213584,0.00006520469,0.000044592445,0.003573427],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.994287,0.0032110938,0.0008469754,0.0003642438,0.000879275,0.00041137636],"domain_scores_gemma":[0.9963444,0.0011076519,0.001958438,0.0002704435,0.00024576377,0.000073294475],"candidate_categories":["bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.001236408,0.0002913151,0.001468942,0.00016907859,0.0016505651,0.00011454167,0.0008266144,0.00041305428,0.00045813055],"category_scores_gemma":[0.0018253505,0.00024014465,0.0009574282,0.15870142,0.00077442825,0.00031342416,0.00012476211,0.0009591394,0.00003200254],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000029831075,0.00017074287,0.026693879,0.00045391326,0.00010467949,0.000001218607,0.09520609,2.2034216e-7,1.3017354e-8,0.003145606,0.0026235546,0.8715702],"study_design_scores_gemma":[0.00012579428,0.000005840171,0.5707884,0.0005159315,0.00027013267,2.666104e-7,0.0062670927,1.4634292e-8,1.9340641e-8,0.00065775635,0.42111266,0.00025608626],"about_ca_topic_score_codex":0.0016147462,"about_ca_topic_score_gemma":0.025428068,"teacher_disagreement_score":0.87131417,"about_ca_system_score_codex":0.0003612503,"about_ca_system_score_gemma":0.00069449533,"threshold_uncertainty_score":0.99964917},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2132615533","doi":"10.1002/asi.20349","title":"The place of serials in referencing practices: Comparing natural sciences and engineering with social sciences and humanities","year":2006,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":294,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"Natural science; Citation index; Citation; Science Citation Index; Social science; Index (typography); The arts; Digital humanities; Bibliometrics; Citation analysis; Social Sciences Citation Index; Library science; Sociology; Humanities; Computer science; Political science; Epistemology; Philosophy; World Wide Web","score_opus":0.2802135795098643,"score_gpt":0.4847581285676076,"score_spread":0.20454454905774327,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2132615533","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9953043,0.00040042875,0.00016537648,0.0037394068,0.000073843745,0.00011358935,0.000001856142,0.0000039871175,0.00019719715],"genre_scores_gemma":[0.9968065,0.0001924588,0.0028888516,0.000082643375,0.000017430355,0.000002274214,3.497029e-8,0.0000011737488,0.000008609317],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9966777,0.000027054246,0.0005481834,0.00013526046,0.0023324802,0.00027934264],"domain_scores_gemma":[0.99428993,0.0019609788,0.0019771538,0.00009659831,0.001650492,0.000024855119],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["sts"],"category_scores_codex":[0.021178816,0.00007076069,0.00023232581,0.005429512,0.0013895062,0.0014790497,0.0012060286,0.00002790762,2.3772908e-7],"category_scores_gemma":[0.009161904,0.00003278094,0.00003626956,0.041962195,0.007014023,0.002738237,0.0003946512,0.0001952862,7.257341e-8],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025691738,0.00009866897,0.55102897,0.000121407305,0.000080871214,0.0000012328303,0.01170226,0.0018581348,0.03741667,0.1973272,0.003427963,0.19667971],"study_design_scores_gemma":[0.002141481,0.0025014982,0.6369707,0.0001661279,0.00004123221,0.00044787012,0.2087575,0.09776947,0.008233329,0.013513653,0.028967353,0.0004897672],"about_ca_topic_score_codex":0.000101802245,"about_ca_topic_score_gemma":0.000039307764,"teacher_disagreement_score":0.19705525,"about_ca_system_score_codex":0.00005978278,"about_ca_system_score_gemma":0.00044339764,"threshold_uncertainty_score":0.99991053},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2133948460","doi":"10.5860/crl.75.5.664","title":"Exploring the Veterinary Literature: A Bibliometric Methodology for Identifying Interdisciplinary and Collaborative Publications","year":2014,"lang":"en","type":"article","venue":"College & Research Libraries","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Library of Parliament","funders":"","keywords":"Subject (documents); Publication; Accreditation; Veterinary medicine; Bibliometrics; Work (physics); Scopus; Institution; Library science; MEDLINE; Medical education; Medicine; Political science; Computer science","score_opus":0.9170081740482853,"score_gpt":0.6480557156310761,"score_spread":0.26895245841720916,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133948460","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8303041,0.022876278,0.06501492,0.05516383,0.002227151,0.003891841,0.00078902947,0.00018917008,0.019543692],"genre_scores_gemma":[0.9666497,0.0014081133,0.02386839,0.00015274697,0.0004025413,0.0012429853,0.000022688253,0.00003826681,0.0062145577],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9866926,0.0046335966,0.0007975126,0.0011520222,0.0055769635,0.0011472803],"domain_scores_gemma":[0.81915855,0.16638684,0.00030808113,0.001998062,0.01131292,0.00083555357],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.07957655,0.00022233716,0.00046869778,0.34431803,0.0029006137,0.014836229,0.0036965357,0.00013292761,0.000094908355],"category_scores_gemma":[0.19961172,0.0001345169,0.0001485618,0.74725837,0.0015108634,0.0061461036,0.0052559846,0.0007836302,0.000059670343],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046295562,0.00018078314,0.002973816,0.000100013014,0.00008846841,0.000017706532,0.0058502546,0.0000055313035,0.0028525987,0.41572785,0.2293239,0.34241614],"study_design_scores_gemma":[0.0008860344,0.0016480536,0.11274911,0.00005623795,0.000009777636,0.00004098372,0.015815046,0.0035174529,0.0023922762,0.41470206,0.44784254,0.0003404346],"about_ca_topic_score_codex":0.000008149759,"about_ca_topic_score_gemma":0.000008479704,"teacher_disagreement_score":0.40294033,"about_ca_system_score_codex":0.000079723104,"about_ca_system_score_gemma":0.00051383,"threshold_uncertainty_score":0.99839747},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2134386090","doi":"10.1016/j.joi.2010.04.001","title":"The development of an AI journal ranking based on the revealed preference approach","year":2010,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":56,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lakehead University","funders":"","keywords":"Ranking (information retrieval); Journal ranking; Impact factor; Citation; Preference; Computer science; Index (typography); Information retrieval; Field (mathematics); Citation analysis; Domain (mathematical analysis); Library science; Data science; Operations research; Statistics; World Wide Web; Mathematics; Political science","score_opus":0.578649466269294,"score_gpt":0.5307756849711357,"score_spread":0.047873781298158335,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2134386090","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9258365,0.0001783194,0.05949329,0.0012445912,0.0016322633,0.00025201324,0.0000045456936,0.000004447551,0.011354001],"genre_scores_gemma":[0.9684832,0.00007673225,0.030764908,0.00027567046,0.00017870929,0.0000019728832,3.8600305e-7,0.000008352406,0.00021008316],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.98194844,0.00029310005,0.002368455,0.00018161055,0.014692408,0.00051597465],"domain_scores_gemma":[0.97017926,0.014804917,0.0027784656,0.0008279893,0.010904237,0.00050510635],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.10649229,0.00016240048,0.00036751365,0.04023067,0.0010362114,0.003272895,0.005699869,0.00013954286,0.00012183549],"category_scores_gemma":[0.13048086,0.0000635925,0.0002634895,0.112865664,0.00022542883,0.0010144822,0.000268404,0.00200126,0.000016381568],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024905172,0.0003844371,0.024839312,0.000008932821,0.000048664937,0.00001010842,0.00070714985,0.0008745607,0.00063838705,0.001544114,0.007483589,0.9632117],"study_design_scores_gemma":[0.00437344,0.0028759788,0.57301503,0.00011713478,0.00004966682,0.0005630559,0.005224383,0.13341539,0.010527979,0.013850848,0.25525907,0.0007280212],"about_ca_topic_score_codex":0.0000017913493,"about_ca_topic_score_gemma":0.0000060879393,"teacher_disagreement_score":0.96248364,"about_ca_system_score_codex":0.000115140734,"about_ca_system_score_gemma":0.0020007158,"threshold_uncertainty_score":0.99967974},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2144516260","doi":"10.1002/asi.21493","title":"Determining the impact factors of secondary journals: A retrospective cohort study","year":2011,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"Health Sciences Research and Education","field":"Health Professions","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Journal club; Relevance (law); Retrospective cohort study; Bibliometrics; Impact factor; Medicine; MEDLINE; Inclusion (mineral); Family medicine; Library science; Psychology; Medical education; Computer science; Political science; Internal medicine","score_opus":0.10672512031900643,"score_gpt":0.48016782519942625,"score_spread":0.3734427048804198,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2144516260","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9975522,0.000013208384,0.00013785978,0.0010078107,0.00023922003,0.00063177163,0.0000072755843,0.000005867657,0.00040477884],"genre_scores_gemma":[0.999141,0.00006519531,0.00042222382,0.000313836,0.00003139208,0.000016538934,1.2850506e-7,0.0000019491258,0.000007741346],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9982655,0.00006880218,0.0006292799,0.00006464597,0.00064368267,0.0003280778],"domain_scores_gemma":[0.9950573,0.00023503593,0.00195125,0.0001923671,0.0024691143,0.0000949499],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.007308651,0.00006417409,0.00023330543,0.0003007662,0.0013994101,0.000022931545,0.00072280073,0.000039903116,0.00001833118],"category_scores_gemma":[0.0026987686,0.00002846764,0.00011534449,0.0024379662,0.0017706627,0.0013290454,0.00014096407,0.0005902554,0.0000010042359],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012012195,0.000026049325,0.96969765,0.000007440634,0.000021582167,1.3408381e-8,0.021966176,4.0941467e-7,0.000069171525,0.000053488675,0.0021044556,0.0060415687],"study_design_scores_gemma":[0.00019794893,0.00068743713,0.8542504,0.000016501628,0.000009639101,0.0000038911226,0.14383525,0.000092403934,0.000077412944,0.00042283494,0.00037930012,0.00002702317],"about_ca_topic_score_codex":0.00016635172,"about_ca_topic_score_gemma":0.0000073653546,"teacher_disagreement_score":0.12186908,"about_ca_system_score_codex":0.0002897748,"about_ca_system_score_gemma":0.0022463317,"threshold_uncertainty_score":0.99990064},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2146918704","doi":"10.1016/j.respol.2008.11.011","title":"Research collaboration and research output: A longitudinal study of 65 biomedical scientists in a New Zealand university","year":2009,"lang":"en","type":"article","venue":"Research Policy","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":272,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Universiteit van Tilburg; University of Auckland","keywords":"Causal inference; Coding (social sciences); Longitudinal study; Longitudinal data; Sociology; Psychology; Social science; Medicine; Econometrics; Economics; Demography","score_opus":0.7862779829075962,"score_gpt":0.7031571443127773,"score_spread":0.08312083859481889,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2146918704","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97924984,0.00029267286,0.0000349394,0.013537649,0.00006502938,0.001311289,0.0000355659,0.000012626308,0.00546038],"genre_scores_gemma":[0.9807648,0.00028891306,0.000184821,0.000012353078,0.0002277506,0.0000047041667,0.0000033191554,0.00001012233,0.018503208],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9454972,0.008164673,0.000955415,0.0017193161,0.0409476,0.0027157925],"domain_scores_gemma":[0.95687574,0.017611917,0.00015441768,0.0017765886,0.021130221,0.0024511102],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.1844882,0.00016585602,0.0004970072,0.24216838,0.0011096859,0.0025881792,0.0036942936,0.00024903848,0.00015074894],"category_scores_gemma":[0.12709504,0.00013424725,0.000059496262,0.5092384,0.0020722114,0.0009045127,0.0025584958,0.0020824957,0.00016818827],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011574801,0.005328098,0.4738782,0.000026710066,0.000030411578,0.00067785766,0.013154518,0.000018370016,0.0020644462,0.020001171,0.27221662,0.2114461],"study_design_scores_gemma":[0.0028653683,0.0047742,0.88778496,0.00004687319,0.0000015313855,0.000011095576,0.014680718,0.00090681104,0.00013274324,0.02636861,0.06226656,0.00016053507],"about_ca_topic_score_codex":0.033150382,"about_ca_topic_score_gemma":0.0053169094,"teacher_disagreement_score":0.41390675,"about_ca_system_score_codex":0.001143691,"about_ca_system_score_gemma":0.0077477936,"threshold_uncertainty_score":0.99844724},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2150354467","doi":"10.6000/1929-7092.2015.04.11","title":"Bibliometric Rankings of Journals Based on the Thomson Reuters Citations Database","year":2015,"lang":"en","type":"article","venue":"Journal of Reviews on Global Economics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Impact factor; Database; Web of science; Ranking (information retrieval); Computer science; Information retrieval; Library science; Political science; MEDLINE","score_opus":0.7579617406582054,"score_gpt":0.587304561522659,"score_spread":0.17065717913554646,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2150354467","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8594081,0.05416258,0.023516273,0.03249533,0.004650164,0.0012890182,0.0005821538,0.000011286117,0.023885136],"genre_scores_gemma":[0.91414994,0.060123675,0.017848888,0.0071147406,0.0005164305,0.000008936981,0.000004773501,0.000026386735,0.0002062124],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99169403,0.00083003624,0.002581735,0.00033167572,0.004181929,0.00038059757],"domain_scores_gemma":[0.9834929,0.006601452,0.003856671,0.0010485341,0.0042887283,0.000711719],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.08233699,0.00019770644,0.0008721403,0.18749063,0.00013228925,0.0013750921,0.003328664,0.000086290296,0.00024836953],"category_scores_gemma":[0.10689491,0.000103220955,0.0006146646,0.41280785,0.00014573192,0.0008320256,0.00020081257,0.00044408874,0.00022576828],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032667932,0.000512943,0.036628902,0.0000215661,0.00006565521,0.000020696214,0.00007043777,0.0055549163,0.000029290642,0.00520004,0.47417283,0.47739604],"study_design_scores_gemma":[0.0022710296,0.0019364402,0.032812618,0.00042989093,0.00006401827,0.00008296535,0.0003947386,0.016308296,0.0002479344,0.016602715,0.92851514,0.0003342104],"about_ca_topic_score_codex":0.000012947197,"about_ca_topic_score_gemma":0.0000038409003,"teacher_disagreement_score":0.47706184,"about_ca_system_score_codex":0.000455901,"about_ca_system_score_gemma":0.000682096,"threshold_uncertainty_score":0.99966156},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2153046986","doi":"10.1007/s11192-012-0891-6","title":"Discovering and assessing fields of expertise in nanomedicine: a patent co-citation network perspective","year":2012,"lang":"en","type":"article","venue":"Scientometrics","topic":"Intellectual Property and Patents","field":"Business, Management and Accounting","cited_by":33,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Quality (philosophy); Patent analysis; Modularity (biology); Data science; Computer science; Patent office; Perspective (graphical); Patent visualisation; Intellectual property; Simple (philosophy); Business; Citation; Patent portfolio; Knowledge management; Industrial organization; Engineering; Artificial intelligence; World Wide Web","score_opus":0.239983206839301,"score_gpt":0.3122728714126446,"score_spread":0.0722896645733436,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2153046986","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9834958,0.0012501895,0.002587848,0.00009591523,0.00060946686,0.0001259596,4.471991e-7,0.00001362792,0.011820765],"genre_scores_gemma":[0.9991127,0.000026361133,0.00013957769,0.00027474295,0.00035343601,0.00000418554,0.0000038384765,0.0000069661073,0.000078179495],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9990631,0.0000113728875,0.00018114902,0.00013910854,0.00032241963,0.00028289633],"domain_scores_gemma":[0.99959785,0.00006835815,0.00010456601,0.00007591235,0.00013579031,0.000017520535],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00089223677,0.00008142502,0.00012794982,0.0013258187,0.00009224899,0.00012582485,0.000085702704,0.00004400291,0.000053479864],"category_scores_gemma":[0.000804904,0.00006433035,0.00002348443,0.0045457445,0.000060949587,0.0012773375,0.00007621756,0.00008983237,0.000014106032],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000056873418,0.00023755348,0.9558433,0.00021608753,0.00001809571,0.0000039338606,0.00883526,0.0006738151,0.0009906695,0.0051672016,0.001340813,0.02661639],"study_design_scores_gemma":[0.0017392021,0.00008383198,0.92582256,0.0005105312,0.000053165335,0.00000431235,0.013901752,0.05158628,0.000816323,0.0015995807,0.003302328,0.00058015535],"about_ca_topic_score_codex":0.0006089257,"about_ca_topic_score_gemma":0.000018425737,"teacher_disagreement_score":0.050912466,"about_ca_system_score_codex":0.00006443504,"about_ca_system_score_gemma":0.000010175337,"threshold_uncertainty_score":0.26233155},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2156897283","doi":"10.1002/asi.22695","title":"Author name disambiguation: What difference does it make in author‐based citation analysis?","year":2012,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":134,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Workaround; Citation; Computer science; Ambiguity; Ranking (information retrieval); Field (mathematics); Information retrieval; Citation analysis; Journal ranking; Publication; Data science; World Wide Web; Political science; Mathematics","score_opus":0.09078199573774924,"score_gpt":0.4091928979380296,"score_spread":0.31841090220028034,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2156897283","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8012107,0.000054753753,0.10133003,0.09650245,0.0005437273,0.00025599648,0.000017425044,0.000015889886,0.00006904471],"genre_scores_gemma":[0.99050653,0.000042334537,0.006253454,0.0030643258,0.000024249992,0.000012037797,0.00000227845,0.0000015927948,0.00009318952],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.997117,0.00006428365,0.0008877515,0.00012735183,0.0015257964,0.0002778135],"domain_scores_gemma":[0.99633515,0.0004701557,0.0017376984,0.0003898343,0.0009873743,0.000079803656],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012186533,0.00009163265,0.00029784805,0.0012013484,0.00034191468,0.0006052165,0.0011482148,0.00004997226,0.000008541886],"category_scores_gemma":[0.0042016404,0.00004900637,0.00021719353,0.009250639,0.0014194024,0.0054898052,0.00021739994,0.00017378166,0.0000048996176],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000051615858,0.00011203617,0.04618249,0.000019187704,0.00011431309,1.2933687e-7,0.011564542,0.00044732986,0.0005035694,0.031289794,0.004732374,0.9049826],"study_design_scores_gemma":[0.0010772519,0.0003516331,0.39763194,0.00007015421,0.00030202253,0.000013529035,0.33107638,0.04945956,0.001191119,0.035377383,0.18306804,0.0003809643],"about_ca_topic_score_codex":0.000011011733,"about_ca_topic_score_gemma":0.000025999654,"teacher_disagreement_score":0.90460163,"about_ca_system_score_codex":0.0001254322,"about_ca_system_score_gemma":0.0001692373,"threshold_uncertainty_score":0.58361167},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2165099835","doi":"10.1111/j.1741-6787.2010.00197.x","title":"Mapping the Knowledge Utilization Field in Nursing from 1945 to 2004: A Bibliometric Analysis","year":2010,"lang":"en","type":"article","venue":"Worldviews on Evidence-Based Nursing","topic":"Health Sciences Research and Education","field":"Health Professions","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; University of Alberta","funders":"Canadian Institutes of Health Research","keywords":"Field (mathematics); Sociology of scientific knowledge; Discipline; Bibliometrics; Data science; Nursing literature; Order (exchange); Product (mathematics); Knowledge management; Computer science; Sociology; Library science; Social science; Medicine; Business; Alternative medicine","score_opus":0.32830527708112234,"score_gpt":0.5435531604224096,"score_spread":0.2152478833412873,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2165099835","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.88861793,0.0046446654,0.018483663,0.07353909,0.0045439843,0.002544234,0.000010416702,0.0001039308,0.007512069],"genre_scores_gemma":[0.9913898,0.00019685138,0.0017623026,0.0050984332,0.00080561254,0.00027379268,0.000021646227,0.000019632575,0.00043189156],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99495405,0.0015418368,0.00093533186,0.0006691512,0.00076721987,0.0011323878],"domain_scores_gemma":[0.99029076,0.007687679,0.00028602083,0.00084641116,0.000347952,0.0005411768],"candidate_categories":["bibliometrics","sts","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0055571166,0.00022696047,0.0004225983,0.04637442,0.0013604909,0.000112623515,0.0006793943,0.00020082999,0.0012124416],"category_scores_gemma":[0.0071229483,0.00016826011,0.00013731036,0.18945083,0.00013805255,0.00043160815,0.00002906465,0.0015275958,0.000657892],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001558309,0.000377934,0.14329502,0.00008357141,0.000013857476,0.0000017011818,0.01629615,0.00040520116,0.0008605816,0.00014750748,0.04220328,0.7961593],"study_design_scores_gemma":[0.0004429992,0.00021021992,0.92751414,0.01272078,0.00006627972,2.7000397e-7,0.004950746,0.021622898,0.00029221168,0.00083362724,0.031001339,0.00034450705],"about_ca_topic_score_codex":0.00198668,"about_ca_topic_score_gemma":0.0053065033,"teacher_disagreement_score":0.7958149,"about_ca_system_score_codex":0.00084216543,"about_ca_system_score_gemma":0.0015127696,"threshold_uncertainty_score":0.9999396},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2196064469","doi":"10.1097/j.pain.0000000000000403","title":"A comprehensive categorical and bibliometric analysis of published research articles on pediatric pain from 1975 to 2010","year":2015,"lang":"en","type":"review","venue":"Pain","topic":"Pediatric Pain Management Techniques","field":"Medicine","cited_by":61,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"Canadian Institutes of Health Research","keywords":"Categorical variable; Bibliometrics; Medicine; Acute pain; Psychology; MEDLINE; Chronic pain; Physical therapy; Library science; Political science","score_opus":0.19141654250498555,"score_gpt":0.43622551482315564,"score_spread":0.2448089723181701,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2196064469","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00079166563,0.9921479,0.00394272,0.00025791329,0.00006030011,0.0021065685,0.00009708745,0.00016932457,0.00042647906],"genre_scores_gemma":[0.001301545,0.994889,0.0019048332,0.00029847084,0.0004978751,0.00029788192,0.00044956725,0.00006249865,0.00029829887],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9882168,0.0072289323,0.0010755919,0.0009974301,0.001851665,0.0006295351],"domain_scores_gemma":[0.9837406,0.013217176,0.0003797344,0.0012178752,0.00080910267,0.0006354747],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.03216011,0.00044211277,0.0030460681,0.21624866,0.000059848433,0.00012134196,0.00051918963,0.0004360195,0.00012167843],"category_scores_gemma":[0.018161187,0.00035362734,0.0004785488,0.29105997,0.0000991808,0.00008688017,0.00045265313,0.00093914795,0.00006443751],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000035226723,0.00026311222,0.0063434527,0.0044677905,0.00095997757,0.000068520436,0.000040774656,5.703317e-7,5.992548e-7,0.000020840635,0.21072432,0.7770748],"study_design_scores_gemma":[0.00033022583,0.0009897229,0.00945062,0.00063423184,0.01048916,0.0000013452538,0.000065141154,0.0006679421,3.5917068e-7,0.000441957,0.976491,0.00043830642],"about_ca_topic_score_codex":0.0009932995,"about_ca_topic_score_gemma":0.000016352811,"teacher_disagreement_score":0.7766365,"about_ca_system_score_codex":0.00029461377,"about_ca_system_score_gemma":0.00031786127,"threshold_uncertainty_score":0.9998916},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2212616545","doi":"10.1371/journal.pone.0139005","title":"A Comprehensive Analysis of Authorship in Radiology Journals","year":2015,"lang":"en","type":"article","venue":"PLoS ONE","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University; Ottawa Hospital; University of Ottawa","funders":"University of Ottawa","keywords":"Medical journal; Categorization; Medicine; MEDLINE; Bibliometrics; Library science; Univariate; Meta-analysis; Impact factor; Computer science; Family medicine; Statistics; Pathology; Political science; Mathematics; Law; Multivariate statistics","score_opus":0.9232797035653724,"score_gpt":0.610283124809877,"score_spread":0.31299657875549536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2212616545","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9943604,0.0021846364,0.00038144967,0.00070323807,0.00005029391,0.00010215735,0.00001842112,0.0000053318117,0.0021941077],"genre_scores_gemma":[0.997625,0.0001674197,0.0016236561,0.000069331574,0.000024976373,0.000003651631,0.0000022415445,0.0000034604489,0.00048025989],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9907665,0.0006487393,0.0007932728,0.00036458226,0.0070646913,0.0003621814],"domain_scores_gemma":[0.9898177,0.004382423,0.00029377174,0.00051623565,0.0045558475,0.00043403124],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.020390214,0.000076236276,0.0007186024,0.13369334,0.00003021507,0.00029178173,0.0013283716,0.000108912674,0.0005154668],"category_scores_gemma":[0.05405388,0.000054132837,0.00013755998,0.33888322,0.00013048129,0.00020548025,0.00030116155,0.00024662432,0.00011321284],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004639382,0.00088999124,0.9857949,0.000003871991,0.00048709626,0.000019301142,0.0005155759,0.00034858813,0.005421778,0.00022816277,0.0010742544,0.005170062],"study_design_scores_gemma":[0.0005693833,0.00023048329,0.9527949,0.000014044908,0.00014382819,0.0000017469519,0.0005566769,0.031088512,0.0022514418,0.011844091,0.00039308518,0.00011177279],"about_ca_topic_score_codex":0.000117127696,"about_ca_topic_score_gemma":0.00004228628,"teacher_disagreement_score":0.20518988,"about_ca_system_score_codex":0.00007868741,"about_ca_system_score_gemma":0.00014398563,"threshold_uncertainty_score":0.9539142},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2226524777","doi":"10.1007/s11192-015-1825-x","title":"How to boost scientific production? A statistical analysis of research funding and other influencing factors","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":137,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Quality (philosophy); Work (physics); Production (economics); Relation (database); Impact factor; Statistical analysis; Business; Marketing; Political science; Economics; Computer science; Statistics; Mathematics; Engineering; Microeconomics","score_opus":0.7558819682205911,"score_gpt":0.6277056305859157,"score_spread":0.1281763376346754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2226524777","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97538346,0.0002480292,0.020243814,0.001788594,0.0009827956,0.00041431043,0.00033213166,0.000027903794,0.0005789458],"genre_scores_gemma":[0.9891614,0.000017919803,0.0023091617,0.000027378368,0.00005451293,0.0000104403225,0.0000021506107,0.000014889618,0.008402177],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.96931726,0.0005634705,0.00087940705,0.0020031785,0.02569866,0.001538015],"domain_scores_gemma":[0.96522546,0.01815615,0.00035888527,0.0017940674,0.012726195,0.0017392618],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.09974855,0.00020882227,0.0006025062,0.4767476,0.0009271588,0.0074615646,0.0025741048,0.00012406956,0.00034432148],"category_scores_gemma":[0.3991574,0.000117539435,0.00016430025,0.7982325,0.0018128299,0.0011355258,0.001701463,0.00026152498,0.000098767356],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016677262,0.00010814637,0.9263775,0.000008778137,0.000083605184,0.0000029118844,0.0007880807,0.000027860131,0.019895807,0.0045163003,0.0046243216,0.043550026],"study_design_scores_gemma":[0.00024930958,0.0002517567,0.9469564,0.000026717704,0.000057071327,0.0000018969605,0.0022264076,0.0006100925,0.009076896,0.0023162907,0.0379461,0.00028109134],"about_ca_topic_score_codex":0.00012039453,"about_ca_topic_score_gemma":0.00009319081,"teacher_disagreement_score":0.3214849,"about_ca_system_score_codex":0.00043435112,"about_ca_system_score_gemma":0.00036772818,"threshold_uncertainty_score":0.9935688},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2231640728","doi":"10.4103/2320-0057.135407","title":"A bibliometric analysis of a national Journal: The case of the Turkish Journal of Psychology","year":2013,"lang":"en","type":"article","venue":"Journal of Scientometric Research","topic":"Families in Therapy and Culture","field":"Psychology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Turkish; Impact factor; Social science; Publishing; Citation; Psychology; Library science; Science Citation Index; Multidisciplinary approach; Web of science; Sociology; Political science; MEDLINE; Law; Philosophy; Computer science","score_opus":0.19639891335076606,"score_gpt":0.5319963834047735,"score_spread":0.3355974700540074,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2231640728","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9805955,0.010656296,0.00016635563,0.0020189728,0.0017153128,0.00019306556,0.000014319592,0.0000013365585,0.0046388116],"genre_scores_gemma":[0.9969828,0.0012518462,0.00028348764,0.000200797,0.0003811683,0.0000030903614,2.3792543e-7,0.000012464001,0.0008841221],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99102277,0.0019796223,0.0019351349,0.00018407902,0.0043373136,0.0005410556],"domain_scores_gemma":[0.9786206,0.0028231586,0.0028068563,0.0005454422,0.014937451,0.0002665037],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.026628241,0.00015138197,0.0006881504,0.23290534,0.0003651261,0.00020309747,0.0029015823,0.00020842635,0.0032672652],"category_scores_gemma":[0.0077602407,0.00007131619,0.0009989975,0.5180313,0.0012809938,0.00044660707,0.00019030324,0.002161794,0.000013946663],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010112079,0.003710577,0.121425256,0.00007333023,0.013199516,0.0009473486,0.015052831,0.0011927683,0.01562858,0.0022208672,0.61224896,0.21328877],"study_design_scores_gemma":[0.0036045427,0.0028020611,0.9238925,0.00010860291,0.0006694091,0.02795194,0.0151168,0.0002205685,0.0006895011,0.0066686664,0.018076267,0.00019914747],"about_ca_topic_score_codex":0.00009034755,"about_ca_topic_score_gemma":0.000007962477,"teacher_disagreement_score":0.8024672,"about_ca_system_score_codex":0.00018791796,"about_ca_system_score_gemma":0.00047978509,"threshold_uncertainty_score":0.9976439},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2236608270","doi":"","title":"Collaborative Citation Analysis and Accuracy Degree of Vancouver Style Application in References Of The Scientific Journal of Birjand University Of Medical Sciences: Short Communication","year":2015,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Ophthalmology and Visual Health Research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Degree (music); Style (visual arts); Citation analysis; Computer science; Library science; Information retrieval; Geography; Archaeology; Physics","score_opus":0.6020774233915351,"score_gpt":0.6558915468924948,"score_spread":0.05381412350095971,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2236608270","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9946016,0.0039198766,0.00031574068,0.00025188221,0.0000780272,0.00033486547,0.000015981772,0.0000010304365,0.00048096213],"genre_scores_gemma":[0.9965696,0.0030433373,0.00033295623,0.00000793476,0.00000903552,0.0000013651635,0.000003978044,0.0000032799385,0.000028535238],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9969395,0.00074670627,0.00072835706,0.00015778716,0.0013064166,0.00012123637],"domain_scores_gemma":[0.9956974,0.0007588301,0.001265173,0.00029083938,0.001830352,0.00015742183],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008373133,0.0000791701,0.0005687657,0.0012328541,0.00015914612,0.00004522532,0.0013204247,0.00010602069,0.00028020958],"category_scores_gemma":[0.0015018601,0.000056497982,0.00006922076,0.004445154,0.0013464693,0.0007562189,0.0003732398,0.00031965642,1.7136256e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005070274,0.00032416134,0.9861505,0.00013534218,0.00014426275,0.0000025529882,0.0014601528,0.00006768857,0.0036714023,0.00008873606,0.00072866544,0.006719487],"study_design_scores_gemma":[0.0007457146,0.000093538874,0.986908,0.00048011015,0.00016501892,0.000006155864,0.00402742,0.00077531,0.0055515277,0.0011028715,0.000095025825,0.00004930498],"about_ca_topic_score_codex":0.0017021982,"about_ca_topic_score_gemma":0.0021665907,"teacher_disagreement_score":0.0068712733,"about_ca_system_score_codex":0.00007041846,"about_ca_system_score_gemma":0.001515789,"threshold_uncertainty_score":0.4961122},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2243575632","doi":"","title":"Bibliometric factor maps for knowledge discovery in digital libraries","year":2009,"lang":"en","type":"article","venue":"International Conference on Electronic Publishing","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Digital library; Visualization; Information retrieval; Exploratory search; Field (mathematics); Citation; Domain (mathematical analysis); Co-citation; Data science; Information visualization; World Wide Web; Data mining","score_opus":0.036269693073406224,"score_gpt":0.3078835157846224,"score_spread":0.2716138227112162,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2243575632","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4277057,0.00039173334,0.14125282,0.0073760427,0.0004621404,0.0007799248,0.0005848199,0.00033781354,0.421109],"genre_scores_gemma":[0.9975985,0.000010407845,0.00013676165,0.00008590158,0.00039398076,0.000041809606,0.0004891012,0.000012602937,0.0012309351],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9986414,0.000016239785,0.00029903324,0.00034199396,0.00025550407,0.00044583066],"domain_scores_gemma":[0.9992163,0.00021729045,0.0001329203,0.00017655276,0.00020527611,0.00005164565],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.00016788059,0.00017733107,0.00020398793,0.007345821,0.000057840432,0.0048644757,0.0007546124,0.000048058635,0.00021978225],"category_scores_gemma":[0.00012711392,0.00017233557,0.00014050303,0.0061475295,0.00002293452,0.006184179,0.00007373243,0.00036190887,0.000007918387],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000032779477,0.00017189888,0.0060174637,0.0000011925245,0.00004594618,3.0737561e-7,0.000055842924,0.000011984832,0.00009808172,0.8524756,0.0027862692,0.13830264],"study_design_scores_gemma":[0.0006735385,0.00022639723,0.009358704,0.00006972092,0.000008439513,6.198551e-7,0.0001229226,0.009444387,0.0007368222,0.955636,0.023366712,0.00035572227],"about_ca_topic_score_codex":0.00003967104,"about_ca_topic_score_gemma":0.00002493271,"teacher_disagreement_score":0.56989276,"about_ca_system_score_codex":0.0001819941,"about_ca_system_score_gemma":0.0002869359,"threshold_uncertainty_score":0.99616855},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2259068601","doi":"10.1016/j.ijsu.2016.02.029","title":"Worldwide productivity in the hand and wrist literature: A bibliometric analysis of four highly cited subspecialty journals","year":2016,"lang":"en","type":"article","venue":"International Journal of Surgery","topic":"Orthopedic Surgery and Rehabilitation","field":"Medicine","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Subspecialty; Medicine; Productivity; Bibliometrics; China; Demography; Web of science; Population; Scopus; Geography; Family medicine; MEDLINE; Library science; Environmental health; Economic growth; Political science; Meta-analysis; Internal medicine","score_opus":0.04457388255179528,"score_gpt":0.32011566907472644,"score_spread":0.27554178652293115,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2259068601","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98434055,0.0028658528,0.00023803087,0.011739598,0.0006073935,0.00007321527,0.0000189893,0.0000026888474,0.00011369909],"genre_scores_gemma":[0.9972208,0.0017372725,0.0000687439,0.00013713312,0.0006271502,0.0000019883614,0.000004607314,0.0000066314374,0.00019565618],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9974911,0.00027432796,0.0009364538,0.00014030821,0.0010274308,0.00013032818],"domain_scores_gemma":[0.99441624,0.003192162,0.0007786922,0.00013656041,0.0013920576,0.00008427519],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.004713234,0.00010665445,0.00052841776,0.05845687,0.00003117734,0.00008041059,0.0001226053,0.00006290528,0.00014610609],"category_scores_gemma":[0.0048460364,0.000053591903,0.00039857396,0.03745103,0.00015427628,0.0004575668,0.000020995034,0.00021395339,0.000001185295],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00059853576,0.0002658834,0.8545413,0.000020107796,0.0009494771,0.00033093177,0.00031655966,0.000004823305,0.0064649554,0.000025528288,0.0006308673,0.13585104],"study_design_scores_gemma":[0.0005168587,0.000052797106,0.9941562,0.0005773334,0.00028029265,0.0005035974,0.0000463383,0.000009090799,0.0008571733,0.00020093845,0.0027387291,0.00006068246],"about_ca_topic_score_codex":0.000012595799,"about_ca_topic_score_gemma":0.000011370359,"teacher_disagreement_score":0.13961488,"about_ca_system_score_codex":0.000067656794,"about_ca_system_score_gemma":0.0001484852,"threshold_uncertainty_score":0.98300844},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2260143504","doi":"10.1016/j.wneu.2016.01.087","title":"Social Media Metrics and Bibliometric Profiles of Neurosurgical Departments and Journals: Is There a Relationship?","year":2016,"lang":"en","type":"article","venue":"World Neurosurgery","topic":"Social Media in Health Education","field":"Social Sciences","cited_by":68,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Michael's Hospital; Toronto Western Hospital; University of Toronto","funders":"","keywords":"Social media; Accreditation; Medicine; Bibliometrics; Medical education; Library science; World Wide Web; Computer science","score_opus":0.22885418847871541,"score_gpt":0.414867894221078,"score_spread":0.1860137057423626,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2260143504","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9824025,0.002080314,0.000005151064,0.012450174,0.0020044926,0.00038034178,0.000020822728,0.000043038595,0.0006131802],"genre_scores_gemma":[0.99524623,0.0035954686,0.000049001217,0.00022857648,0.0006401542,0.0000811298,8.770802e-7,0.000019135676,0.00013943158],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9974291,0.0006821209,0.0004951618,0.00027875925,0.0007548959,0.00035996482],"domain_scores_gemma":[0.97767484,0.02139642,0.00038210826,0.00010593031,0.00016931751,0.00027140073],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0016886799,0.00012218113,0.00028317882,0.009312577,0.0005436088,0.000085611326,0.00011791448,0.000112625494,0.00013439516],"category_scores_gemma":[0.02116254,0.00009759446,0.000064545624,0.027121196,0.0005078531,0.00033182936,0.000049946735,0.0001710369,0.0000073586684],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001617298,0.00005133623,0.95039356,0.000022793372,0.000005277935,0.0000040725226,0.0025451349,7.4176936e-9,0.00007332437,0.0018144732,0.003165202,0.04190862],"study_design_scores_gemma":[0.00020734643,0.000013187472,0.9660642,0.000033738765,0.000018825313,0.000002111896,0.00017678358,7.4354796e-7,0.00011948759,0.0017816861,0.03147894,0.000102917285],"about_ca_topic_score_codex":0.00007729436,"about_ca_topic_score_gemma":0.000108904394,"teacher_disagreement_score":0.041805707,"about_ca_system_score_codex":0.00005894629,"about_ca_system_score_gemma":0.0002088982,"threshold_uncertainty_score":0.9935578},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2264463174","doi":"10.1007/s11192-016-1846-0","title":"Factors influencing research collaboration in LIS schools in South Africa","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":31,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"University of South Africa","keywords":"Bureaucracy; Library science; Productivity; Medical education; Population; Public relations; Psychology; Sociology; Political science; Medicine; Economic growth; Computer science","score_opus":0.7350202013577418,"score_gpt":0.6208490315127927,"score_spread":0.11417116984494913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2264463174","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9906318,0.00073485594,0.0006557208,0.0009051833,0.0007326345,0.00056053215,0.0000675769,0.000033469005,0.0056782207],"genre_scores_gemma":[0.99639803,0.000086808315,0.00041810327,0.000027011158,0.000057018027,0.00003033521,0.0000016917224,0.000018713892,0.0029622575],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9657006,0.0011704097,0.00150628,0.0017150487,0.027746128,0.002161516],"domain_scores_gemma":[0.971301,0.016063169,0.00036653606,0.001599673,0.009592143,0.001077432],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.11080867,0.0002549729,0.00051031256,0.49142045,0.0003508513,0.004645818,0.0043446654,0.0002920899,0.00040166415],"category_scores_gemma":[0.3862859,0.00014706452,0.00011687501,0.8414483,0.0003983643,0.002547789,0.0014909299,0.0007956754,0.0009874034],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021088566,0.00014015404,0.96061635,0.0000033644578,0.0000022060267,0.00001669694,0.0037783762,0.000046527886,0.0017449015,0.00061443076,0.0026039619,0.030411916],"study_design_scores_gemma":[0.0012986293,0.00023008013,0.9503058,0.0000624748,0.0000012514514,5.751936e-7,0.0120748775,0.0007106786,0.003132249,0.009093488,0.022717817,0.00037203188],"about_ca_topic_score_codex":0.00013789732,"about_ca_topic_score_gemma":0.00024373917,"teacher_disagreement_score":0.35002786,"about_ca_system_score_codex":0.0019613714,"about_ca_system_score_gemma":0.0012376616,"threshold_uncertainty_score":0.99979043},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2279214036","doi":"10.1016/j.jclinepi.2015.02.019","title":"Knowledge synthesis research: a bibliometric analysis","year":2016,"lang":"en","type":"review","venue":"Journal of Clinical Epidemiology","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":35,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada Research Chairs; University of Toronto; St. Michael's Hospital; Institute for Work & Health","funders":"Canadian Institutes of Health Research; University of Toronto","keywords":"Subject (documents); Bibliometrics; Computer science; Publishing; Medical research; Medical literature; Data science; Management science; Medicine; Library science; Pathology; Political science","score_opus":0.9890212536511169,"score_gpt":0.8813268878723645,"score_spread":0.10769436577875247,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2279214036","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":"methods","model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000036068162,0.9801172,0.0023247278,0.008906812,0.003534103,0.0008625598,0.00013950428,0.00002036928,0.0040586432],"genre_scores_gemma":[0.000029882607,0.98654175,0.0020998502,0.0034864452,0.006038951,0.00017187741,0.0000025837087,0.000048336387,0.0015803032],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.68945545,0.27410543,0.032320958,0.0008135345,0.0010103136,0.0022943136],"domain_scores_gemma":[0.09786092,0.86868733,0.026566995,0.0012045599,0.003573403,0.0021067814],"candidate_categories":["metaresearch","metaepi_broad","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","research_integrity","insufficient_payload"],"category_scores_codex":[0.533461,0.00044040458,0.019864859,0.18011564,0.00067991053,0.000010171904,0.0023554543,0.0023388388,0.004321504],"category_scores_gemma":[0.880622,0.00023600127,0.0050405697,0.20398402,0.0010023466,0.00022054269,0.00064148277,0.008160124,0.0058176126],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000020172469,0.000057309557,0.0077348156,0.0045417976,0.0010691135,0.000012133542,0.000066395936,7.7596646e-8,3.3555434e-9,0.0009074614,0.12030975,0.865281],"study_design_scores_gemma":[0.00033513026,0.0002795728,0.008851078,0.011234243,0.0017795566,0.000016787977,0.00004936541,0.0000036933884,3.6735868e-9,0.001570615,0.97569734,0.00018263546],"about_ca_topic_score_codex":0.00007774356,"about_ca_topic_score_gemma":0.000070512135,"teacher_disagreement_score":0.86509836,"about_ca_system_score_codex":0.0007257044,"about_ca_system_score_gemma":0.014751791,"threshold_uncertainty_score":0.9989563},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2299574032","doi":"10.18438/b8wd1b","title":"Faculty Decisions on Serials Subscriptions Differ Significantly from Decisions Predicted by a Bibliometric Tool","year":2016,"lang":"en","type":"article","venue":"Evidence Based Library and Information Practice","topic":"Library Collection Development and Digital Resources","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Valuation (finance); Computer science; Bibliometrics; Library science; Accounting; Economics","score_opus":0.03616895166796097,"score_gpt":0.25970761184757846,"score_spread":0.2235386601796175,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2299574032","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16212177,0.000447692,0.6932942,0.12924203,0.0009420315,0.0010960115,0.0016794518,0.0013029737,0.009873822],"genre_scores_gemma":[0.856512,0.005250419,0.056732032,0.07582539,0.00017400554,0.0002298586,0.00074558397,0.00003070868,0.004499993],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977144,0.00023789854,0.00067674625,0.00035787636,0.0007439856,0.00026911538],"domain_scores_gemma":[0.9860164,0.0127134845,0.0003758187,0.0004987138,0.00011436816,0.00028120403],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.0003418227,0.00022460202,0.00019237699,0.0048823613,0.0004244596,0.0026527075,0.00072987756,0.00011682956,0.0004562124],"category_scores_gemma":[0.010515513,0.00014811776,0.00006566276,0.015315422,0.000056615736,0.3144444,0.00029477998,0.00013670677,0.00025588291],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001022018,0.00033438802,0.003401863,0.000010872681,0.00008218098,0.000010605736,0.0003581817,0.00004053246,0.0015005708,0.08041924,0.5785516,0.33426794],"study_design_scores_gemma":[0.0009279639,0.00027313686,0.062295184,0.0003790692,0.000015201648,0.000006742112,0.00007212741,0.0016849183,0.007855311,0.001301627,0.92484295,0.0003457361],"about_ca_topic_score_codex":0.0000033809704,"about_ca_topic_score_gemma":2.2499654e-8,"teacher_disagreement_score":0.69439024,"about_ca_system_score_codex":0.000021007267,"about_ca_system_score_gemma":0.00030193292,"threshold_uncertainty_score":0.9983826},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2303156206","doi":"10.1007/s11873-013-0210-2","title":"La carrière des publications d'Ettore Majorana: une étude bibliométrique","year":2013,"lang":"fr","type":"article","venue":"Revue de Synthèse","topic":"Philosophy, Science, and History","field":"Arts and Humanities","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec","funders":"","keywords":"Humanities; MAJORANA; Philosophy; Physics","score_opus":0.029971551916118288,"score_gpt":0.23908542608626712,"score_spread":0.20911387417014884,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2303156206","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.123315066,0.03947547,0.00023057676,0.045255385,0.0020035934,0.0006724808,0.00038339343,0.00032148507,0.78834254],"genre_scores_gemma":[0.9310266,0.0015962508,0.0005050754,0.0004472232,0.0024658446,0.00017725921,0.00002247839,0.000055973636,0.063703336],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99810165,0.00020884362,0.00039546276,0.00043842036,0.00022011947,0.0006354846],"domain_scores_gemma":[0.99823916,0.00022765322,0.00015899066,0.0005819339,0.0003850812,0.00040719385],"candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0004579638,0.00028676132,0.00032551558,0.0013618177,0.001123266,0.00078380085,0.00057358394,0.00019809819,0.003646303],"category_scores_gemma":[0.00016344302,0.00030083655,0.00022154568,0.0012071736,0.0021472936,0.0012363166,0.00006301756,0.00041872452,0.0015707096],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004139524,0.00026827503,0.0011070118,0.00043856032,0.000061253675,0.000029691895,0.019565819,0.000016442747,0.00028001706,0.9060101,0.059267838,0.012950856],"study_design_scores_gemma":[0.00017012295,0.000051703573,0.008537923,0.00029549823,0.00007999068,0.00007845729,0.0019969004,0.0002963344,0.000036784266,0.02490535,0.96316266,0.00038830278],"about_ca_topic_score_codex":0.010754825,"about_ca_topic_score_gemma":0.0093855,"teacher_disagreement_score":0.9038948,"about_ca_system_score_codex":0.00034309278,"about_ca_system_score_gemma":0.00024285504,"threshold_uncertainty_score":0.9999444},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2303709641","doi":"","title":"Impact of family medicine research: Bibliometrics and beyond.","year":2016,"lang":"en","type":"article","venue":"PubMed","topic":"Primary Care and Health Outcomes","field":"Health Professions","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University; College of Family Physicians of Canada","funders":"","keywords":"Bibliometrics; Legitimacy; Work (physics); Academic medicine; Social science; Sociology; Engineering ethics; Political science; Medicine; Medical education; Library science; Engineering; Law; Computer science","score_opus":0.42785090164056583,"score_gpt":0.5475574184010801,"score_spread":0.11970651676051425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2303709641","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.82809806,0.0036299813,0.00004701594,0.037181433,0.0006368046,0.0015060593,0.00007466854,0.000067269786,0.12875868],"genre_scores_gemma":[0.9854124,0.004875031,0.000047431768,0.004096513,0.00037239635,0.0006402056,0.0000033686126,0.000020753476,0.0045318715],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9973218,0.00050859153,0.0005003591,0.00021178974,0.00055646646,0.000900993],"domain_scores_gemma":[0.9945368,0.004067606,0.00015177063,0.0003445588,0.00048639067,0.00041284613],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004710071,0.000100774356,0.00035595606,0.0051055034,0.0002664732,0.0000027174547,0.0001925124,0.00015647308,0.0002476606],"category_scores_gemma":[0.0030326403,0.00005098896,0.00004813106,0.005716208,0.00026160118,0.0001545679,0.00019920288,0.00034830216,0.00004367322],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006910728,0.00001791566,0.36793053,0.00011251964,0.000021924387,0.0000024147125,0.0002298264,6.2239915e-9,0.0001560962,0.00080815644,0.3887099,0.2419416],"study_design_scores_gemma":[0.001178173,0.00012686962,0.92718863,0.000022350743,0.000007480977,6.444269e-7,0.00013298384,1.3052814e-7,0.0000065632125,0.002153047,0.06912728,0.000055853867],"about_ca_topic_score_codex":0.00097269134,"about_ca_topic_score_gemma":0.00003774969,"teacher_disagreement_score":0.5592581,"about_ca_system_score_codex":0.00045147663,"about_ca_system_score_gemma":0.00054352195,"threshold_uncertainty_score":0.45555878},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2320350537","doi":"10.1097/ccm.0000000000001466","title":"Determinants of Citation Impact in Large Clinical Trials in Critical Care","year":2015,"lang":"en","type":"review","venue":"Critical Care Medicine","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University; St. Michael's Hospital","funders":"","keywords":"Medicine; Interquartile range; Clinical trial; Randomized controlled trial; MEDLINE; Internal medicine","score_opus":0.9465369354342983,"score_gpt":0.7688826758742692,"score_spread":0.17765425956002912,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2320350537","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00043332367,0.9936108,0.00040843332,0.00020773645,0.0014022754,0.0018772818,0.00033445793,0.000004362656,0.0017213592],"genre_scores_gemma":[0.16133283,0.83516306,0.00090364594,0.00023108431,0.0015292477,0.00028915712,0.00030315603,0.00008839619,0.00015943992],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.8228267,0.11362803,0.052921575,0.0019709351,0.00788015,0.00077261345],"domain_scores_gemma":[0.7754975,0.20404124,0.0066835373,0.0050151553,0.0076632733,0.0010992904],"candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.42663315,0.0008164316,0.041933723,0.0019422657,0.00004426395,0.00016866095,0.0019879788,0.0009155121,0.0059114387],"category_scores_gemma":[0.88068324,0.00034918427,0.0063880514,0.0030881113,0.000643635,0.00018940191,0.00023811398,0.0010976763,0.00053111254],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001672463,0.00011295911,0.015786665,0.027284764,0.000045477213,0.00020110249,0.001301921,1.868078e-7,3.9019017e-8,0.0009252406,0.006129884,0.94819504],"study_design_scores_gemma":[0.002120538,0.001377133,0.0035780258,0.051170245,0.0076731173,0.00007519451,0.025172496,0.00025937002,1.5657608e-7,0.0043546003,0.903364,0.0008551227],"about_ca_topic_score_codex":0.00005468638,"about_ca_topic_score_gemma":0.00027109435,"teacher_disagreement_score":0.9473399,"about_ca_system_score_codex":0.00030321977,"about_ca_system_score_gemma":0.0010838539,"threshold_uncertainty_score":0.999896},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2333110603","doi":"10.1177/000841740907600201","title":"Are We Ready for an Impact Factor?","year":2009,"lang":"en","type":"article","venue":"Canadian Journal of Occupational Therapy","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Medicine","score_opus":0.9241897391965286,"score_gpt":0.6855952147089429,"score_spread":0.23859452448758567,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2333110603","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9892718,0.0018513674,0.0048976964,0.0025448173,0.00043960026,0.00018360413,0.00045005803,0.0000032102703,0.0003578025],"genre_scores_gemma":[0.9978948,0.00013034191,0.0007586642,0.0006419183,0.00032332467,9.742234e-7,0.0000060081506,0.000006406861,0.00023750852],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99598837,0.00011216247,0.00057398505,0.00020646374,0.002714724,0.0004043138],"domain_scores_gemma":[0.99133277,0.0010422986,0.00063456706,0.00026588832,0.005285984,0.0014384935],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0052973675,0.000111038964,0.0002509153,0.026287185,0.0002449766,0.0012942535,0.0013498088,0.00008148016,0.00082749326],"category_scores_gemma":[0.004656937,0.00006998977,0.00022602179,0.020902228,0.000056515764,0.00082585396,0.0000051792235,0.00016316821,0.000018002691],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003315365,0.000082796265,0.3590796,5.4361885e-7,0.00003322727,0.000028002069,0.00024575973,0.00031905618,0.0004056463,0.0008876472,0.08487003,0.5537161],"study_design_scores_gemma":[0.0005591129,0.0010338355,0.8755556,0.000005069536,0.0000011352843,0.000027638958,0.00010446017,0.00042341914,0.00009587395,0.014965365,0.10713443,0.000094080686],"about_ca_topic_score_codex":0.00036979714,"about_ca_topic_score_gemma":0.0011775407,"teacher_disagreement_score":0.55362207,"about_ca_system_score_codex":0.00025248123,"about_ca_system_score_gemma":0.0023613113,"threshold_uncertainty_score":0.999909},"labels":[{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"grok","categories":["metaresearch","bibliometrics","scholarly_communication"],"domain":"evaluation","study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"opus","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2337101923","doi":"","title":"Systematic reviews on reports of hip fractures in Web of Science: a bibliometric analysis of publication activity.","year":2014,"lang":"en","type":"article","venue":"PubMed","topic":"Hip and Femur Fractures","field":"Medicine","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Web of science; Systematic review; China; Citation; Bibliometrics; Scopus; Medicine; Ranking (information retrieval); Library science; Science Citation Index; Meta-analysis; MEDLINE; Geography; Political science; Information retrieval; Computer science; Pathology; Archaeology; Law","score_opus":0.038489523980474755,"score_gpt":0.323310451873466,"score_spread":0.2848209278929913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2337101923","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99320304,0.0008259964,0.00007512677,0.00017562433,0.00005552745,0.0014311908,0.000002133703,0.000008589856,0.004222757],"genre_scores_gemma":[0.9993042,0.000113972295,0.000043909866,0.000090572794,0.000017318163,0.00035567896,0.000005230098,0.000005548621,0.00006361629],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9977461,0.0001631593,0.0009411614,0.00023489552,0.0007227614,0.00019189139],"domain_scores_gemma":[0.99707985,0.00041368103,0.0013522832,0.0006994003,0.00035060645,0.00010415422],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.006251432,0.00009712892,0.0010914496,0.07788405,0.000016067257,0.0000103754965,0.00013427615,0.00007078509,0.000027335924],"category_scores_gemma":[0.020293714,0.00006409065,0.00018851754,0.1329768,0.00016882684,0.00013209025,0.00002258411,0.00010409345,7.7548077e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025791195,0.002128198,0.6581865,0.04204197,0.0010774696,0.00000950411,0.0011975762,0.00012463736,0.019832876,0.00033157202,0.0012119883,0.2735998],"study_design_scores_gemma":[0.00018639253,0.000047678895,0.983344,0.00046574773,0.00061400473,0.0000029339371,0.000016223934,0.0003187748,0.014734592,0.000020183697,0.00020077154,0.000048662372],"about_ca_topic_score_codex":0.000068232606,"about_ca_topic_score_gemma":0.00001739502,"teacher_disagreement_score":0.32515755,"about_ca_system_score_codex":0.00005557985,"about_ca_system_score_gemma":0.000083164974,"threshold_uncertainty_score":0.9879588},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2337529989","doi":"10.1093/scipol/scv038","title":"What determines researchers’ scientific impact? A case study of Quebec researchers","year":2015,"lang":"en","type":"article","venue":"Science and Public Policy","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université de Montréal; Polytechnique Montréal","funders":"","keywords":"Regional science; Management science; Engineering ethics; Sociology; Economics; Engineering","score_opus":0.8604730270441896,"score_gpt":0.6730168236645624,"score_spread":0.1874562033796272,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2337529989","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99344045,0.0006927962,0.000029846782,0.0031158435,0.0004736256,0.00053166295,0.0000120902505,0.000019538309,0.0016841219],"genre_scores_gemma":[0.9962648,0.00005189807,0.00006729093,0.00006136299,0.0001081809,0.000019424853,6.7585444e-7,0.000009597876,0.0034167431],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.97032607,0.00083160907,0.0008000236,0.0014325241,0.024702452,0.0019073285],"domain_scores_gemma":[0.96573114,0.0030489054,0.00034343117,0.0022876398,0.023991264,0.0045976406],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.13297059,0.0002022393,0.00039233707,0.11373351,0.0012637817,0.037701312,0.004758996,0.00009351599,0.00007676125],"category_scores_gemma":[0.1986515,0.00012776756,0.000090867805,0.36774755,0.0061466503,0.012196435,0.0031157434,0.00037991029,0.00007400691],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000051733965,0.0010857288,0.1639089,0.000009018724,0.000021302196,0.00061168126,0.026057981,0.000019608928,0.0013325235,0.0013411035,0.013034413,0.792526],"study_design_scores_gemma":[0.003551703,0.003955894,0.087085515,0.000030351204,0.0000099516365,0.0014836394,0.8406503,0.013583497,0.00063736556,0.01993225,0.028318893,0.00076060224],"about_ca_topic_score_codex":0.12481403,"about_ca_topic_score_gemma":0.022350175,"teacher_disagreement_score":0.81459236,"about_ca_system_score_codex":0.00070373976,"about_ca_system_score_gemma":0.021745829,"threshold_uncertainty_score":0.99655807},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2338153978","doi":"10.82308/43227","title":"A bibliometric analysis of Quebec's PhD students' contribution to the advancement of knowledge","year":2010,"lang":"en","type":"article","venue":"Open MIND","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Socialization; Workforce; Sociology; Population; Attribution; Medical education; Pedagogy; Psychology; Library science; Social science; Political science; Computer science; Medicine; Social psychology","score_opus":0.45461326270658553,"score_gpt":0.5998687920653639,"score_spread":0.1452555293587784,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2338153978","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9892404,0.00031551195,0.0061529777,0.0005663894,0.00036944117,0.0006834754,0.00015111642,5.517858e-7,0.0025201107],"genre_scores_gemma":[0.99389464,0.000068607784,0.0012528361,0.000018542987,0.000025544392,0.000019472582,0.000008524857,0.000003870896,0.0047079613],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99103755,0.00022060233,0.0009865958,0.0005777659,0.0067976927,0.00037978808],"domain_scores_gemma":[0.988494,0.0036720226,0.00049871125,0.0012443371,0.0058019883,0.0002889559],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.032737333,0.000114570044,0.0005475592,0.6162646,0.00014200734,0.0016175049,0.006683954,0.00007451885,0.005351114],"category_scores_gemma":[0.030429792,0.000064686166,0.0002197616,0.9377577,0.0001420754,0.0004590903,0.0024792855,0.00017109633,0.0004886034],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000040096205,0.0004986383,0.1699577,9.3505184e-7,0.00018684767,8.878564e-7,0.0002562814,0.00008479324,0.0063704196,0.00028113232,0.0009717297,0.8213505],"study_design_scores_gemma":[0.00040590749,0.00014772576,0.690542,0.0000037095494,0.000100144,2.808842e-7,0.00022991405,0.0010442444,0.01206604,0.000048956015,0.2953247,0.00008638498],"about_ca_topic_score_codex":0.0011840159,"about_ca_topic_score_gemma":0.009251222,"teacher_disagreement_score":0.82126415,"about_ca_system_score_codex":0.000058373837,"about_ca_system_score_gemma":0.0002469461,"threshold_uncertainty_score":0.9994189},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W2346102634","doi":"10.3109/13561820.2016.1142430","title":"Examining semantics in interprofessional research: A bibliometric study","year":2016,"lang":"en","type":"article","venue":"Journal of Interprofessional Care","topic":"Interprofessional Education and Collaboration","field":"Health Professions","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Ontario Council of University Libraries","funders":"","keywords":"Terminology; Multidisciplinary approach; Search engine indexing; CLARITY; Bibliometrics; Computer science; Semantics (computer science); Field (mathematics); Subject (documents); Data science; Psychology; Information retrieval; Sociology; Linguistics; Library science; Social science; Mathematics","score_opus":0.21841468693692853,"score_gpt":0.560398135774358,"score_spread":0.3419834488374295,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2346102634","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9821557,0.00053211296,0.0002957658,0.003973992,0.009178757,0.0009881129,0.000021754002,0.000028731849,0.0028250993],"genre_scores_gemma":[0.9917474,0.00005048454,0.00071879994,0.00063331914,0.0011265926,0.00015421567,0.0000070287097,0.00005872238,0.0055034584],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9895887,0.0044720443,0.0024440042,0.00045244404,0.0022274007,0.00081543945],"domain_scores_gemma":[0.9883701,0.0047585797,0.0011758759,0.0004568338,0.0048974473,0.0003411254],"candidate_categories":["bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.006176673,0.00032806833,0.0007068383,0.026985582,0.0007413453,0.000030965246,0.00093506067,0.00040200585,0.0022953206],"category_scores_gemma":[0.0037179946,0.00018790204,0.00015971105,0.01906337,0.00015962533,0.00082730444,0.0006346586,0.002725796,0.0003777229],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005074313,0.0009548525,0.92662317,0.00013886177,0.00006812567,0.00015245732,0.024683345,0.0000018239573,0.0079213595,0.0008638768,0.015939593,0.017578254],"study_design_scores_gemma":[0.0040649953,0.0015600206,0.5720609,0.014027137,0.000022994662,0.000024713927,0.40424663,0.000011042082,0.00018666888,0.0014927324,0.0020043892,0.0002978366],"about_ca_topic_score_codex":0.00010617649,"about_ca_topic_score_gemma":0.0006343319,"teacher_disagreement_score":0.37956327,"about_ca_system_score_codex":0.0015965049,"about_ca_system_score_gemma":0.0059069563,"threshold_uncertainty_score":0.9997286},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2351115647","doi":"","title":"Bibliometric analysis of literatures correlated with disease of synovial membrane from 1995 to 2004","year":2006,"lang":"en","type":"article","venue":"Chongqing yixue","topic":"Traditional Chinese Medicine Studies","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Synovial membrane; MEDLINE; Osteoarthritis; English language; Medicine; Bibliometrics; Rheumatoid arthritis; Library science; Disease; Pathology; History; Alternative medicine; Political science; Computer science; Linguistics; Internal medicine; Philosophy","score_opus":0.009129332057380828,"score_gpt":0.25530924362770735,"score_spread":0.2461799115703265,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2351115647","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99413,0.00318043,0.00047046965,0.00054583495,0.00012435972,0.00023599798,0.0004053791,0.00004897128,0.0008585286],"genre_scores_gemma":[0.99779505,0.000020210802,0.0008400045,0.00013974262,0.0002638024,0.000012175597,0.0006176066,0.000021748527,0.00028963757],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99842644,0.000025377607,0.00042994902,0.0003006456,0.0006326093,0.0001849931],"domain_scores_gemma":[0.99868906,0.0002907297,0.00016190784,0.00033763843,0.00033379166,0.00018689394],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00010465132,0.00019398658,0.0007440777,0.03481357,0.000041854913,0.000010674683,0.000101340214,0.00004991907,0.00012318559],"category_scores_gemma":[0.00041092595,0.0001353303,0.00017307319,0.10740696,0.000125312,0.00006546212,0.000027696771,0.00016085485,0.0000045488077],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019536146,0.0007383763,0.95739996,0.00033632977,0.0033923106,0.00024514084,0.00045364004,0.009517615,0.020040624,0.00025902828,0.005036533,0.0006268186],"study_design_scores_gemma":[0.0014993864,0.00021464837,0.991701,0.00048407522,0.003336081,0.0000034683208,0.000027216278,0.0005348601,0.0017703987,0.00008174762,0.00021999286,0.0001271188],"about_ca_topic_score_codex":0.0010696409,"about_ca_topic_score_gemma":0.00007683135,"teacher_disagreement_score":0.07259339,"about_ca_system_score_codex":0.000045840352,"about_ca_system_score_gemma":0.00008138283,"threshold_uncertainty_score":0.976126},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2363093906","doi":"10.29173/cais516","title":"Is Grey Literature Ever Used? Using Citation Analysis to Measure the Impact of GESAMP, an International Marine Scientific Advisory Body","year":2013,"lang":"en","type":"article","venue":"Proceedings of the Annual Conference of CAIS / Actes du congrès annuel de l ACSI","topic":"Grey System Theory Applications","field":"Decision Sciences","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"","keywords":"Citation; Measure (data warehouse); Grey literature; Web of science; Library science; Scientific literature; Citation analysis; Political science; Computer science; MEDLINE; Database; Law; Geology","score_opus":0.08727320027593506,"score_gpt":0.3578664483545574,"score_spread":0.27059324807862234,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2363093906","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9953554,0.000028380842,0.00030567017,0.0013492533,0.00015044397,0.00071004935,0.00041423133,0.000018004897,0.0016685601],"genre_scores_gemma":[0.9983394,0.0000031069599,0.0007098011,0.00007966876,0.0000630534,0.00004697919,0.00001408933,0.000015200884,0.0007286619],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9960684,0.00013115468,0.0010277064,0.00060511346,0.0018453121,0.00032227935],"domain_scores_gemma":[0.8771666,0.00047199416,0.0015818409,0.00075722067,0.11983429,0.00018800703],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.003817184,0.0002759479,0.00057065947,0.0010483356,0.00026354278,0.0075836726,0.004142626,0.0001448466,0.00037761297],"category_scores_gemma":[0.02911861,0.00015982096,0.00049483916,0.0043377304,0.0004982,0.0131027335,0.000720325,0.00023331902,0.000016503614],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000119075354,0.00019568914,0.7879527,0.000035158835,0.0006266947,1.8869399e-7,0.06970111,0.00037387104,0.12860978,0.0061537144,0.0026469268,0.0035850434],"study_design_scores_gemma":[0.0002750494,0.00014523404,0.9496822,0.00012813743,0.00023630283,0.00001250411,0.0049232105,0.010163871,0.010305943,0.022505052,0.0013760177,0.00024651535],"about_ca_topic_score_codex":0.00066862593,"about_ca_topic_score_gemma":0.000041218922,"teacher_disagreement_score":0.16172941,"about_ca_system_score_codex":0.00009949465,"about_ca_system_score_gemma":0.00028353688,"threshold_uncertainty_score":0.9934465},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2382594364","doi":"","title":"Status and Trends of Biomass Energy Based on Bibliometrics","year":2014,"lang":"en","type":"article","venue":"Anhui nongye kexue","topic":"Power Systems and Renewable Energy","field":"Energy","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Biomass (ecology); Bibliometrics; China; Subject (documents); Index (typography); Web of science; Energy (signal processing); Environmental science; Environmental resource management; Agricultural engineering; Computer science; Geography; Ecology; Political science; Library science; Mathematics; Biology; Statistics; World Wide Web; Engineering; MEDLINE","score_opus":0.012900775079382733,"score_gpt":0.23633000807884583,"score_spread":0.2234292329994631,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2382594364","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.39790088,0.001852619,0.015878512,0.00035841943,0.0026117533,0.00008080082,0.00007576446,0.00030660105,0.58093464],"genre_scores_gemma":[0.9941358,0.00009496854,0.00028232718,0.00023994627,0.00020563733,0.000011535314,0.00004201084,0.000044712197,0.0049430435],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99826264,0.00013725377,0.00038112138,0.00037752223,0.00038658656,0.00045487867],"domain_scores_gemma":[0.9986945,0.00026662392,0.00017948513,0.00053421175,0.00008832083,0.00023687776],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00028788834,0.00024427092,0.0004034325,0.0058600195,0.00007940497,0.00004887824,0.00018949821,0.00019686579,0.00014561322],"category_scores_gemma":[0.000092076385,0.00019855019,0.00010509098,0.006996493,0.000089882655,0.00009511915,0.00005524488,0.00006997351,0.000009062009],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027795043,0.00082744757,0.017563263,0.0002777529,0.00027151126,0.00004188568,0.00017676536,0.017568283,0.026189294,0.19715175,0.016593853,0.72306025],"study_design_scores_gemma":[0.002163334,0.0008909769,0.016543746,0.00008728602,0.000039030187,0.0000058972855,0.000036012752,0.023415249,0.048641384,0.00062350224,0.9070165,0.00053708913],"about_ca_topic_score_codex":0.019951578,"about_ca_topic_score_gemma":0.0019376028,"teacher_disagreement_score":0.89042264,"about_ca_system_score_codex":0.00003245004,"about_ca_system_score_gemma":0.000035557296,"threshold_uncertainty_score":0.98657465},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2383316764","doi":"","title":"Bibliometric analysis on clinical research of lung transplantation from PubMed database","year":2011,"lang":"en","type":"article","venue":"Chinese Journal of Transplantation","topic":"Medical Research and Treatments","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Lung transplantation; Transplantation; Bronchiolitis obliterans; Medicine; China; Intensive care medicine; Family medicine; Political science; Internal medicine; Law","score_opus":0.22840597608702987,"score_gpt":0.48555386928892236,"score_spread":0.25714789320189246,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2383316764","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99056935,0.0004755388,0.0070731617,0.000100319055,0.0001627174,0.00029070798,0.00025539903,0.000006906907,0.0010658912],"genre_scores_gemma":[0.9864495,0.011274181,0.0015824178,0.0000382559,0.0001861568,0.000008932309,0.0004187359,0.000012896711,0.000028940714],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9961957,0.0004673265,0.0010450897,0.00021763337,0.0017837095,0.00029056484],"domain_scores_gemma":[0.9967054,0.001512557,0.00032570277,0.00025044396,0.00064429047,0.0005615742],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0033111246,0.00014133507,0.00061281794,0.03834197,0.000053480166,0.000013288189,0.00019285771,0.00012025465,0.00039354683],"category_scores_gemma":[0.0011153627,0.0000869958,0.00039116046,0.040599477,0.000108946835,0.00026723195,0.000005899037,0.00063071493,0.000009593157],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0116807055,0.0015042147,0.9624641,0.0003220768,0.0024439418,0.0015060942,0.00074969704,0.00000473311,0.00096291985,0.000029289378,0.00013285635,0.018199384],"study_design_scores_gemma":[0.0050621834,0.001176077,0.98730123,0.00023543545,0.0024034118,0.000063169675,0.000029394794,0.00018356796,0.0032505332,0.00022709835,0.000006394597,0.00006147781],"about_ca_topic_score_codex":0.00026134853,"about_ca_topic_score_gemma":0.000048916045,"teacher_disagreement_score":0.02483717,"about_ca_system_score_codex":0.000053514115,"about_ca_system_score_gemma":0.00019668759,"threshold_uncertainty_score":0.9797931},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2385673000","doi":"","title":"Bibliometric analysis of systematic review on tuberculosis literature in Web of science","year":2011,"lang":"en","type":"article","venue":"Chinese Journal of Disease Control and Prevention","topic":"Viral Infections and Vectors","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Tuberculosis; Web of science; Systematic review; Citation; Science Citation Index; Medicine; Bibliometrics; Citation analysis; MEDLINE; Original research; Family medicine; Meta-analysis; Political science; Library science; Pathology; Computer science","score_opus":0.01746753585444181,"score_gpt":0.31383190107005443,"score_spread":0.2963643652156126,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2385673000","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9356577,0.06377132,0.000074130825,0.00005673553,0.000044021606,0.00029576154,0.000011665485,0.0000015999786,0.000087070206],"genre_scores_gemma":[0.9917273,0.008164108,0.000020818314,0.000050256695,0.000014940823,0.0000045004163,0.0000021249964,0.0000027533763,0.000013218188],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9988083,0.0001098822,0.0006477711,0.00007963024,0.00028811346,0.000066261244],"domain_scores_gemma":[0.99868137,0.000052545125,0.0005265687,0.00014255181,0.00046431052,0.0001326356],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.001129411,0.000082472034,0.00075925066,0.031020733,0.000020722962,0.000009092809,0.000061067585,0.000022699938,0.00003459524],"category_scores_gemma":[0.0010564504,0.00004454336,0.00038246327,0.057203338,0.00006258322,0.00021224178,0.000007823041,0.000089577625,2.2891793e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001159875,0.0015799489,0.83660734,0.15161188,0.0020708137,0.00003218095,0.00028978032,0.00003077315,0.004631167,0.00013716266,0.000024079072,0.001825],"study_design_scores_gemma":[0.0010511815,0.00050557335,0.9385153,0.05235849,0.0066885184,0.000009181503,0.000008942458,0.00063201576,0.000022533564,0.00016878803,3.8119e-7,0.000039107126],"about_ca_topic_score_codex":0.000004440577,"about_ca_topic_score_gemma":0.0000027708147,"teacher_disagreement_score":0.101907946,"about_ca_system_score_codex":0.000019980287,"about_ca_system_score_gemma":0.00009205499,"threshold_uncertainty_score":0.9799618},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2392690500","doi":"","title":"A bibliometric analysis of systematic review on tuberculosis at home and abroad","year":2013,"lang":"en","type":"article","venue":"Xiandai yufang yixue","topic":"Tuberculosis Research and Epidemiology","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Tuberculosis; China; Web of science; Citation database; Bibliometrics; Systematic review; Science Citation Index; Grey literature; Medicine; Citation analysis; MEDLINE; Library science; Geography; Meta-analysis; Political science; Scopus; Pathology","score_opus":0.033336768450147866,"score_gpt":0.35600369920247527,"score_spread":0.3226669307523274,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2392690500","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9065697,0.08467368,0.000016849193,0.004812398,0.000036482183,0.0015965016,0.0000382684,0.000041926236,0.002214151],"genre_scores_gemma":[0.94362324,0.050660677,0.000138423,0.003778353,0.00003384823,0.00027884348,0.00009161022,0.000022881604,0.0013721285],"study_design_codex":"systematic_review","study_design_gemma":"observational","domain_scores_codex":[0.9976468,0.00031766246,0.00075409684,0.00037182373,0.00049777696,0.0004118164],"domain_scores_gemma":[0.9976555,0.00080938515,0.00020385916,0.0007021503,0.00025682905,0.00037228822],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0011000644,0.00020562412,0.0019360302,0.035436008,0.00006297588,0.000019470357,0.000138446,0.00012517873,0.0011405513],"category_scores_gemma":[0.0028236092,0.00013485477,0.00040346527,0.06610597,0.00010313321,0.000091359594,0.0000966565,0.0001876656,0.00045457203],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017074068,0.0005552764,0.31702513,0.545817,0.01205189,0.000057542613,0.00022548248,0.0000156039,0.004149503,0.00029117343,0.11528281,0.0043578693],"study_design_scores_gemma":[0.0009199965,0.0007210612,0.9509102,0.0355832,0.00887834,0.000081011545,0.000037811224,0.0017526927,0.00027280618,0.000043128184,0.0004990875,0.00030067845],"about_ca_topic_score_codex":0.00047359508,"about_ca_topic_score_gemma":0.000025430134,"teacher_disagreement_score":0.6338851,"about_ca_system_score_codex":0.00010903754,"about_ca_system_score_gemma":0.000027557167,"threshold_uncertainty_score":0.99977255},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2401719874","doi":"","title":"[Bibliometrical analysis of Medline publications in neuro-ophthalmology].","year":2005,"lang":"en","type":"article","venue":"PubMed","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Neuro-ophthalmology; Medicine; Ocular Motility Disorders; Ophthalmology; Optic nerve; Optic neuritis; Optic neuropathy; Optometry; Eye movement; Psychiatry; Glaucoma; Multiple sclerosis","score_opus":0.07894194979511747,"score_gpt":0.3153437314797627,"score_spread":0.2364017816846452,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2401719874","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98309684,0.00018669174,0.000039901122,0.0037703621,0.0000358465,0.00045000028,0.000114824776,0.00009799555,0.012207523],"genre_scores_gemma":[0.9976674,0.000007656926,0.00048736937,0.00011131367,0.000051988096,0.0011202242,0.000097078024,0.000030300776,0.00042668972],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9979197,0.00016301808,0.0005967626,0.0003667292,0.000430561,0.00052323757],"domain_scores_gemma":[0.99825865,0.00053692015,0.00021503579,0.00059564755,0.00018567914,0.00020808869],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.001178125,0.00013311989,0.000452511,0.040206566,0.000023484117,0.000028515613,0.0004410541,0.000097832024,0.00042026036],"category_scores_gemma":[0.003639775,0.00012747411,0.0001850378,0.12402143,0.000114761606,0.0003053652,0.00010052731,0.00018407726,0.00018267382],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000030707724,0.00089369,0.9690512,0.0000035927822,0.00044839893,0.000008045292,0.000038851023,0.0019795264,0.00013898005,0.0006838227,0.004644167,0.022079045],"study_design_scores_gemma":[0.00038921885,0.0000064616693,0.98896104,3.048586e-7,0.00033229028,0.000013526781,0.0000056334698,0.00461744,0.00020130385,0.00004650736,0.0053217467,0.000104497],"about_ca_topic_score_codex":0.00007208858,"about_ca_topic_score_gemma":0.00011357365,"teacher_disagreement_score":0.08381487,"about_ca_system_score_codex":0.00014628723,"about_ca_system_score_gemma":0.000027013857,"threshold_uncertainty_score":0.9706719},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2404777889","doi":"10.1525/jer.2006.1.4.77","title":"Trends in Canadian Sociology Master's Theses in Relation to Research Ethics Review, 1995–2004","year":2006,"lang":"en","type":"article","venue":"Journal of Empirical Research on Human Research Ethics","topic":"Contemporary Sociological Theory and Practice","field":"Social Sciences","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of New Brunswick","funders":"Medical Research Council","keywords":"Sociology; Research ethics; Field (mathematics); Qualitative research; Relation (database); Social science; Work (physics); Engineering ethics","score_opus":0.8652001765787157,"score_gpt":0.7007556206004423,"score_spread":0.16444455597827345,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2404777889","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2390643,0.006446407,0.000005529371,0.5625256,0.00017035891,0.00097015005,0.000015412244,0.000015515117,0.19078675],"genre_scores_gemma":[0.98368245,0.0064923596,0.0000774126,0.003002098,0.0009113973,0.000056995967,0.000012397168,0.000030250572,0.005734649],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.86194175,0.12345522,0.001575631,0.0007341385,0.009080934,0.003212328],"domain_scores_gemma":[0.88510406,0.107301034,0.00021953155,0.0005609338,0.0053665596,0.001447851],"candidate_categories":["metaresearch","sts","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","sts","research_integrity"],"category_scores_codex":[0.41856754,0.00022451638,0.00070032314,0.006269429,0.003596821,0.00032816903,0.0020310616,0.0019712967,0.0015494364],"category_scores_gemma":[0.22637995,0.00018796792,0.00022960284,0.006941913,0.0049444074,0.00064574316,0.00036170825,0.04524271,0.00024211036],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.0012454945,0.0010032794,0.04273519,0.0002708727,0.00003848819,0.0019898722,0.07915537,0.00016436176,0.00018468931,0.74756444,0.11966707,0.005980875],"study_design_scores_gemma":[0.00046082243,0.0018847829,0.0505326,0.0015417137,0.0000028270463,0.0000049239293,0.010258994,0.000007417517,0.00000904612,0.29136837,0.64372736,0.00020112103],"about_ca_topic_score_codex":0.31042948,"about_ca_topic_score_gemma":0.6853127,"teacher_disagreement_score":0.7446181,"about_ca_system_score_codex":0.0046780584,"about_ca_system_score_gemma":0.011306366,"threshold_uncertainty_score":0.9993633},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W2423415261","doi":"10.1016/j.respol.2016.05.009","title":"Which gender gap? Factors affecting researchers’ scientific impact in science and medicine","year":2016,"lang":"en","type":"article","venue":"Research Policy","topic":"Health and Medical Research Impacts","field":"Medicine","cited_by":185,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Polytechnique Montréal","funders":"","keywords":"Gender gap; Psychology; Sociology; Demographic economics; Economics","score_opus":0.6838478894611898,"score_gpt":0.6503567219247052,"score_spread":0.03349116753648462,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2423415261","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.913521,0.00034731143,0.000010823196,0.06914721,0.00005935441,0.0010031356,0.000008927449,0.00003933584,0.015862929],"genre_scores_gemma":[0.9970373,0.00045903007,0.000036035628,0.00043909572,0.00051591895,0.000021898162,0.0000026056332,0.000030330195,0.0014577871],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9863351,0.0008122507,0.00039636632,0.0008306471,0.0068810517,0.004744577],"domain_scores_gemma":[0.95216966,0.008423207,0.00009873187,0.0017664845,0.006448668,0.031093264],"candidate_categories":["metaresearch","sts"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.05528185,0.00019987012,0.0004190041,0.005378654,0.00082241255,0.00015662659,0.00046998946,0.00015872497,0.00044841578],"category_scores_gemma":[0.6148646,0.00010198665,0.00003940639,0.010556845,0.006009784,0.00048744443,0.0004883672,0.0013469197,0.0001134741],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.0007832059,0.0004150591,0.33784792,0.0009900542,0.000045954475,0.00024347923,0.011602537,3.088789e-7,0.45034406,0.0036872553,0.05518405,0.1388561],"study_design_scores_gemma":[0.0048475796,0.002326857,0.9666799,0.0015877588,0.0000047827953,0.000090554575,0.002195869,0.00012708566,0.0075120665,0.003180834,0.011226272,0.00022045021],"about_ca_topic_score_codex":0.006891842,"about_ca_topic_score_gemma":0.0003894531,"teacher_disagreement_score":0.628832,"about_ca_system_score_codex":0.0039647305,"about_ca_system_score_gemma":0.051647794,"threshold_uncertainty_score":0.99985886},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2486556987","doi":"10.1016/j.jcct.2016.07.013","title":"The 100 most-cited original articles in cardiac computed tomography: A bibliometric analysis","year":2016,"lang":"en","type":"review","venue":"Journal of cardiovascular computed tomography","topic":"Advanced X-ray and CT Imaging","field":"Engineering","cited_by":33,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vancouver General Hospital","funders":"","keywords":"Medicine; Bibliometrics; Web of science; Citation analysis; Scopus; Citation; Sample size determination; Subject (documents); Library science; MEDLINE; Meta-analysis; Statistics; Computer science; Pathology; Mathematics","score_opus":0.015324503304082169,"score_gpt":0.25754682969635834,"score_spread":0.24222232639227617,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2486556987","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00020016318,0.89783937,0.09977517,0.000009030185,0.0013445353,0.0004798751,0.00005017742,0.00017432625,0.000127328],"genre_scores_gemma":[0.0022170772,0.9941396,0.0025317394,0.00000924124,0.000902767,0.00003189294,0.000026540225,0.00013725604,0.000003869099],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9936255,0.00091795716,0.0022657644,0.00053826266,0.0016495835,0.0010028972],"domain_scores_gemma":[0.9959401,0.00097640936,0.00073925,0.001296131,0.0006600786,0.000388042],"candidate_categories":["metaepi_narrow","metaepi_broad","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0026201126,0.0009819639,0.005234086,0.1308525,0.0002451217,0.00037835524,0.0013531736,0.00038085954,0.00000854408],"category_scores_gemma":[0.0000715638,0.0006459503,0.015227662,0.29557708,0.00020340148,0.00047597208,0.00013944946,0.0014704484,0.000012253646],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012696367,0.000042953005,0.00041158276,0.0011637001,0.06063742,0.00025843136,0.000023847813,0.03798575,6.339922e-7,0.000024993786,0.0003685722,0.8990694],"study_design_scores_gemma":[0.000869604,0.00006556784,0.0011630374,0.004511316,0.014790901,0.0002016451,0.00001681378,0.0019052278,0.0000062201557,0.00012059859,0.9755294,0.0008197047],"about_ca_topic_score_codex":0.0000117259215,"about_ca_topic_score_gemma":0.0000023267598,"teacher_disagreement_score":0.9751608,"about_ca_system_score_codex":0.00021206168,"about_ca_system_score_gemma":0.00014520927,"threshold_uncertainty_score":0.99959916},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2507633606","doi":"","title":"Bibliometric study of Dr. R.C.Sinha, a plant pathologist","year":2001,"lang":"en","type":"article","venue":"Annals of Library and Information Studies","topic":"Plant Virus Research Studies","field":"Agricultural and Biological Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Science Citation Index; Library science; Citation; Citation analysis; Bibliometrics; Index (typography); Geography; Social science; History; Sociology; Computer science","score_opus":0.19567759517939828,"score_gpt":0.34513745638875504,"score_spread":0.14945986120935675,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2507633606","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99201787,0.0032154787,1.8217412e-7,0.001691111,0.000024115983,0.00018191728,0.0003057578,0.00002779147,0.0025357588],"genre_scores_gemma":[0.94701153,0.052507482,0.000007840839,0.0003093643,0.000022747328,0.000012697035,0.00004537208,2.6171273e-7,0.00008269856],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.999101,0.000055943172,0.00037793387,0.0000768463,0.00023539993,0.00015288932],"domain_scores_gemma":[0.99927485,0.0003657993,0.00019192825,0.000036925794,0.00009195684,0.000038546976],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00020867806,0.0000906278,0.00025943897,0.0011754993,0.0001443659,0.000032004977,0.00012643011,0.00002804628,0.00003462131],"category_scores_gemma":[0.00015451711,0.000032997286,0.00003449015,0.00920491,0.00012619524,0.003519554,0.00024270901,0.00005093406,0.0000050188637],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010454903,0.001293213,0.55781496,0.00032391524,0.0007770712,0.00003104346,0.008749134,0.000012505812,0.006738093,0.0021495842,0.1360962,0.28496882],"study_design_scores_gemma":[0.00021268344,0.0016281685,0.9120652,0.00003693632,0.000007821917,0.0000070160763,0.015636975,0.000010405663,0.0030032322,0.00019154273,0.06708876,0.00011127203],"about_ca_topic_score_codex":0.000018596476,"about_ca_topic_score_gemma":0.000004041395,"teacher_disagreement_score":0.35425025,"about_ca_system_score_codex":6.1073763e-7,"about_ca_system_score_gemma":0.0000031268175,"threshold_uncertainty_score":0.44226542},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2511751023","doi":"10.1186/s40064-016-3165-6","title":"Bibliometric profile of the global scientific research on autism spectrum disorders","year":2016,"lang":"en","type":"article","venue":"SpringerPlus","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":83,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Autism; Scopus; Citation; Bibliometrics; Library science; MEDLINE; Psychology; Medicine; Political science; Psychiatry; Computer science","score_opus":0.47844034927722484,"score_gpt":0.5618977958917807,"score_spread":0.08345744661455584,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2511751023","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9342546,0.00060315785,0.00069131487,0.014739445,0.0028069513,0.0006004771,0.00012278809,0.000035547066,0.04614574],"genre_scores_gemma":[0.98923403,0.000112445625,0.00013749598,0.000023224551,0.00006275977,0.000018118497,3.467463e-7,0.00001269869,0.010398871],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.97637284,0.0006556888,0.0007387757,0.0012001301,0.019769229,0.0012633473],"domain_scores_gemma":[0.9901704,0.0054070884,0.00029861036,0.0025276435,0.001157677,0.00043855974],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.038997848,0.000180673,0.00030836687,0.3939531,0.0008376112,0.0019063562,0.0060659507,0.00014141227,0.0013072971],"category_scores_gemma":[0.037076153,0.00008099148,0.00028860653,0.88153934,0.0015915708,0.0004464787,0.0024757297,0.00035857985,0.0022524446],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000055576613,0.00034292488,0.4506551,0.0000089219875,0.000017472194,0.0000045053243,0.00003886722,0.000012901258,0.0015234798,0.105715014,0.019152952,0.4224723],"study_design_scores_gemma":[0.00046210067,0.00021839734,0.8180109,0.000035747904,0.000002111298,0.0000020617338,0.000043689757,0.00017244993,0.00533789,0.119023755,0.05655034,0.00014052476],"about_ca_topic_score_codex":0.00017326903,"about_ca_topic_score_gemma":0.00012765652,"teacher_disagreement_score":0.48758623,"about_ca_system_score_codex":0.0005054498,"about_ca_system_score_gemma":0.00063922704,"threshold_uncertainty_score":0.99960566},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2513105786","doi":"10.1136/emermed-2016-205893","title":"Analysis of h-index and other bibliometric markers of productivity and repercussion of a selected sample of worldwide emergency medicine researchers","year":2016,"lang":"en","type":"article","venue":"Emergency Medicine Journal","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Medicine; Sample (material); Index (typography); Productivity; Emergency medicine; Family medicine; World Wide Web; Chromatography","score_opus":0.42254164564654606,"score_gpt":0.5466975217501402,"score_spread":0.12415587610359413,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2513105786","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9798153,0.010293062,0.0055045984,0.0018180513,0.000577732,0.0002174596,0.0000601655,0.0000041443773,0.0017094985],"genre_scores_gemma":[0.9766949,0.019341642,0.0001769105,0.0000014165101,0.00012746122,0.0000021386477,0.0000012227791,0.000013623147,0.0036406757],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9850484,0.0009223851,0.0031870347,0.0006405232,0.009693507,0.0005081591],"domain_scores_gemma":[0.9806989,0.0048860526,0.0024510988,0.00082022784,0.010513085,0.0006306745],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.049161963,0.00021494833,0.0013061462,0.39925143,0.00011236019,0.000010150306,0.0010868433,0.00011665072,0.03061717],"category_scores_gemma":[0.21457341,0.00010252651,0.00021498055,0.66610014,0.0010486315,0.00041792318,0.00034055847,0.00031079704,6.1408565e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003081938,0.00014594189,0.870708,0.00006801377,0.00045739178,5.414011e-7,0.0005206023,0.0000050997282,0.06381589,0.000020358279,0.025015572,0.038934402],"study_design_scores_gemma":[0.0010737454,0.001361963,0.98973197,0.0002063266,0.00022865093,0.000008190136,0.0007189006,0.00051640806,0.0018077301,0.0018531556,0.0023789753,0.00011397759],"about_ca_topic_score_codex":0.001515959,"about_ca_topic_score_gemma":0.00021202365,"teacher_disagreement_score":0.2668487,"about_ca_system_score_codex":0.000048705053,"about_ca_system_score_gemma":0.0002029518,"threshold_uncertainty_score":0.9790878},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2517558721","doi":"10.1007/s11192-016-2098-8","title":"University–industry R&amp;D linkage metrics: validity and applicability in world university rankings","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"Innovation and Knowledge Management","field":"Business, Management and Accounting","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Directorate-General for Education and Culture; Newfoundland and Labrador; Ministerie van Onderwijs, Cultuur en Wetenschap; European Commission","keywords":"Ranking (information retrieval); Linkage (software); Metric (unit); Computer science; Data science; Index (typography); Strengths and weaknesses; Information retrieval; Operations research; Data mining; Mathematics; Marketing; Psychology; Business; World Wide Web","score_opus":0.05807529621679081,"score_gpt":0.25030281545862426,"score_spread":0.19222751924183346,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2517558721","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.67716384,0.000022731472,0.0051755765,0.0016652239,0.000548836,0.00038930855,0.000009254714,0.00016809157,0.31485713],"genre_scores_gemma":[0.98152345,0.000014977438,0.0003783974,0.0003203903,0.0001393381,3.909426e-7,0.0000048294824,0.000008418319,0.017609816],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99853355,0.000031175998,0.00017538424,0.00046603335,0.00045458053,0.00033925506],"domain_scores_gemma":[0.99897736,0.0002080653,0.00015168166,0.000293761,0.00033633513,0.000032809185],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0026268528,0.00014029682,0.00017140049,0.010013203,0.00020616195,0.0001543717,0.00040120652,0.00011470019,0.00020621838],"category_scores_gemma":[0.00086624443,0.00012477642,0.000045757573,0.038387477,0.00016490939,0.0008167187,0.0006124422,0.00021560836,0.00023844161],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003854123,0.0001971645,0.71600384,0.000114888935,0.000010280785,0.00001951598,0.000041610005,0.0000020951777,0.00009393642,0.24945953,0.003582604,0.030435989],"study_design_scores_gemma":[0.001345325,0.000004858448,0.1292542,0.000030852632,0.000024838075,2.5596788e-7,0.00034561125,0.00021689312,0.00001890902,0.0015482953,0.86693823,0.0002717024],"about_ca_topic_score_codex":0.00008886478,"about_ca_topic_score_gemma":0.00026555316,"teacher_disagreement_score":0.86335564,"about_ca_system_score_codex":0.00031697552,"about_ca_system_score_gemma":0.0000251967,"threshold_uncertainty_score":0.9820521},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2518185335","doi":"","title":"Produção científica acerca do modelo Calgary de avaliação da família: um estudo bibliométrico","year":2015,"lang":"pt","type":"article","venue":"Redalyc (Universidad Autónoma del Estado de México)","topic":"Rural Development and Agriculture","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Psychology; Political science; Humanities; Philosophy","score_opus":0.024290461624305707,"score_gpt":0.2361485106268789,"score_spread":0.21185804900257318,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2518185335","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95436704,0.0015862373,0.004729205,0.004566342,0.0010442426,0.001373628,0.00011252247,0.0004060769,0.031814683],"genre_scores_gemma":[0.90861076,0.00075421983,0.007754882,0.0005346827,0.0002700569,0.00003644537,0.00012925,0.00011831988,0.08179137],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.992768,0.00044928517,0.0007630029,0.0018547904,0.0018287181,0.0023361864],"domain_scores_gemma":[0.99611455,0.00010555029,0.000521904,0.001121736,0.00018365003,0.001952607],"candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0012246958,0.0011274907,0.00088860677,0.001646188,0.0009124538,0.0006722318,0.002064582,0.0009620094,0.0018775053],"category_scores_gemma":[0.00015219503,0.0010624669,0.000451324,0.009318044,0.00051828026,0.0019977586,0.001580891,0.001182612,0.0024178035],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008533859,0.0030690478,0.23159254,0.00034129628,0.000706517,0.005309021,0.028566265,0.018769128,0.02768598,0.0037953884,0.60080457,0.07850684],"study_design_scores_gemma":[0.010529461,0.0014728653,0.5994924,0.0008521744,0.001211199,0.0019291574,0.03622952,0.20397659,0.0029828222,0.0016566162,0.13296767,0.0066995826],"about_ca_topic_score_codex":0.0010698565,"about_ca_topic_score_gemma":0.00017644705,"teacher_disagreement_score":0.46783695,"about_ca_system_score_codex":0.0035189814,"about_ca_system_score_gemma":0.0008277662,"threshold_uncertainty_score":0.9991826},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2522666184","doi":"10.1097/01.gox.0000503206.60532.9c","title":"Abstract: The Canadian Contribution to the Global Plastic Surgery Literature: A 10-Year Bibliometric Analysis","year":2016,"lang":"en","type":"article","venue":"Plastic & Reconstructive Surgery Global Open","topic":"Biomedical Ethics and Regulation","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Specialty; Inclusion (mineral); Medicine; Family medicine; Library science; MEDLINE; Psychology; Political science; Social science; Sociology","score_opus":0.024748079878717348,"score_gpt":0.29149304376663365,"score_spread":0.2667449638879163,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2522666184","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93124753,0.001132912,0.0040623504,0.04341704,0.0075909393,0.0020796717,0.0057692803,0.00010314911,0.004597142],"genre_scores_gemma":[0.9985843,0.000084253625,0.00007034024,0.0005876407,0.0003398146,0.00005779747,0.00019569314,0.000012371928,0.00006782492],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99674964,0.00023057258,0.0007235556,0.00062249386,0.0008944735,0.0007792661],"domain_scores_gemma":[0.98794687,0.009235885,0.0003126203,0.00054120837,0.0010138942,0.0009495523],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0022968338,0.00032776012,0.00080586167,0.006425658,0.00054104615,0.0005253958,0.00044263457,0.00038509522,0.0012833902],"category_scores_gemma":[0.013467087,0.00015418697,0.00048223184,0.07938244,0.0008402492,0.00029479063,0.0001813692,0.00029186817,0.00037017727],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009983018,0.00009381676,0.84943265,0.000038776176,0.00214814,0.00013874123,0.000031567182,0.000024541532,0.000018557046,0.01117934,0.022872603,0.11302295],"study_design_scores_gemma":[0.00041854597,0.000037729616,0.977838,0.0003816455,0.0005825638,0.00015593212,0.000034178203,0.00007688262,0.000012277652,0.0018692756,0.018352222,0.00024076218],"about_ca_topic_score_codex":0.014524948,"about_ca_topic_score_gemma":0.06966074,"teacher_disagreement_score":0.12840532,"about_ca_system_score_codex":0.0013095101,"about_ca_system_score_gemma":0.002393099,"threshold_uncertainty_score":0.99962956},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2524647697","doi":"10.29173/cais283","title":"A Concise Review On The Role Of Author Self-Citations In Information Science, Bibliometrics And Science Policy","year":2013,"lang":"fr","type":"review","venue":"Proceedings of the Annual Conference of CAIS / Actes du congrès annuel de l ACSI","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Citation; Interpretation (philosophy); Humanities; Epistemology; Sociology; Library science; Philosophy; Computer science; Linguistics","score_opus":0.2356972037868541,"score_gpt":0.46773803466837877,"score_spread":0.23204083088152466,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2524647697","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":"evaluation","model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.33887187,0.6049777,0.000017501472,0.0107355,0.0006075855,0.0057707946,0.00107495,0.00003255823,0.037911523],"genre_scores_gemma":[0.40064725,0.59823364,0.0005153039,0.0002503284,0.000039312163,0.0001065605,0.0000021753856,0.000014523087,0.0001909075],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.98290175,0.00019434703,0.0029515317,0.000926067,0.01173099,0.0012953324],"domain_scores_gemma":[0.3312202,0.013807357,0.011537852,0.0020555947,0.6397456,0.0016333862],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.045850568,0.00058435526,0.0018978583,0.16021524,0.00072537456,0.018923394,0.013652999,0.00033548966,0.00010551],"category_scores_gemma":[0.6337134,0.00032973647,0.0004230326,0.5368283,0.0072029843,0.03599976,0.0043288944,0.0010183959,0.000045124245],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019599524,0.00033953093,0.02075864,0.007663431,0.000049600447,3.6486446e-7,0.008093997,0.0000016779923,0.00060219434,0.1369368,0.0022256179,0.8233085],"study_design_scores_gemma":[0.0005572544,0.0009402238,0.056154244,0.02072025,0.00024024163,0.000054337535,0.005369681,0.002544943,0.0025720536,0.0142700635,0.89579695,0.0007797597],"about_ca_topic_score_codex":0.00076415617,"about_ca_topic_score_gemma":0.0000047418102,"teacher_disagreement_score":0.8935713,"about_ca_system_score_codex":0.0005355266,"about_ca_system_score_gemma":0.009922325,"threshold_uncertainty_score":0.9999155},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2526349856","doi":"10.18438/b8mm0j","title":"Retroactive Collection Development: Gap Identification through Citation Analysis","year":2016,"lang":"en","type":"article","venue":"Evidence Based Library and Information Practice","topic":"Academic Writing and Publishing","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"University of Northern British Columbia","funders":"","keywords":"Collection development; Identification (biology); Computer science; Citation; Data science; Information retrieval; Library science","score_opus":0.040432228185220145,"score_gpt":0.2586946566664715,"score_spread":0.21826242848125138,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2526349856","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06443964,0.0016071423,0.287391,0.35755754,0.0014897956,0.0012015827,0.00010859566,0.0013117353,0.28489298],"genre_scores_gemma":[0.97880495,0.00060488394,0.0029729758,0.01380284,0.00025236202,0.000047934733,0.00014591862,0.000008248345,0.0033599052],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987917,0.00019674386,0.00046086917,0.00014719927,0.00027518722,0.00012830207],"domain_scores_gemma":[0.9975808,0.0015276195,0.00054981513,0.0001259291,0.00016826404,0.000047524747],"candidate_categories":["scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.00074826885,0.00010243012,0.00010576413,0.00036623346,0.0005783099,0.0011032085,0.00011218166,0.00006721046,0.00082701753],"category_scores_gemma":[0.002007888,0.000077680466,0.000038573005,0.00044428094,0.000076186596,0.3487514,0.000028064145,0.00013546576,0.00009136486],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002856366,0.00002773327,0.0009903193,0.000071309885,0.00015991012,4.710548e-7,0.02198283,0.00003577825,0.000077118326,0.9313752,0.0127511015,0.032242592],"study_design_scores_gemma":[0.00023998065,0.00003782273,0.0067210584,0.0001747055,0.00017049817,0.000002117162,0.0070402683,0.0008476705,0.0016051467,0.0008067635,0.9821681,0.00018586316],"about_ca_topic_score_codex":0.000017171238,"about_ca_topic_score_gemma":4.9576647e-7,"teacher_disagreement_score":0.969417,"about_ca_system_score_codex":0.000029678517,"about_ca_system_score_gemma":0.00012782335,"threshold_uncertainty_score":0.9999337},"labels":[{"model":"gemma","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2531842628","doi":"10.1186/s41073-016-0022-7","title":"Is it becoming harder to secure reviewers for peer review? A test with data from five ecology journals","year":2016,"lang":"en","type":"article","venue":"Research Integrity and Peer Review","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"World Wildlife Fund Canada; University of British Columbia; Women's Health Research Institute","funders":"","keywords":"Ecology; Community; Publishing; Functional ecology; Evolutionary ecology; Rhetorical question; Biology; Habitat; Political science; Ecosystem; Law","score_opus":0.9221730707033521,"score_gpt":0.7096993867356763,"score_spread":0.21247368396767574,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2531842628","genre_codex":"commentary","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0005110418,0.34053174,0.0014475455,0.65045696,0.00015405026,0.002460596,0.003103672,0.000009509171,0.0013248941],"genre_scores_gemma":[0.001295789,0.90540236,0.004976423,0.020568965,0.00029994838,0.00027006606,0.00009847949,0.000030421204,0.06705754],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9717272,0.0017873342,0.0015532733,0.0019223058,0.021658812,0.0013510691],"domain_scores_gemma":[0.6633531,0.13572043,0.0015033262,0.011049611,0.18310234,0.005271216],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","insufficient_payload"],"category_scores_codex":[0.22905879,0.00031132856,0.0014028784,0.008199491,0.0005751274,0.0018189815,0.0058880993,0.00020723142,0.009945498],"category_scores_gemma":[0.6551285,0.00014581611,0.00023535629,0.04588677,0.0003807069,0.0010729847,0.002725561,0.001550264,0.0017139163],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000022627122,0.000071562594,0.0042054285,0.00096979976,0.000036505924,0.000012503712,0.000059838872,1.21300054e-8,0.000024506942,0.00005459077,0.82216835,0.17237426],"study_design_scores_gemma":[0.00042553712,0.00042062256,0.0031072118,0.019074518,0.000047578,0.000028095124,0.00013084522,0.00005704773,0.000027551963,0.0020120258,0.97442335,0.00024563877],"about_ca_topic_score_codex":0.00031066744,"about_ca_topic_score_gemma":0.00045767368,"teacher_disagreement_score":0.629888,"about_ca_system_score_codex":0.00018788362,"about_ca_system_score_gemma":0.00092774985,"threshold_uncertainty_score":0.9994905},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2531929339","doi":"","title":"Bibliometric Analysis of National Cancer Institute of Canada-Supported Research","year":2008,"lang":"en","type":"article","venue":"","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Political science; Geography; Regional science; Library science; Computer science","score_opus":0.8147265308711052,"score_gpt":0.6424995894648025,"score_spread":0.1722269414063028,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2531929339","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9555601,0.0006097442,0.0006842964,0.00047506628,0.00031090988,0.00013500897,0.00023064826,0.0000056254175,0.04198859],"genre_scores_gemma":[0.98712754,0.00041506873,0.00041768298,0.00004824426,0.000028897386,0.000009207771,0.0000090879785,0.00000459439,0.011939686],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.96133906,0.00022343743,0.0011714073,0.00056011404,0.036099456,0.00060653465],"domain_scores_gemma":[0.96204656,0.0054780645,0.00035741966,0.0006236568,0.031078432,0.00041584464],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.029366605,0.000102970116,0.00057163043,0.91774285,0.00026020722,0.00014795912,0.002351923,0.00008777718,0.005309349],"category_scores_gemma":[0.06060376,0.0000708344,0.00021306243,0.9878314,0.00057102705,0.0004927162,0.0005276447,0.0002346808,0.000013614221],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000018841354,0.00017199847,0.7859941,0.0000048354477,0.00036184196,0.000014312981,0.00006151029,0.0017614482,0.0010856057,0.0010318095,0.19380817,0.015685517],"study_design_scores_gemma":[0.00020041878,0.000048092134,0.9632586,0.0000020093191,0.000016782504,0.0000016923734,0.00008173731,0.009857437,0.0038548245,0.0005599638,0.022031872,0.00008653654],"about_ca_topic_score_codex":0.5247912,"about_ca_topic_score_gemma":0.27629018,"teacher_disagreement_score":0.248501,"about_ca_system_score_codex":0.00035067202,"about_ca_system_score_gemma":0.008179093,"threshold_uncertainty_score":0.99947137},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W2553147471","doi":"10.1186/s12913-016-1904-6","title":"Methods used to address fidelity of receipt in health intervention research: a citation analysis and systematic review","year":2016,"lang":"en","type":"review","venue":"BMC Health Services Research","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":120,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital","funders":"City, University of London","keywords":"Receipt; Fidelity; Citation; Health informatics; Psychological intervention; Medicine; Data extraction; Systematic review; Intervention (counseling); Biostatistics; Nursing research; Public health; MEDLINE; Computer science; Nursing; World Wide Web; Political science","score_opus":0.8768176192761763,"score_gpt":0.8135927768787335,"score_spread":0.06322484239744275,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2553147471","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00003798263,0.96008974,0.0020101243,0.0044825478,0.00017808433,0.03248675,0.00052743923,0.000047560516,0.00013975234],"genre_scores_gemma":[0.000028388355,0.97959435,0.006354507,0.0030537222,0.00011894541,0.0101631265,0.0001462272,0.00007546199,0.00046529042],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.7348617,0.24354552,0.012207221,0.001769797,0.004094049,0.0035217195],"domain_scores_gemma":[0.93409634,0.05355746,0.00472767,0.0023197667,0.0033539438,0.0019448018],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","research_integrity"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.34669733,0.000515479,0.007231952,0.009849807,0.0014137358,0.00006458577,0.0017200778,0.0004915367,0.00052647444],"category_scores_gemma":[0.01669248,0.0003807912,0.00046476527,0.023997916,0.0002249563,0.0003829674,0.0012240625,0.0028034074,0.00047465513],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000016450333,0.000043613643,0.00019333213,0.8664023,0.00007888093,5.4795174e-7,0.009494971,1.3094797e-7,1.5370118e-7,0.00013067138,0.00038393392,0.12325497],"study_design_scores_gemma":[0.0003793933,0.00031239836,0.00085943856,0.7930966,0.000111192676,0.0000011389311,0.0047917687,0.000036459423,1.1787655e-7,0.00010381486,0.20010756,0.00020011463],"about_ca_topic_score_codex":0.025883565,"about_ca_topic_score_gemma":0.041004676,"teacher_disagreement_score":0.33000484,"about_ca_system_score_codex":0.004550843,"about_ca_system_score_gemma":0.012949251,"threshold_uncertainty_score":0.9998863},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2556861239","doi":"10.1186/s12968-016-0303-9","title":"Top 100 cited articles in cardiovascular magnetic resonance: a bibliometric analysis","year":2016,"lang":"en","type":"review","venue":"Journal of Cardiovascular Magnetic Resonance","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":88,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vancouver General Hospital; University of British Columbia","funders":"","keywords":"Scopus; Web of science; Bibliometrics; Medicine; Citation; MEDLINE; Citation analysis; Library science; Family medicine; Meta-analysis; Computer science; Political science; Internal medicine","score_opus":0.07319567428049205,"score_gpt":0.3701037668851918,"score_spread":0.29690809260469975,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2556861239","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00057141774,0.9949289,0.0008326631,0.00009449187,0.0010901723,0.00088197173,0.00009133458,0.000033234213,0.0014758308],"genre_scores_gemma":[0.0003822884,0.9952545,0.00081026176,0.0000685058,0.0011713053,0.00008016746,0.000004719374,0.000118058524,0.0021101825],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.97154474,0.0052832966,0.0067179487,0.0021016386,0.012879473,0.0014729124],"domain_scores_gemma":[0.98588526,0.0037931914,0.002202082,0.00497991,0.0023971256,0.0007424452],"candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","bibliometrics","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","bibliometrics","research_integrity"],"category_scores_codex":[0.042682532,0.0012176511,0.012034601,0.13844386,0.00020796592,0.0022554104,0.008980818,0.0012981325,0.00076768355],"category_scores_gemma":[0.021192007,0.0007729624,0.020710552,0.32588413,0.00045669454,0.0020885426,0.0010048823,0.0025282153,0.00030477825],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000035188583,0.000066471395,0.00087314256,0.00030435322,0.0028273687,0.0009957249,0.000066675,0.00017312991,5.6566183e-8,0.00001960523,0.004724405,0.9899139],"study_design_scores_gemma":[0.001466165,0.0002082609,0.013133241,0.005164119,0.010538958,0.00071087416,0.000055482124,0.00011217746,8.1515765e-7,0.0005587186,0.9671428,0.00090834],"about_ca_topic_score_codex":0.00017569633,"about_ca_topic_score_gemma":0.000011378407,"teacher_disagreement_score":0.98900557,"about_ca_system_score_codex":0.00059836695,"about_ca_system_score_gemma":0.0013788124,"threshold_uncertainty_score":0.9999984},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2579812090","doi":"10.1186/s13104-016-2371-1","title":"A bibliometric analysis of two decades of aromatherapy research","year":2017,"lang":"en","type":"article","venue":"BMC Research Notes","topic":"Olfactory and Sensory Function Studies","field":"Neuroscience","cited_by":125,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Ontario","funders":"","keywords":"Aromatherapy; Web of science; Medicine; Citation; Bibliometrics; Science Citation Index; MEDLINE; Alternative medicine; Medical education; Computer science; Library science; Meta-analysis; Pathology; Political science","score_opus":0.8178627288745213,"score_gpt":0.5691392540613007,"score_spread":0.24872347481322055,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2579812090","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9940803,0.0002915371,0.00009850263,0.00015325483,0.000081385544,0.00023025666,0.00004431243,0.00001772711,0.005002707],"genre_scores_gemma":[0.997363,0.00063234824,0.0001964732,0.000007205649,0.00005832476,0.000014155742,6.633866e-7,0.000011672618,0.0017161947],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9948865,0.0013988941,0.0003639027,0.0004842224,0.00230322,0.00056322233],"domain_scores_gemma":[0.9855291,0.011922708,0.00018625257,0.0012038064,0.001040087,0.000118049706],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0057695233,0.00010641212,0.00043589636,0.100121446,0.0012029965,0.00018103988,0.0010322508,0.00006403039,0.00037328733],"category_scores_gemma":[0.040714785,0.00008326264,0.00021738368,0.12151664,0.0020035976,0.0003092969,0.0004173456,0.00041524292,0.00008935257],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022433644,0.00024394668,0.4209071,0.00006546422,0.00014909731,0.0000067273295,0.0004854559,0.00008895498,0.5712587,0.002703479,0.000456279,0.0034104253],"study_design_scores_gemma":[0.00038728208,0.00024167616,0.47701293,0.00001942917,0.000028665592,6.4854845e-7,0.00019659735,0.0006732449,0.51893735,0.0012074893,0.0012165991,0.000078065525],"about_ca_topic_score_codex":0.0004962005,"about_ca_topic_score_gemma":0.00015336853,"teacher_disagreement_score":0.056105834,"about_ca_system_score_codex":0.000031015068,"about_ca_system_score_gemma":0.00012506775,"threshold_uncertainty_score":0.9673657},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2597139062","doi":"10.1177/2325967117694024","title":"The h-Index of Editorial Board Members Correlates Positively With the Impact Factor of Sports Medicine Journals","year":2017,"lang":"en","type":"article","venue":"Orthopaedic Journal of Sports Medicine","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Impact factor; Editorial board; Medicine; Index (typography); Bibliometrics; Interquartile range; Library science; Statistics; Internal medicine; Mathematics; Political science; Computer science; World Wide Web","score_opus":0.18094606262248883,"score_gpt":0.49283091591464406,"score_spread":0.3118848532921552,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2597139062","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9511468,0.009916226,0.0006981899,0.014989767,0.019276582,0.00036353475,0.000021374495,0.000005266549,0.0035823064],"genre_scores_gemma":[0.98199034,0.006394885,0.0000312224,0.000051082843,0.010837875,0.0000010113953,8.56688e-7,0.000022575927,0.0006701642],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.96937484,0.00023854552,0.0028619594,0.0003835852,0.026447058,0.00069404073],"domain_scores_gemma":[0.9702709,0.0060797897,0.009380488,0.0013307687,0.012055094,0.00088299386],"candidate_categories":["metaresearch","bibliometrics","sts","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.054875925,0.0003559022,0.0014455469,0.012253895,0.0008738811,0.0003504825,0.003998798,0.00019616474,0.0014067218],"category_scores_gemma":[0.033934727,0.00011887064,0.00049147266,0.017002009,0.0035496226,0.0006297498,0.00031986373,0.0012364754,0.0000017988934],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001096975,0.00007594057,0.6853396,0.000009366463,0.00022132165,0.00033911288,0.0007097818,0.00007440205,0.00034049802,0.000056452936,0.13317758,0.17855896],"study_design_scores_gemma":[0.003080174,0.0017554959,0.9771229,0.0011359218,0.00014386442,0.00039580654,0.0026766348,0.00011731299,0.00016315086,0.0019301993,0.011327972,0.00015055393],"about_ca_topic_score_codex":0.00034232816,"about_ca_topic_score_gemma":0.000015640115,"teacher_disagreement_score":0.2917833,"about_ca_system_score_codex":0.00010438762,"about_ca_system_score_gemma":0.0009855551,"threshold_uncertainty_score":0.9995061},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W2599097535","doi":"10.1097/md.0000000000004822","title":"The top-cited systematic reviews/meta-analyses in tuberculosis research","year":2017,"lang":"en","type":"review","venue":"Medicine","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Department of Science and Technology of Sichuan Province; National Natural Science Foundation of China","keywords":"Medicine; Tuberculosis; Systematic review; Meta-analysis; Web of science; Citation; MEDLINE; Original research; Science Citation Index; Bibliometrics; Citation analysis; Family medicine; Library science; Pathology","score_opus":0.9922855861369225,"score_gpt":0.7722278828780111,"score_spread":0.22005770325891139,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2599097535","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[5.4162695e-7,0.97620076,0.00008681349,0.0024109036,0.0006479998,0.010742282,0.000014859299,0.000006169493,0.00988965],"genre_scores_gemma":[0.000010527009,0.93519527,0.00008591164,0.00010506885,0.00047819898,0.002947825,0.000029511028,0.000049621132,0.06109804],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.60832304,0.2865372,0.07109173,0.0039220923,0.028632341,0.0014935975],"domain_scores_gemma":[0.7494713,0.16641381,0.0398103,0.0393264,0.004373365,0.00060484174],"candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","metaepi_narrow","metaepi_broad","insufficient_payload"],"category_scores_codex":[0.7045087,0.0013835246,0.06962358,0.003626436,0.0010439799,0.0022982813,0.014108125,0.00047314048,0.013276994],"category_scores_gemma":[0.660269,0.00034537225,0.015910795,0.008287948,0.00075832853,0.0002301002,0.0007424893,0.0017260363,0.017337218],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000013784721,0.000036985326,0.00001540841,0.17779113,0.008871342,0.00003929725,0.0002111139,3.987773e-7,7.178925e-8,0.0005890309,0.29510555,0.5173383],"study_design_scores_gemma":[0.00008239877,0.00003192105,0.000010379891,0.06827598,0.04249391,0.000024565139,0.00019410382,0.000044242177,1.2746367e-8,0.00080681825,0.8877743,0.00026135438],"about_ca_topic_score_codex":0.00015805381,"about_ca_topic_score_gemma":0.00025008866,"teacher_disagreement_score":0.5926688,"about_ca_system_score_codex":0.00023823457,"about_ca_system_score_gemma":0.0002990379,"threshold_uncertainty_score":0.9998998},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2599395542","doi":"10.1177/0306312717694512","title":"From ‘just the facts’ to ‘more theory and methods, please’: The evolution of the research article in <i>Administrative Science Quarterly</i> , 1956–2008","year":2017,"lang":"en","type":"article","venue":"Social Studies of Science","topic":"Management and Organizational Studies","field":"Business, Management and Accounting","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Discipline; Section (typography); Epistemology; Period (music); Sociology; Resizing; Positive economics; Social science; Computer science; Economics; Philosophy","score_opus":0.11762719503584612,"score_gpt":0.42968128385042054,"score_spread":0.3120540888145744,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2599395542","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9730543,0.0005064513,0.000096949705,0.017790431,0.00023079856,0.00042104645,0.000003694115,0.0000062704303,0.007890095],"genre_scores_gemma":[0.9993736,0.00001454573,0.00011364551,0.00022166138,0.00018713811,0.000013643185,8.3520476e-8,0.0000030625392,0.00007261459],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.998393,0.00007748286,0.00018840459,0.0002517049,0.00081802683,0.00027139141],"domain_scores_gemma":[0.9984295,0.00042747543,0.00021631391,0.0002787221,0.0006394741,0.000008554723],"candidate_categories":["sts"],"consensus_categories":["sts"],"category_scores_codex":[0.0062642503,0.00008065771,0.00013876597,0.000096847754,0.0075239376,0.00027710665,0.0013725176,0.000013878036,0.0000070471033],"category_scores_gemma":[0.004079184,0.000040352515,0.000026002792,0.0016130797,0.0156084215,0.000699181,0.0012411068,0.00011114724,0.0000037758853],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007091804,0.0000790519,0.11003306,0.00005208545,0.000049039492,6.18245e-7,0.06522162,0.00000888523,0.011381277,0.8021829,0.0022477694,0.008672764],"study_design_scores_gemma":[0.00009063836,0.00001040104,0.730291,0.000023730214,0.000013500829,2.6285578e-8,0.21226043,0.000035131845,0.00046115386,0.05638632,0.00037869276,0.00004893871],"about_ca_topic_score_codex":0.0023555665,"about_ca_topic_score_gemma":0.0014188461,"teacher_disagreement_score":0.74579656,"about_ca_system_score_codex":0.000075173026,"about_ca_system_score_gemma":0.00009076288,"threshold_uncertainty_score":0.99376816},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","scholarly_communication"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2599750828","doi":"10.1002/asi.23758","title":"A webometric analysis of the online vaccination debate","year":2017,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Webometrics; World Wide Web; PageRank; Computer science; Information retrieval; Variety (cybernetics); Data science; Artificial intelligence","score_opus":0.015447240406642611,"score_gpt":0.28669971135156236,"score_spread":0.2712524709449197,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2599750828","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9268255,0.00004027194,0.039010044,0.030615974,0.00126449,0.000349512,0.000025104368,0.00002215804,0.0018468936],"genre_scores_gemma":[0.9982576,0.000028866312,0.0014917762,0.00016597255,0.0000085358915,0.0000010814829,2.4699216e-7,5.516802e-7,0.00004536411],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9986057,0.000013270735,0.00048581173,0.000053292813,0.0007233822,0.00011850505],"domain_scores_gemma":[0.99365896,0.00012490524,0.0030315458,0.00046319544,0.0027006683,0.000020733742],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.003589383,0.00004410026,0.0001500849,0.0022725123,0.0007138535,0.00029082457,0.0022500448,0.0000787049,6.2164963e-7],"category_scores_gemma":[0.015336323,0.000024426665,0.000108983324,0.007927842,0.000109975495,0.0046598716,0.00038431975,0.00012785471,4.920434e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010595041,0.00008819717,0.41696924,0.000026044654,0.0002947641,5.236983e-8,0.0011396279,0.000616276,0.0005323709,0.2500948,0.0008013203,0.3294267],"study_design_scores_gemma":[0.00039254807,0.00007457021,0.9323374,0.00000772051,0.000102217564,0.0000043511873,0.00012583622,0.049989108,0.0037570419,0.006550251,0.0066098925,0.0000490649],"about_ca_topic_score_codex":0.0000050484696,"about_ca_topic_score_gemma":0.000010956121,"teacher_disagreement_score":0.51536816,"about_ca_system_score_codex":0.00021471345,"about_ca_system_score_gemma":0.00025943082,"threshold_uncertainty_score":0.9929579},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2600432762","doi":"10.21815/jde.016.011","title":"Top‐Cited Articles in Problem‐Based Learning: A Bibliometric Analysis and Quality of Evidence Assessment","year":2017,"lang":"en","type":"review","venue":"Journal of Dental Education","topic":"Problem and Project Based Learning","field":"Social Sciences","cited_by":44,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Bibliometrics; Evidence-based medicine; Ranking (information retrieval); MEDLINE; Quality of evidence; Family medicine; Medicine; Psychology; Library science; Alternative medicine; Political science; Information retrieval; Meta-analysis; Computer science; Pathology","score_opus":0.3906450100511481,"score_gpt":0.5746661852518006,"score_spread":0.18402117520065248,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2600432762","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017523272,0.981063,0.00013463556,0.00012668753,0.00017624142,0.00032903661,0.0000017524634,0.0000047703115,0.00064059143],"genre_scores_gemma":[0.1025987,0.89515126,0.0018125592,0.0000078327785,0.00009996645,0.000012488552,0.000005365633,0.000007335804,0.00030447156],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99618906,0.0017146495,0.0010596665,0.00016790733,0.0006995225,0.00016919148],"domain_scores_gemma":[0.99514645,0.000855782,0.0033927583,0.00014096571,0.00034676158,0.00011727159],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0076529086,0.0001333915,0.0009978581,0.037438076,0.00019567039,0.00022202206,0.00033174132,0.00014913653,0.00008834373],"category_scores_gemma":[0.0028563985,0.0001018749,0.0003600288,0.039976962,0.00011018201,0.0005443302,0.000031861993,0.00051548035,0.0000018050393],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000008000974,0.00016729784,0.13324934,0.0017112838,0.000119264834,9.94972e-7,0.0007858005,0.000012259563,0.0000038179646,0.00005619693,0.000015597912,0.86387014],"study_design_scores_gemma":[0.000623239,0.0004883573,0.32550836,0.034353197,0.0044531603,0.000013905095,0.0049389806,0.00005411192,0.000010768774,0.00020272826,0.62873405,0.00061913044],"about_ca_topic_score_codex":0.0033779338,"about_ca_topic_score_gemma":0.0008911632,"teacher_disagreement_score":0.86325103,"about_ca_system_score_codex":0.00037325226,"about_ca_system_score_gemma":0.003967309,"threshold_uncertainty_score":0.9804288},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2600545932","doi":"10.21815/jde.016.006","title":"Do Standard Bibliometric Measures Correlate with Academic Rank of Full‐Time Pediatric Dentistry Faculty Members?","year":2017,"lang":"en","type":"article","venue":"Journal of Dental Education","topic":"Health and Medical Research Impacts","field":"Medicine","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Promotion (chess); Rank (graph theory); Index (typography); Bibliometrics; Accreditation; Sample (material); Medicine; Sample size determination; Psychology; Medical education; Statistics; Mathematics; Library science; Computer science; Political science","score_opus":0.11290308193609938,"score_gpt":0.4742496441281981,"score_spread":0.3613465621920987,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2600545932","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9922695,0.002865857,0.00012924471,0.0023178402,0.0008578798,0.00024435396,0.000017021113,0.0000048742886,0.001293423],"genre_scores_gemma":[0.99425226,0.0029742748,0.0005918043,0.00017758383,0.0009499433,0.0000026136963,0.000008232957,0.0000145997765,0.0010286869],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99688816,0.00006134067,0.00057411497,0.00011585736,0.0019886172,0.0003719252],"domain_scores_gemma":[0.9954944,0.00010916932,0.0010098068,0.00022757868,0.0009265923,0.0022324235],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.00204695,0.00011322556,0.00038007664,0.0052800397,0.00013441713,0.00005715851,0.0002611997,0.00015174155,0.00039182548],"category_scores_gemma":[0.01948462,0.00007642221,0.000103471895,0.0030532482,0.00011022232,0.00043749058,0.000032770935,0.00082200684,0.000045267207],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0066015553,0.0007714691,0.50928235,0.0011535137,0.00024222975,0.0002205556,0.00068954576,0.0000051162865,0.008453721,0.000014166782,0.1763928,0.29617295],"study_design_scores_gemma":[0.0073848665,0.0034743324,0.96054554,0.0012424091,0.0005778153,0.005024028,0.0010970708,0.000019514291,0.0022064894,0.0000600727,0.01818634,0.0001815045],"about_ca_topic_score_codex":0.000046214394,"about_ca_topic_score_gemma":0.0000030192164,"teacher_disagreement_score":0.4512632,"about_ca_system_score_codex":0.000321914,"about_ca_system_score_gemma":0.008070846,"threshold_uncertainty_score":0.99755245},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2609236485","doi":"10.22037/jps.v8i2.13762","title":"Scientometric Analysis of Medical and Non-Medical Highly Cited Papers of Iran in Essential Science Indicator (ESI)","year":2017,"lang":"en","type":"article","venue":"Journal of paramedical sciences.","topic":"Ophthalmology and Visual Health Research","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Subject (documents); Index (typography); Medical science; Science Citation Index; Citation; Descriptive statistics; Bibliometrics; Population; Rank (graph theory); Citation index; Library science; Medicine; Medical education; Geography; Statistics; Computer science; Mathematics; Environmental health","score_opus":0.1180031619362123,"score_gpt":0.552923031496838,"score_spread":0.43491986956062567,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2609236485","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99179995,0.0004526566,0.0000730854,0.006378027,0.0003395025,0.0001308453,0.0000027308974,0.000002291698,0.0008209047],"genre_scores_gemma":[0.99878985,0.0005732624,0.00028502155,0.00020801295,0.00012253843,0.0000014422114,6.135024e-7,0.00000385282,0.000015394924],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98873127,0.00014303412,0.0012453386,0.00037690534,0.008870492,0.00063295965],"domain_scores_gemma":[0.99554443,0.0008988867,0.0009815423,0.00033327303,0.00031422032,0.0019276432],"candidate_categories":["metaresearch","sts"],"consensus_categories":[],"category_scores_codex":[0.022292439,0.00013015201,0.0009870623,0.0064332443,0.0003880092,0.000052421667,0.0018291187,0.00036550887,0.00087968295],"category_scores_gemma":[0.02468718,0.00008512487,0.00017906976,0.009211412,0.01690416,0.00028332244,0.00039118546,0.0010414622,0.0000024651952],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021432406,0.000769807,0.9839263,0.0001327226,0.00012163799,0.00073780207,0.00027258275,0.0000028963993,0.0011427932,0.0002800052,0.00007598944,0.0123231225],"study_design_scores_gemma":[0.0020508382,0.0013260683,0.98605126,0.0003862558,0.00016923904,0.00034144174,0.00019296496,0.00870874,0.00060081424,0.00005000444,0.000048106955,0.000074269126],"about_ca_topic_score_codex":0.00042621407,"about_ca_topic_score_gemma":0.00003658899,"teacher_disagreement_score":0.01651615,"about_ca_system_score_codex":0.000091403104,"about_ca_system_score_gemma":0.005162762,"threshold_uncertainty_score":0.98577124},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2610741382","doi":"10.1016/j.wneu.2017.04.157","title":"The Spectrum of Altmetrics in Neurosurgery: The Top 100 “Trending” Articles in Neurosurgical Journals","year":2017,"lang":"en","type":"review","venue":"World Neurosurgery","topic":"Social Media in Health Education","field":"Social Sciences","cited_by":98,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Western Hospital; Hospital for Sick Children; University of Toronto","funders":"","keywords":"Altmetrics; Social media; Medicine; Neurosurgery; Citation; Mainstream; Bibliometrics; Outreach; Medical education; Library science; World Wide Web; Surgery; Computer science","score_opus":0.41313739206947037,"score_gpt":0.49475298336685625,"score_spread":0.08161559129738588,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2610741382","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004219275,0.9585164,2.2868701e-7,0.011234188,0.022400232,0.0021478762,0.00001683648,0.0000452414,0.0014197085],"genre_scores_gemma":[0.010513924,0.9865433,0.0000020940186,0.00021707483,0.0016853515,0.0005074731,0.0000034831166,0.0000734612,0.0004538512],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.98912185,0.0050779493,0.0024998053,0.00063665526,0.0013608763,0.0013028752],"domain_scores_gemma":[0.94346315,0.05267152,0.0024177094,0.0010813382,0.00008517458,0.00028111218],"candidate_categories":["metaresearch","metaepi_narrow","sts"],"consensus_categories":[],"category_scores_codex":[0.011126484,0.00044985578,0.0019069862,0.0020037729,0.001379189,0.0006141363,0.0016534901,0.00029602612,0.00008678757],"category_scores_gemma":[0.04069273,0.00030177413,0.0007207623,0.0067774043,0.0010821817,0.00027231761,0.00019569558,0.0019298843,0.000029952753],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016494403,0.00015608573,0.047519986,0.0005016867,0.0000109047,0.000087798355,0.0008503961,0.0000017631515,1.6429202e-7,0.0016449938,0.002096913,0.9471128],"study_design_scores_gemma":[0.00007796227,0.000011146001,0.03832578,0.0018296548,0.000051065046,0.0000060736897,0.000072255236,0.0000020220298,7.4908996e-7,0.00039040964,0.9589963,0.00023661164],"about_ca_topic_score_codex":0.0017016451,"about_ca_topic_score_gemma":0.006112929,"teacher_disagreement_score":0.95689934,"about_ca_system_score_codex":0.0003836448,"about_ca_system_score_gemma":0.001954194,"threshold_uncertainty_score":0.99994344},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2620664699","doi":"10.1017/cjn.2017.84","title":"C.06 Retraction of scientific publications in neurosurgery","year":2017,"lang":"en","type":"article","venue":"Canadian Journal of Neurological Sciences / Journal Canadien des Sciences Neurologiques","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"Systems, Applications & Products in Data Processing (Canada); Toronto Public Health","funders":"","keywords":"Impact factor; Neurosurgery; MEDLINE; Medicine; Data extraction; Misattribution of memory; Scientific misconduct; Family medicine; Alternative medicine; Surgery; Political science; Psychiatry; Pathology; Law","score_opus":0.0743255195744261,"score_gpt":0.32388761057128634,"score_spread":0.24956209099686025,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2620664699","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95754385,0.00033653184,0.000017255988,0.018608224,0.002478144,0.00013389469,0.000011455493,0.000011738928,0.020858899],"genre_scores_gemma":[0.9976512,0.00041632628,0.00059916003,0.0007384767,0.00037029252,0.0000017303165,3.0598815e-7,0.0000065033983,0.00021599873],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.994277,0.0016002806,0.0010770574,0.00048270856,0.0012585281,0.0013044483],"domain_scores_gemma":[0.99444425,0.0010713193,0.0018711973,0.00027046766,0.0009410189,0.0014017449],"candidate_categories":["metaresearch","sts","scholarly_communication","research_integrity"],"consensus_categories":["sts"],"category_scores_codex":[0.021140413,0.00022216662,0.0004720764,0.001765098,0.0111190835,0.0023160062,0.004193152,0.0005685617,0.00023735019],"category_scores_gemma":[0.023693027,0.00016805455,0.00024189978,0.0015104507,0.023200398,0.0032997478,0.00007987393,0.0029729102,0.0000049016658],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016974827,0.000029008002,0.9787042,0.0000034881327,0.0000041912817,0.0005269606,0.0013474901,0.0002686875,0.000048446003,0.014116363,0.0021761814,0.002758012],"study_design_scores_gemma":[0.00026281408,0.006469312,0.86308455,0.0000887334,0.000026643156,0.0026462309,0.0014333695,0.00047040018,0.00012272227,0.07035516,0.05466868,0.0003714104],"about_ca_topic_score_codex":0.014873269,"about_ca_topic_score_gemma":0.32834655,"teacher_disagreement_score":0.31347328,"about_ca_system_score_codex":0.00026757235,"about_ca_system_score_gemma":0.007880698,"threshold_uncertainty_score":0.99932724},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"evaluation","study_design":"systematic_review","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics","scholarly_communication","research_integrity"],"domain":"evaluation","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2621304307","doi":"","title":"Bibliometric Indicators For Environment Canada’s First Priority Area: Reducing The Impacts Of Contaminants And Other Environmental Stressors On The Natural Environment","year":2014,"lang":"en","type":"article","venue":"","topic":"Water Quality and Pollution Assessment","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Natural (archaeology); Stressor; Environmental planning; Environmental science; Environmental resource management; Contamination; Environmental protection; Environmental impact assessment; Environmental monitoring; Geography; Environmental engineering; Ecology","score_opus":0.015898302031842754,"score_gpt":0.23239820677147816,"score_spread":0.2164999047396354,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2621304307","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9949458,0.0000855334,0.00019406705,0.003160765,0.000101929894,0.0007466075,0.00012503231,0.000007818477,0.0006323937],"genre_scores_gemma":[0.99817324,0.00010996101,0.0001303182,0.0010491331,0.000027934082,0.00004937012,0.000006882557,0.000016455335,0.00043671764],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99796104,0.00018273004,0.00036679464,0.00038638426,0.0007067189,0.00039634344],"domain_scores_gemma":[0.9985224,0.0005921379,0.00026388172,0.00047636952,8.496257e-7,0.00014434099],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0009554294,0.0002451116,0.0002144072,0.00048849056,0.00045499307,0.000034139368,0.00035394615,0.00006355643,0.0014746865],"category_scores_gemma":[0.000052654235,0.00013252976,0.00006977195,0.00084587774,0.00047928598,0.000084076935,0.0002405665,0.00017586276,0.000026372361],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020977447,0.0009692962,0.9164856,0.000078669815,0.0002710501,0.0000027265874,0.0040818956,0.0037607937,0.008709081,0.00281706,0.019532388,0.043081664],"study_design_scores_gemma":[0.0005795505,0.00019453435,0.85539466,0.000017399587,0.00003652986,0.0000032105922,0.00048649215,0.0010739168,0.011848051,0.000071157185,0.13004854,0.00024594535],"about_ca_topic_score_codex":0.08324447,"about_ca_topic_score_gemma":0.03382864,"teacher_disagreement_score":0.11051615,"about_ca_system_score_codex":0.0005342692,"about_ca_system_score_gemma":0.000023349356,"threshold_uncertainty_score":0.9994381},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W2621787917","doi":"10.1017/s1742170517000308","title":"Who's afraid of Allan Savory? Scientometric polarization on Holistic Management as competing understandings","year":2017,"lang":"en","type":"article","venue":"Renewable Agriculture and Food Systems","topic":"Forest Management and Policy","field":"Environmental Science","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto and Region Conservation Authority; Dalhousie University","funders":"","keywords":"Terminology; Sustainability; Sociology; Sustainability science; Scientometrics; Ecology; Social science; Philosophy","score_opus":0.02840076346361369,"score_gpt":0.24843246126632343,"score_spread":0.22003169780270973,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2621787917","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.42518342,0.00025701828,0.00023000214,0.00025027842,0.00031631612,0.00057645736,0.000013032342,0.000040472973,0.573133],"genre_scores_gemma":[0.97861326,0.00012430534,0.000039528422,0.000055035227,0.000075153825,0.000007899681,0.0000130953385,0.0000080119935,0.021063706],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99879706,0.0000334208,0.00020184359,0.00029410343,0.00042042593,0.00025316936],"domain_scores_gemma":[0.999279,0.000023057673,0.00027679274,0.0003126089,0.000011784697,0.00009672727],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002984276,0.00015268126,0.00019387968,0.00016084492,0.0005181585,0.00029607592,0.00032532075,0.0000671336,0.000054984524],"category_scores_gemma":[0.000032385436,0.000107179745,0.000040921113,0.0004571774,0.00013437934,0.00021410687,0.00022168353,0.000053068274,0.00008514598],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017403733,0.0007459845,0.3937531,0.0034246198,0.00082500355,0.00008041839,0.0048951167,0.04631979,0.012867151,0.35889167,0.1731146,0.0049084863],"study_design_scores_gemma":[0.004776953,0.0030687193,0.6147285,0.0020217246,0.00056421076,0.00007278316,0.005198086,0.004878063,0.0016277207,0.0033581923,0.35753345,0.0021715786],"about_ca_topic_score_codex":0.002452674,"about_ca_topic_score_gemma":0.0001601084,"teacher_disagreement_score":0.55342984,"about_ca_system_score_codex":0.000092616516,"about_ca_system_score_gemma":0.0000025794848,"threshold_uncertainty_score":0.43706632},"labels":[{"model":"gemma","categories":["bibliometrics","sts"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2625069236","doi":"10.19255/jmpm01304","title":"Building Information Modeling (BIM) in Brazil's Architecture, Engineering and Construction (AEC) Industry: A Review and a Bibliometric Study","year":2017,"lang":"en","type":"review","venue":"Swinburne Research Bank (Swinburne University of Technology)","topic":"BIM and Construction Integration","field":"Engineering","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Trois-Rivières","funders":"","keywords":"Building information modeling; Scope (computer science); Interoperability; Architecture; Process (computing); Engineering management; Engineering; Construction industry; Work (physics); Sustainability; Construction engineering; Facility management; Architectural engineering; Process management; Knowledge management; Systems engineering; Computer science; Business; Operations management; World Wide Web","score_opus":0.055250643404401494,"score_gpt":0.3341857389141401,"score_spread":0.2789350955097386,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2625069236","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0053109853,0.98976254,0.0022923993,0.00011244435,0.000110376524,0.0017232463,0.00003104795,0.0002736025,0.00038336447],"genre_scores_gemma":[0.005604127,0.99107504,0.0032056693,0.0000020506284,0.000027536737,0.000017357313,0.000014234295,0.00003209075,0.000021872338],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977853,0.000108368215,0.000673458,0.0004461283,0.00045672318,0.00053007767],"domain_scores_gemma":[0.99855375,0.00011031294,0.00024640607,0.00063661195,0.00031436945,0.00013853991],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.0013342986,0.00043605862,0.0014805574,0.047856566,0.00026985168,0.00011537618,0.0006638539,0.0011393285,0.000031963504],"category_scores_gemma":[0.00057223445,0.0004716958,0.00013092872,0.01872117,0.00041078377,0.00082391495,0.00047046912,0.0034221595,0.0000040621567],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000051646007,0.00002169809,0.00015284662,0.010742408,0.00017306613,0.000016974953,0.000049422124,0.00019063413,0.0000023564526,0.00084127526,0.00007634301,0.9877278],"study_design_scores_gemma":[0.0020078484,0.00040140684,0.00017946804,0.05007628,0.0011211014,0.00068983797,0.0016010536,0.011648948,0.000010043039,0.0016245146,0.92930824,0.0013312382],"about_ca_topic_score_codex":0.00021931261,"about_ca_topic_score_gemma":0.00004562113,"teacher_disagreement_score":0.98639655,"about_ca_system_score_codex":0.00031247595,"about_ca_system_score_gemma":0.00019507484,"threshold_uncertainty_score":0.9997735},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2700484896","doi":"10.29173/cais188","title":"Science in the European Union – “Before and After” A Scientometric Approach to Measure the Structural Transformation of Science in Central and Eastern Europe after the Political-Economical Changes","year":2013,"lang":"fr","type":"article","venue":"Proceedings of the Annual Conference of CAIS / Actes du congrès annuel de l ACSI","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Humanities; European union; Scientometrics; Political science; Politics; Geography; Regional science; Sociology; Economics; Social science; Philosophy; Law; International trade","score_opus":0.12207664046591275,"score_gpt":0.3635625514400739,"score_spread":0.2414859109741611,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2700484896","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9725176,0.0005668544,0.000040171264,0.020309204,0.00023629163,0.0011158854,0.0000897228,0.000003902585,0.0051203794],"genre_scores_gemma":[0.9992328,0.000122193,0.00012060731,0.0003018329,0.000052230393,0.00003469255,3.566602e-7,0.000010603004,0.00012468407],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9926071,0.0003914129,0.0008677467,0.00073619286,0.0041091675,0.001288376],"domain_scores_gemma":[0.96419066,0.0008017495,0.0004943428,0.00044615747,0.033617888,0.00044919766],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.028088644,0.00027971782,0.00041935287,0.014206829,0.000401752,0.013551457,0.0065910667,0.000088808345,0.000013249684],"category_scores_gemma":[0.0537518,0.0001311883,0.00008148696,0.086914495,0.009692587,0.010548292,0.0021394473,0.0005285046,0.0000029645141],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009817545,0.00016776814,0.7352747,0.00018244327,0.000009124275,9.3589233e-7,0.08696652,0.000011782789,0.0026932606,0.021486819,0.000052404466,0.15305607],"study_design_scores_gemma":[0.0003437346,0.00031991772,0.97188514,0.00015763103,0.00001565537,0.00004118639,0.015361896,0.008005656,0.00076472585,0.0021977886,0.000751044,0.00015562901],"about_ca_topic_score_codex":0.00092752435,"about_ca_topic_score_gemma":0.00012889295,"teacher_disagreement_score":0.23661044,"about_ca_system_score_codex":0.00014361106,"about_ca_system_score_gemma":0.00067050464,"threshold_uncertainty_score":0.99878377},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2726750016","doi":"","title":"La collaboration dans le secteur spatial canadien - une approche exploratoire combinant bibliométrie et sondages","year":2016,"lang":"fr","type":"article","venue":"PolyPublie (École Polytechnique de Montréal)","topic":"Regional Development and Policy","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Mathematics; Sociology","score_opus":0.015591350671284261,"score_gpt":0.25891027680017736,"score_spread":0.2433189261288931,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2726750016","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.36374047,0.004189394,0.26298198,0.3447025,0.0014406518,0.0017659331,0.00043568105,0.00076202163,0.019981349],"genre_scores_gemma":[0.9346442,0.006552327,0.005747922,0.0020098493,0.0005581792,0.0005144836,0.00007441503,0.00011337765,0.04978521],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99501824,0.0014740368,0.000670936,0.000680451,0.00085134205,0.0013050067],"domain_scores_gemma":[0.9973597,0.00052426255,0.00044363385,0.0005649738,0.0003709625,0.0007364693],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0020898997,0.00055081665,0.00053460064,0.0012865055,0.0010852375,0.00046650233,0.0007964894,0.00079288555,0.00008112043],"category_scores_gemma":[0.0004893726,0.00048424886,0.00019594358,0.0061804047,0.0011169271,0.0015247627,0.00021320881,0.00045181636,0.000100203106],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012873534,0.0004840283,0.022419212,0.00007609434,0.00013659081,0.00023075254,0.030819496,0.00006052698,0.0076993294,0.73670477,0.07353803,0.12770246],"study_design_scores_gemma":[0.0017647145,0.0002974489,0.048855167,0.0005384341,0.00008181682,0.00010422686,0.008264634,0.0010592153,0.00851596,0.029682254,0.899522,0.0013140964],"about_ca_topic_score_codex":0.5002662,"about_ca_topic_score_gemma":0.58655715,"teacher_disagreement_score":0.825984,"about_ca_system_score_codex":0.0016086255,"about_ca_system_score_gemma":0.00472932,"threshold_uncertainty_score":0.9997609},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W2736387275","doi":"10.1016/j.joi.2017.07.003","title":"DataCite as a novel bibliometric source: Coverage, strengths and limitations","year":2017,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Research Data Management Practices","field":"Computer Science","cited_by":34,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Ministerio de Economía y Competitividad; European Commission","keywords":"Bibliometrics; Open data; Value (mathematics); Scientometrics; Production (economics); Open source","score_opus":0.14822043850723518,"score_gpt":0.4057818156424827,"score_spread":0.2575613771352475,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2736387275","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03325908,0.00036496116,0.9441281,0.0016219808,0.00035933344,0.0001113121,0.00001666338,0.000014507123,0.020124054],"genre_scores_gemma":[0.82623017,0.046298936,0.12259447,0.0005813465,0.00020848164,0.0000026919965,0.0000071165055,0.000014856067,0.004061934],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978737,0.000025213996,0.0004853805,0.00015361427,0.0011969991,0.00026508357],"domain_scores_gemma":[0.99396294,0.0027228063,0.0014331597,0.0010407802,0.00059207436,0.00024824045],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics","scholarly_communication"],"category_scores_codex":[0.0024860909,0.00010946318,0.0001828057,0.033697996,0.00044079038,0.017304137,0.003552781,0.000047204347,0.0000035634862],"category_scores_gemma":[0.07102973,0.00008948933,0.000059189486,0.021948336,0.0000723594,0.11098276,0.0015643623,0.0003776359,0.0000134537895],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021145494,0.00014688088,0.0060075573,0.00004342084,0.00012908781,0.000065248154,0.00011980323,0.00011259496,0.000041747215,0.062392987,0.00411018,0.9268094],"study_design_scores_gemma":[0.0014089049,0.0005580144,0.14347853,0.000038102527,0.000031892243,0.00027052243,0.00010735784,0.0046061426,0.000079904356,0.00059218315,0.84863436,0.00019407908],"about_ca_topic_score_codex":0.000032405187,"about_ca_topic_score_gemma":0.0000045038487,"teacher_disagreement_score":0.92661524,"about_ca_system_score_codex":0.00004646216,"about_ca_system_score_gemma":0.00019534807,"threshold_uncertainty_score":0.99884063},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2746253778","doi":"10.1186/s12961-017-0241-5","title":"Health policy and systems research collaboration pathways: lessons from a network science analysis","year":2017,"lang":"en","type":"article","venue":"Health Research Policy and Systems","topic":"Global Health and Surgery","field":"Medicine","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Alliance for Health Policy and Systems Research","keywords":"Health services research; Global health; Health policy; Social connectedness; Citation; Political science; Health informatics; Public health; Economic growth; Medicine; Economics; Psychology; Nursing","score_opus":0.4682911180968649,"score_gpt":0.5992208265458154,"score_spread":0.13092970844895052,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2746253778","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5653295,0.065903105,0.00012273043,0.3486399,0.0010877054,0.0071008285,0.0009533217,0.0001536111,0.010709294],"genre_scores_gemma":[0.98081785,0.012395239,0.00005105854,0.0013537657,0.0039289813,0.00023819995,0.00006008406,0.00002919684,0.0011256292],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9865411,0.0038445995,0.0011704083,0.0010981393,0.0033248954,0.00402085],"domain_scores_gemma":[0.98876697,0.0019808572,0.00050759653,0.0017308074,0.00248197,0.0045317984],"candidate_categories":["metaresearch","sts","scholarly_communication"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.049799535,0.00026277505,0.0014070721,0.0029660359,0.009618187,0.0017201947,0.00047432067,0.00024345871,0.0000030813164],"category_scores_gemma":[0.011388227,0.00022173814,0.00007528152,0.0062267384,0.0018645087,0.00041880147,0.00043145262,0.0011297356,0.000028382063],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015548593,0.00047109794,0.13667446,0.011745016,0.00052107853,0.00017888712,0.012328983,0.00022823067,0.00018883958,0.6929063,0.099948645,0.04325359],"study_design_scores_gemma":[0.0040097097,0.003475856,0.6593528,0.0061925007,0.000061034436,0.00022740292,0.017727468,0.05745628,0.000009723459,0.004133169,0.24664676,0.0007072942],"about_ca_topic_score_codex":0.51700526,"about_ca_topic_score_gemma":0.009257981,"teacher_disagreement_score":0.68877316,"about_ca_system_score_codex":0.00255576,"about_ca_system_score_gemma":0.03181797,"threshold_uncertainty_score":0.9993161},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2747001800","doi":"10.12688/f1000research.12404.1","title":"Exploring Vietnamese co-authorship patterns in social sciences with basic network measures of 2008-2017 Scopus data","year":2017,"lang":"en","type":"preprint","venue":"F1000Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Vietnamese; Clustering coefficient; Scopus; Cluster analysis; Biology; Computer science; Artificial intelligence; MEDLINE","score_opus":0.9720149141063179,"score_gpt":0.6744967559506466,"score_spread":0.2975181581556713,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2747001800","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9678401,0.004605644,0.0017479811,0.0029520916,0.001810097,0.0021143525,0.0017272239,0.0000627729,0.017139735],"genre_scores_gemma":[0.9937695,0.0026489794,0.001085905,0.000025754302,0.0010036222,0.00017006086,0.0001099987,0.000058245387,0.00112788],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9551693,0.0026136225,0.0017223207,0.0034522307,0.034381773,0.002660747],"domain_scores_gemma":[0.98084885,0.006467498,0.0012458588,0.0062350626,0.004501865,0.00070084387],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","bibliometrics","open_science"],"category_scores_codex":[0.14956689,0.0005564939,0.0014759494,0.06429328,0.0019218588,0.008801022,0.033463318,0.00051907497,0.00046497188],"category_scores_gemma":[0.05174103,0.00039163252,0.00024657673,0.07765008,0.0027109035,0.0022037562,0.020681629,0.0032457388,0.00020253233],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002974081,0.00030850235,0.82002753,0.00020001957,0.00008001552,0.0002518448,0.0008951875,0.0015162226,0.000071458024,0.00030865654,0.062858306,0.11318485],"study_design_scores_gemma":[0.0013132556,0.0005633878,0.95071,0.0010862728,0.000025756846,0.00001606947,0.0013985808,0.014408349,0.0004614485,0.015544175,0.0134032555,0.0010694661],"about_ca_topic_score_codex":0.0032105052,"about_ca_topic_score_gemma":0.001761803,"teacher_disagreement_score":0.13068245,"about_ca_system_score_codex":0.0003104286,"about_ca_system_score_gemma":0.0043092314,"threshold_uncertainty_score":0.99985355},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2749249904","doi":"10.1108/ajim-01-2017-0024","title":"Incorporating data sharing to the reward system of science","year":2017,"lang":"en","type":"article","venue":"Aslib Journal of Information Management","topic":"Research Data Management Practices","field":"Computer Science","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Institutes of Health","keywords":"Computer science; Similarity (geometry); Set (abstract data type); Data science; Originality; Data sharing; Information retrieval; Value (mathematics); Data set; Scientometrics; Scale (ratio); World Wide Web; Psychology; Artificial intelligence","score_opus":0.13976620657204658,"score_gpt":0.3809386428785568,"score_spread":0.2411724363065102,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2749249904","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027888883,0.000013325257,0.9571408,0.006557424,0.0005656938,0.0003435592,0.0000043541204,0.000016962838,0.032568995],"genre_scores_gemma":[0.8547153,0.000059422553,0.14483507,0.00021676967,0.000056864425,0.0000038699027,0.000002608324,0.0000028024585,0.00010729546],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9973874,0.000039428654,0.00072690984,0.00015922917,0.001468859,0.00021815086],"domain_scores_gemma":[0.99366057,0.000048323527,0.0021243184,0.003628525,0.0004347557,0.000103491926],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.014029949,0.00007537395,0.00012253836,0.00059193146,0.0008094053,0.0065349406,0.024012676,0.000010209557,0.0000013094368],"category_scores_gemma":[0.0011549682,0.000050807907,0.000026367572,0.00052751246,0.00013273911,0.11337294,0.013156379,0.00013552417,0.000033595275],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019341476,0.000018420767,0.0008884902,0.00029274976,0.0000857898,0.000020300824,0.0006875512,0.0024805686,0.000033122684,0.858923,0.004789569,0.13176109],"study_design_scores_gemma":[0.0021775106,0.0005838192,0.10990025,0.0020780645,0.00014084236,0.00015248454,0.014840038,0.42922646,0.0015165911,0.0030329505,0.43569425,0.00065672817],"about_ca_topic_score_codex":0.000029293355,"about_ca_topic_score_gemma":0.0000035427427,"teacher_disagreement_score":0.85589004,"about_ca_system_score_codex":0.00009117433,"about_ca_system_score_gemma":0.00008405392,"threshold_uncertainty_score":0.994825},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics","open_science"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2751853820","doi":"10.21083/partnership.v12i1.3906","title":"Research Impact Metrics: A Faculty Perspective","year":2017,"lang":"en","type":"article","venue":"Partnership The Canadian Journal of Library and Information Practice and Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Discipline; Library science; Perspective (graphical); Sociology; Psychology; Social science; Computer science","score_opus":0.7298564952173201,"score_gpt":0.6544215829644491,"score_spread":0.07543491225287102,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2751853820","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.23787731,0.006239104,0.000060823368,0.30678838,0.00033956484,0.00057537004,0.000073357274,0.00000616722,0.44803992],"genre_scores_gemma":[0.99545485,0.0019235916,0.00018254477,0.0004451331,0.0002040722,0.000003777428,0.00000213055,0.00000583965,0.001778072],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9886036,0.0021244574,0.00068997464,0.00020985198,0.007528841,0.00084330427],"domain_scores_gemma":[0.96404773,0.019778002,0.00083401415,0.00091592723,0.012128504,0.0022958147],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.10161031,0.00010896747,0.00022655747,0.03879334,0.005858558,0.03357773,0.00297349,0.00014811326,0.0003368008],"category_scores_gemma":[0.21332103,0.000056040717,0.000098449476,0.029847268,0.0017405814,0.05452465,0.0006264305,0.0020986816,0.00011837029],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006700682,0.0000685202,0.096745625,0.000026365795,0.00019833776,0.00018526879,0.017088741,0.000031159292,0.000009282867,0.44930574,0.28617615,0.14949474],"study_design_scores_gemma":[0.0004927839,0.0006112872,0.14250568,0.000023592565,0.000007880345,0.00028494903,0.052164923,0.0007078101,0.00007836379,0.03181339,0.7712012,0.0001081575],"about_ca_topic_score_codex":0.0047852797,"about_ca_topic_score_gemma":0.00028952866,"teacher_disagreement_score":0.75757754,"about_ca_system_score_codex":0.00014139462,"about_ca_system_score_gemma":0.0045615905,"threshold_uncertainty_score":0.99543566},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2762227057","doi":"10.1016/j.joi.2017.08.008","title":"Mapping science using Library of Congress Subject Headings","year":2017,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":15,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Subject (documents); Library of congress; Computer science; Information retrieval; Library science; Library of Congress Classification; Data science; Library classification","score_opus":0.6384368990329121,"score_gpt":0.5833211230820153,"score_spread":0.055115775950896784,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2762227057","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9748696,0.0005421247,0.0044210674,0.00024239764,0.0020624148,0.00009437589,0.000014974763,0.0000044728304,0.01774855],"genre_scores_gemma":[0.98688066,0.00053864165,0.011858002,0.00006332413,0.00013767152,1.7040603e-7,1.5265884e-7,0.000008588626,0.00051281543],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98470527,0.00004818306,0.0018066365,0.00025140063,0.012602639,0.0005858891],"domain_scores_gemma":[0.9825504,0.003134933,0.0044447468,0.0010091577,0.008259776,0.00060102943],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.036379527,0.0001379398,0.0005627316,0.15875909,0.0008683075,0.0065993713,0.008046128,0.000102552665,0.00016132419],"category_scores_gemma":[0.16882387,0.000090465175,0.00028915354,0.18151087,0.0010716353,0.009738696,0.0016195731,0.0004331194,0.000019386485],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000055519893,0.00009712867,0.73172945,0.00002332576,0.000027484037,0.000053145643,0.0002860258,0.00012497231,0.0062505947,0.0008795857,0.005254867,0.2552179],"study_design_scores_gemma":[0.0019365711,0.00087186147,0.83449596,0.00022243733,0.00001910701,0.0004167346,0.0020466389,0.019352091,0.067934565,0.0065845875,0.06565077,0.00046868995],"about_ca_topic_score_codex":0.000038306513,"about_ca_topic_score_gemma":6.147794e-7,"teacher_disagreement_score":0.2547492,"about_ca_system_score_codex":0.0001480388,"about_ca_system_score_gemma":0.0018544471,"threshold_uncertainty_score":0.99732083},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2765492995","doi":"10.1097/md.0000000000008247","title":"Bibliometric analysis of the top-cited articles on islet transplantation","year":2017,"lang":"en","type":"review","venue":"Medicine","topic":"Pancreatic function and diabetes","field":"Medicine","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Science Citation Index; Impact factor; Citation; Transplantation; Bibliometrics; Publishing; Library science; MEDLINE; Citation analysis; Family medicine; Internal medicine; Political science; Law","score_opus":0.1421508592924145,"score_gpt":0.40333522401638167,"score_spread":0.26118436472396717,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2765492995","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0051532798,0.9881922,0.00002848633,0.000504077,0.0003168268,0.00046255768,0.00005294093,0.00002402995,0.005265592],"genre_scores_gemma":[0.004105866,0.9937618,0.000014312404,0.00028922752,0.00021070207,0.000020299087,0.0002724811,0.00001620156,0.0013091479],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983907,0.00010409508,0.0005620489,0.00022941329,0.0005656441,0.00014810011],"domain_scores_gemma":[0.9979426,0.00056027673,0.0004872167,0.0008095675,0.00010653712,0.0000938012],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00045783742,0.00021322082,0.0020289843,0.05933384,0.00006635459,0.0000074988748,0.0001786795,0.00014761083,0.00083211664],"category_scores_gemma":[0.0012739302,0.00009920871,0.0005227826,0.092795536,0.00016584201,0.000022289261,0.0000117396385,0.00021051237,0.00002774879],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000030716055,0.00007448597,0.0022216865,0.0053196386,0.002133615,0.0000083421255,0.00006308793,8.138892e-7,0.000010302679,0.000058020458,0.0020619174,0.9880174],"study_design_scores_gemma":[0.0010596896,0.0002570702,0.051778432,0.015442967,0.057680316,0.000008114955,0.000012687034,0.0000458441,0.000026237893,0.000008709066,0.87356216,0.00011774889],"about_ca_topic_score_codex":0.00005952731,"about_ca_topic_score_gemma":0.000002023353,"teacher_disagreement_score":0.9878996,"about_ca_system_score_codex":0.000040046623,"about_ca_system_score_gemma":0.000081826525,"threshold_uncertainty_score":0.9513278},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2766781696","doi":"10.2214/ajr.17.18077","title":"Bibliometric Analysis of Manuscript Characteristics That Influence Citations: A Comparison of Six Major Radiology Journals","year":2017,"lang":"en","type":"article","venue":"American Journal of Roentgenology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":51,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vancouver General Hospital","funders":"","keywords":"Medicine; Bibliometrics; Medical physics; Library science; Radiology; Computer science","score_opus":0.42414929523449185,"score_gpt":0.5606304783726436,"score_spread":0.13648118313815172,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2766781696","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9882716,0.002509268,0.007293823,0.0010204121,0.00054941943,0.0000978995,0.000083942556,0.0000029380394,0.00017071399],"genre_scores_gemma":[0.9950147,0.002077088,0.002705326,0.00008074466,0.00004696755,0.0000018080843,0.000002103711,0.000010435794,0.00006078772],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99002886,0.00078927504,0.0031144244,0.00041243964,0.005028821,0.0006261798],"domain_scores_gemma":[0.9700585,0.0053510945,0.01325647,0.0012910281,0.009605418,0.00043750813],"candidate_categories":["metaresearch","bibliometrics","open_science"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.014930036,0.00020599544,0.002394904,0.64514935,0.00030509802,0.0005993518,0.0054280297,0.00012517473,0.00037823533],"category_scores_gemma":[0.069086805,0.00014889505,0.0006859102,0.5675114,0.0021594935,0.0007978675,0.00059121416,0.000540674,0.000015841055],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001242546,0.00021948326,0.8344164,0.00000466208,0.00091070135,0.000027633896,0.00022576949,0.00023747857,0.003685457,0.00008544443,0.00054231484,0.15952042],"study_design_scores_gemma":[0.00049931515,0.0017302525,0.9924628,0.000016809343,0.00041301208,0.00008639288,0.0013341485,0.00095841096,0.0010110439,0.0005654037,0.00080366584,0.00011873544],"about_ca_topic_score_codex":0.0003047899,"about_ca_topic_score_gemma":0.00003814633,"teacher_disagreement_score":0.15940168,"about_ca_system_score_codex":0.00009587316,"about_ca_system_score_gemma":0.00032115667,"threshold_uncertainty_score":0.9999531},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2766919312","doi":"10.1136/bmjopen-2016-015317","title":"Global research trends in spinal ultrasound: a systematic bibliometric analysis","year":2017,"lang":"en","type":"article","venue":"BMJ Open","topic":"Scoliosis diagnosis and treatment","field":"Medicine","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Science and Technology Commission of Shanghai Municipality; Changhai Hospital of Shanghai; Second Military Medical University; National Natural Science Foundation of China","keywords":"Medicine; Citation; Web of science; Idiopathic scoliosis; Bibliometrics; Scoliosis; Library science; Meta-analysis; Surgery; Pathology","score_opus":0.37060356855213966,"score_gpt":0.5884811362109733,"score_spread":0.21787756765883365,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2766919312","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9275171,0.00061765476,0.0000040480136,0.0017744371,0.00007203287,0.0052827317,0.00001729011,0.00001360026,0.06470112],"genre_scores_gemma":[0.996114,0.000047782913,0.00020411533,0.00007978826,0.00005857593,0.0025052512,0.000015806914,0.000008815696,0.00096586795],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99778664,0.00023369219,0.00043588204,0.00038966807,0.00077037246,0.00038374963],"domain_scores_gemma":[0.99769425,0.00016907019,0.00019732359,0.00156896,0.00017525245,0.00019513792],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0033867776,0.00013417494,0.00083329837,0.04724749,0.00025635306,0.0008828676,0.00078476715,0.000075704025,0.00037595467],"category_scores_gemma":[0.0017874823,0.000096170326,0.00020240589,0.11150722,0.00006960146,0.00023480179,0.00039230022,0.00010600961,0.00017592969],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021187247,0.00064379897,0.9909162,0.00085645384,0.00084950775,0.0003251711,0.00003809942,0.0000024088029,0.000002798611,0.00039173904,0.0021027825,0.0036591482],"study_design_scores_gemma":[0.0019311288,0.00046053922,0.99334526,0.0029843138,0.00077423226,0.000051984942,0.00013428213,0.000054329365,0.000019569552,0.00010629686,0.000038623675,0.00009940946],"about_ca_topic_score_codex":0.0077112503,"about_ca_topic_score_gemma":0.0044989167,"teacher_disagreement_score":0.06859691,"about_ca_system_score_codex":0.0004038709,"about_ca_system_score_gemma":0.00012806497,"threshold_uncertainty_score":0.9988965},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2775582453","doi":"","title":"Exploring the Music Library Association Mailing List: A Text Mining Approach","year":2017,"lang":"en","type":"article","venue":"Research Commons (University of Waikato)","topic":"Music and Audio Processing","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"University of Illinois at Urbana-Champaign; University of Hong Kong","keywords":"Association (psychology); Computer science; Information retrieval; Association rule learning; World Wide Web; Data science; Data mining; Psychology","score_opus":0.27893830423254173,"score_gpt":0.2931727871556954,"score_spread":0.014234482923153657,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2775582453","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8545045,0.0001306986,0.030262096,0.014618156,0.00016039185,0.0002050121,0.0000044646763,0.00012403153,0.099990636],"genre_scores_gemma":[0.9829607,0.000053116975,0.013652976,0.000038079525,0.00007376669,0.0000016763761,0.0000018178346,0.000008463979,0.0032093944],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9981209,0.0002798744,0.0001005624,0.00031130572,0.00075282925,0.0004345273],"domain_scores_gemma":[0.99802685,0.00047716603,0.00020811199,0.0010271206,0.00015463834,0.00010610803],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.0018083066,0.00008942586,0.00016814224,0.00022219373,0.0034058806,0.0007183229,0.0032412412,0.00005402919,0.000046925852],"category_scores_gemma":[0.00025159543,0.00008596972,0.00008776522,0.000553821,0.00025377003,0.0033231473,0.0023769105,0.00046924272,0.000023991322],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000748342,0.00043065025,0.057879448,0.00037895577,0.00025484082,0.00016479906,0.06890063,0.00017109555,0.001546164,0.07905875,0.088299006,0.7028408],"study_design_scores_gemma":[0.0030666783,0.00030470284,0.38721842,0.0012099752,0.000074745076,0.000036769157,0.047596987,0.35974196,0.003160001,0.009551644,0.1866579,0.001380227],"about_ca_topic_score_codex":0.00034357564,"about_ca_topic_score_gemma":0.00007187497,"teacher_disagreement_score":0.7014606,"about_ca_system_score_codex":0.00009525956,"about_ca_system_score_gemma":0.00019775533,"threshold_uncertainty_score":0.99789155},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2776920184","doi":"10.2106/jbjs.16.01408","title":"A Historical Analysis of Randomized Controlled Trials in Anterior Cruciate Ligament Surgery","year":2017,"lang":"en","type":"review","venue":"Journal of Bone and Joint Surgery","topic":"Knee injuries and reconstruction techniques","field":"Medicine","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Medicine; Consolidated Standards of Reporting Trials; Randomized controlled trial; Confidence interval; Checklist; Anterior cruciate ligament reconstruction; MEDLINE; Randomization; Anterior cruciate ligament; Physical therapy; Internal medicine; Surgery; Psychology","score_opus":0.1492740404499428,"score_gpt":0.3919929937555513,"score_spread":0.2427189533056085,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2776920184","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026147608,0.99486285,0.000056530083,0.00028307852,0.0011385909,0.0009767931,0.0000074341974,0.000008667461,0.00005129659],"genre_scores_gemma":[0.0007195208,0.99833465,0.00017506316,0.00002360905,0.00038428608,0.000055020828,0.0000018270282,0.00002320851,0.0002828123],"study_design_codex":"design_other","study_design_gemma":"systematic_review","domain_scores_codex":[0.9731239,0.00213724,0.023830436,0.00021177727,0.0004818212,0.00021481818],"domain_scores_gemma":[0.9481641,0.011887097,0.039085317,0.00031451025,0.00035658016,0.00019240066],"candidate_categories":["metaresearch","metaepi_broad","bibliometrics"],"consensus_categories":["metaresearch","metaepi_broad"],"category_scores_codex":[0.08011701,0.00039217345,0.120307244,0.011377909,0.000044866,0.00004911696,0.00003448063,0.00038862138,0.00008258861],"category_scores_gemma":[0.033142205,0.0002084588,0.042295717,0.000399088,0.00016623123,0.0000970611,0.000035828445,0.00052691693,3.339556e-7],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.031702098,0.00020365749,0.00042215857,0.011281198,0.018936587,0.0004947832,0.0000318143,6.178262e-8,0.0000061075743,0.0000016248596,0.0017259794,0.93519396],"study_design_scores_gemma":[0.07222408,0.00018648621,0.0003846979,0.43138647,0.14189453,0.0022251478,0.000036280646,0.000008990268,0.00002049844,0.000088432724,0.35096523,0.00057916465],"about_ca_topic_score_codex":0.00007879275,"about_ca_topic_score_gemma":0.0000029949717,"teacher_disagreement_score":0.9346148,"about_ca_system_score_codex":0.00028900968,"about_ca_system_score_gemma":0.0009843,"threshold_uncertainty_score":0.99982727},"labels":[{"model":"gemma","categories":["metaresearch","metaepi_broad"],"domain":"reporting","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","metaepi_broad","bibliometrics"],"domain":"reporting","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2778151157","doi":"10.18438/b8h66r","title":"Effect of Undergraduate Research Output on Faculty Scholarly Research Impact","year":2017,"lang":"en","type":"article","venue":"Evidence Based Library and Information Practice","topic":"Scientific Research and Technology","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Citation; Context (archaeology); Variables; Undergraduate research; Regression analysis; Linear regression; Library science; Statistics; Medical education; Computer science; Psychology; Mathematics education; Mathematics; Medicine; Geography","score_opus":0.11077999805075847,"score_gpt":0.44767616987259407,"score_spread":0.3368961718218356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2778151157","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19121894,0.00060632866,0.014050792,0.7484316,0.0005633318,0.0020686004,0.000042255273,0.00040391207,0.04261421],"genre_scores_gemma":[0.9954754,0.0004205978,0.0027829523,0.000778363,0.000030621635,0.000040164257,0.000010798672,0.0000053271087,0.00045578042],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99464256,0.0021426934,0.00035202314,0.00034526314,0.0019550563,0.00056238717],"domain_scores_gemma":[0.9884024,0.008634913,0.00032206546,0.0017866738,0.00060835,0.0002455619],"candidate_categories":["metaresearch","sts","scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.017111342,0.0001307807,0.00017494055,0.0011302584,0.0015218883,0.0063361716,0.0027648478,0.00013716584,0.00002110722],"category_scores_gemma":[0.040563725,0.000094092575,0.000060456645,0.001130563,0.0006187838,0.35425636,0.0012348782,0.0012951422,0.00033312454],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010675167,0.00006892905,0.0015842088,0.00017358635,0.000025283449,0.000023330138,0.00012263795,0.000040318773,0.00035473486,0.74578524,0.019219445,0.2315348],"study_design_scores_gemma":[0.0032410305,0.01889685,0.056903422,0.0018338653,0.000014224352,0.0000839297,0.0002639706,0.0712055,0.24061246,0.027413242,0.57886916,0.000662318],"about_ca_topic_score_codex":0.000058582118,"about_ca_topic_score_gemma":6.95913e-8,"teacher_disagreement_score":0.80425644,"about_ca_system_score_codex":0.000035759185,"about_ca_system_score_gemma":0.000561863,"threshold_uncertainty_score":0.999778},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2782830987","doi":"10.2147/jmdh.s151745","title":"Exploring the characteristics, global distribution and reasons for retraction of published articles involving human research participants: a literature survey","year":2018,"lang":"en","type":"article","venue":"Journal of Multidisciplinary Healthcare","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":45,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; St. Joseph’s Healthcare Hamilton; Impact; Programs for Assessment of Technology in Health Research Institute","funders":"","keywords":"Scientific misconduct; Misconduct; Human research; Medicine; Research integrity; Research ethics; Psychology; Alternative medicine; Political science; Law; Pathology; Psychiatry; Public relations","score_opus":0.3137519640706948,"score_gpt":0.47320960417727975,"score_spread":0.15945764010658497,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2782830987","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97932386,0.0006061531,0.00017886145,0.018495353,0.0007472023,0.00028056937,0.0003329791,0.000008592391,0.00002644514],"genre_scores_gemma":[0.99748963,0.0008089292,0.00018298415,0.000023741533,0.0014134368,0.000010498448,0.000040001014,0.000005985788,0.000024794024],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9963614,0.0017639464,0.0006132077,0.00016631419,0.0006426407,0.00045251692],"domain_scores_gemma":[0.99468124,0.001194562,0.0004919742,0.0001459372,0.003253248,0.00023303802],"candidate_categories":["metaresearch","sts"],"consensus_categories":[],"category_scores_codex":[0.019114833,0.00008927316,0.00023954913,0.000075950134,0.0018949816,0.00014994829,0.00027031678,0.00031429264,0.000006395994],"category_scores_gemma":[0.010043449,0.00006389211,0.000068061454,0.0005823249,0.0005889768,0.0010422074,0.00008415802,0.0014182548,6.4417117e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016405573,0.00025950017,0.79740167,0.000398708,0.000085518484,0.000015232267,0.14393999,9.65469e-7,0.0011064596,0.030967752,0.0062269634,0.017956695],"study_design_scores_gemma":[0.00026650232,0.00050446374,0.9637245,0.0006904247,0.000017637867,0.000011902404,0.026063602,0.00013812243,0.00015120291,0.0072248415,0.0011271136,0.0000797064],"about_ca_topic_score_codex":0.0032036724,"about_ca_topic_score_gemma":0.007232984,"teacher_disagreement_score":0.16632281,"about_ca_system_score_codex":0.0003002164,"about_ca_system_score_gemma":0.0003758471,"threshold_uncertainty_score":0.99940443},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2784275186","doi":"10.1007/s11192-018-2991-4","title":"Edge Factors: Scientific Frontier Positions of Nations","year":2018,"lang":"en","type":"preprint","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Institute on Aging; National Institutes of Health","keywords":"Frontier; Work (physics); Position (finance); China; Political science; Enhanced Data Rates for GSM Evolution; Data science; Sociology; Epistemology; Public relations; Computer science; Engineering; Business; Law; Telecommunications","score_opus":0.6933349039972806,"score_gpt":0.6011293296615132,"score_spread":0.09220557433576737,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2784275186","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8979332,0.0031797665,0.033309385,0.00050351897,0.022732448,0.0012125255,0.003409441,0.00012341453,0.037596337],"genre_scores_gemma":[0.9709293,0.00014907883,0.008144205,0.000034747714,0.0003484687,0.000033022603,0.00025339396,0.000044873294,0.020062868],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9555929,0.0005347067,0.0024249677,0.0030055528,0.036915086,0.0015267918],"domain_scores_gemma":[0.9462385,0.00840401,0.0020214678,0.0048170937,0.037091125,0.0014277706],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","sts","open_science","insufficient_payload"],"category_scores_codex":[0.053577498,0.0005760624,0.0011582659,0.52635723,0.0015819465,0.012970526,0.011868752,0.0007754094,0.0027281912],"category_scores_gemma":[0.19140348,0.00044042733,0.0009515544,0.701194,0.0028271282,0.0010429792,0.009692999,0.0011652807,0.0010479243],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002664888,0.0022581255,0.39396006,0.00017011739,0.00018895455,0.000013743357,0.0023504826,0.0005941267,0.0017739418,0.0058294428,0.5652619,0.027572457],"study_design_scores_gemma":[0.0013401198,0.000727473,0.58576685,0.0002724732,0.00019471029,0.000010501634,0.0017085158,0.017906899,0.018023362,0.09850065,0.27316105,0.002387385],"about_ca_topic_score_codex":0.000112900176,"about_ca_topic_score_gemma":0.000042762713,"teacher_disagreement_score":0.29210085,"about_ca_system_score_codex":0.000673702,"about_ca_system_score_gemma":0.0024070058,"threshold_uncertainty_score":0.99988663},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2789392886","doi":"10.1067/j.cpradiol.2018.03.005","title":"Top 100 Cited articles on Radiation Exposure in Medical Imaging: A Bibliometric Analysis","year":2018,"lang":"en","type":"article","venue":"Current Problems in Diagnostic Radiology","topic":"Radiation Dose and Imaging","field":"Medicine","cited_by":11,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vancouver General Hospital; St. John’s Health Sciences Centre; University of British Columbia; Memorial University of Newfoundland","funders":"","keywords":"Medicine; Medical radiation; Scopus; Modalities; Medical physics; MEDLINE; Descriptive statistics; Web of science; Bibliometrics; Family medicine; Library science; Meta-analysis; Statistics; Pathology; Social science; Computer science","score_opus":0.027069640692170598,"score_gpt":0.33887309552023787,"score_spread":0.31180345482806726,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2789392886","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9842809,0.009204652,0.0014664645,0.0031970497,0.0007028809,0.00053827435,0.000004838433,0.000065419204,0.00053955795],"genre_scores_gemma":[0.9954091,0.0029873298,0.000050421455,0.0008122889,0.0004778065,0.0001272285,0.00010167431,0.000018636945,0.000015522093],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9972574,0.00027511723,0.0008623752,0.00055655406,0.00045850629,0.00059002906],"domain_scores_gemma":[0.99694043,0.0021421055,0.00017910886,0.0003924805,0.00010034882,0.00024552178],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0012512339,0.00021545222,0.000580577,0.055070594,0.00005230778,0.000034394336,0.00021248471,0.00014426955,0.000668138],"category_scores_gemma":[0.0132428,0.00019207493,0.00011746481,0.07700675,0.00023404417,0.00017153358,0.000049259615,0.0004731761,0.00012148289],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000025808391,0.00036236996,0.71620893,0.000024014726,0.0000605681,0.000052272964,0.00035537555,0.00038249968,0.00003940468,0.00013955239,0.0017993813,0.28054985],"study_design_scores_gemma":[0.002729434,0.0002944021,0.95025766,0.0002400855,0.00014693853,0.00005718245,0.00003838967,0.04103736,0.0001670957,0.00040107092,0.004442745,0.00018766096],"about_ca_topic_score_codex":0.00008624432,"about_ca_topic_score_gemma":0.000059264985,"teacher_disagreement_score":0.2803622,"about_ca_system_score_codex":0.00021509109,"about_ca_system_score_gemma":0.00017705966,"threshold_uncertainty_score":0.9950691},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2792318286","doi":"10.5334/ijic.s1112","title":"Current performance and future trends of integrated care: a scientometric analysis","year":2018,"lang":"en","type":"article","venue":"International Journal of Integrated Care","topic":"Chronic Disease Management Strategies","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Popularity; Integrated care; Incentive; Health care; Public relations; Political science; Business; Medicine; Economics","score_opus":0.017290947889591783,"score_gpt":0.33878278297642983,"score_spread":0.32149183508683804,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2792318286","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9834953,0.0076959194,0.00034744327,0.00041723478,0.0016392876,0.000082595885,0.00014470825,0.000020928428,0.006156599],"genre_scores_gemma":[0.9972078,0.0012928908,0.00035375735,0.00006191046,0.0006799384,0.000002411741,0.00024900172,0.000013014303,0.00013925388],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99794304,0.00003822053,0.0006160991,0.00021582733,0.0010068294,0.0001799895],"domain_scores_gemma":[0.99395126,0.000023674105,0.0004875011,0.0001870521,0.005191365,0.00015914417],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00019016118,0.00020254572,0.00043466824,0.00502967,0.000045634064,0.00008662643,0.0003788594,0.00006761689,0.000811074],"category_scores_gemma":[0.000103667844,0.0001439766,0.0003039809,0.0044389036,0.00025006704,0.00025395086,0.00007185109,0.00032309696,0.0000045936235],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012184259,0.00026289397,0.28379488,0.00023149968,0.004190627,0.000103804545,0.0030998336,0.00006728335,0.0004741758,0.00044946143,0.003647812,0.7024593],"study_design_scores_gemma":[0.007839386,0.0032128969,0.6954362,0.0017199109,0.005962387,0.0002798361,0.07554737,0.0033753214,0.0058376472,0.00003746209,0.20018446,0.000567124],"about_ca_topic_score_codex":0.000099136254,"about_ca_topic_score_gemma":0.000055635137,"teacher_disagreement_score":0.7018922,"about_ca_system_score_codex":0.0009299381,"about_ca_system_score_gemma":0.0007646401,"threshold_uncertainty_score":0.88806915},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2794223847","doi":"10.1002/asi.24000","title":"A new approach to web co‐link analysis","year":2018,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Link (geometry); Link analysis; Computer science; Data link; Variety (cybernetics); Data science; World Wide Web; Data mining; Computer network","score_opus":0.01557187086333424,"score_gpt":0.2774964609037368,"score_spread":0.2619245900404026,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2794223847","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.026874883,0.000009046097,0.92897296,0.022865297,0.00057922624,0.0002922489,0.0000053192,0.00005026487,0.020350778],"genre_scores_gemma":[0.93694896,0.000007218548,0.06141035,0.0012860415,0.000075070435,0.0000025250397,3.7432372e-7,9.84636e-7,0.00026846825],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986272,0.000009626782,0.00042251946,0.00007902629,0.0006851432,0.00017650526],"domain_scores_gemma":[0.9966003,0.000069049696,0.0007529578,0.00026595293,0.0022372582,0.00007450899],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00383031,0.000050954397,0.00013954336,0.0022702035,0.00034779325,0.0003259119,0.0013038628,0.000088129775,0.0000016195374],"category_scores_gemma":[0.004054183,0.000033931527,0.000069010515,0.011273768,0.00009574378,0.0035578178,0.00018474953,0.000112215326,0.000015768195],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002485196,0.000048298836,0.03889364,0.000013281352,0.00024458914,5.0303747e-8,0.0047687986,0.00029984905,0.00077633746,0.5745225,0.0508374,0.32957038],"study_design_scores_gemma":[0.0014682908,0.00072381523,0.043662094,0.000013911412,0.00016386806,0.000039668263,0.0007833175,0.12627915,0.012857734,0.025500052,0.7882042,0.0003039433],"about_ca_topic_score_codex":0.0000023469004,"about_ca_topic_score_gemma":0.0000020455047,"teacher_disagreement_score":0.91007406,"about_ca_system_score_codex":0.00024737895,"about_ca_system_score_gemma":0.0005598929,"threshold_uncertainty_score":0.54166716},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2796216221","doi":"10.1177/0963662518761733","title":"Identifying diffusion patterns of research articles on Twitter: A case study of online engagement with open access articles","year":2018,"lang":"en","type":"article","venue":"Public Understanding of Science","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":76,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; Simon Fraser University","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"Open science; Social media; Science communication; Sociology; Scholarly communication; Citizen science; Ask price; World Wide Web; Altmetrics; Online community; Public engagement; Data science; Public relations; Computer science; Political science; Biology; Science education; Publishing; Business","score_opus":0.50896915729909,"score_gpt":0.49662623253611704,"score_spread":0.01234292476297294,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2796216221","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96721405,0.0000018435338,0.031666577,0.00014610196,0.000017818847,0.00043485084,0.000005407892,0.000013108581,0.0005002314],"genre_scores_gemma":[0.9992802,7.697605e-7,0.0006444073,0.000006315426,0.00003245316,0.000012994395,0.0000010831084,0.000007875936,0.000013901553],"study_design_codex":"observational","study_design_gemma":"qualitative","domain_scores_codex":[0.9978277,0.00016571718,0.00037132128,0.0003587165,0.00094658334,0.0003299293],"domain_scores_gemma":[0.9984764,0.00019617479,0.00028400868,0.0004961622,0.00045727956,0.00009000579],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002881206,0.000097280354,0.00023178803,0.00074365933,0.0005672697,0.0004536173,0.001341408,0.000009673142,0.00012861924],"category_scores_gemma":[0.00003649474,0.000074935975,0.0000300079,0.0024379669,0.0011704317,0.00085516536,0.0016804052,0.00013520113,3.5891742e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006484748,0.0030802498,0.94370943,0.00003387287,0.00008888314,0.000014531839,0.015173807,0.000026970743,0.011692062,0.02102828,0.00006047397,0.0050265607],"study_design_scores_gemma":[0.0026663686,0.004180921,0.08483757,0.00087415334,0.00010988908,0.000011658533,0.74957347,0.013003196,0.10700311,0.037282575,0.000019638108,0.00043747746],"about_ca_topic_score_codex":0.005020816,"about_ca_topic_score_gemma":0.0012619218,"teacher_disagreement_score":0.8588719,"about_ca_system_score_codex":0.000115033916,"about_ca_system_score_gemma":0.00013726972,"threshold_uncertainty_score":0.7590005},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2802617914","doi":"10.1016/j.jpurol.2018.04.004","title":"Historical bibliometric analysis of the top cited articles on vesicoureteral reflux 1950–2016, and incorporation of a novel impact index","year":2018,"lang":"en","type":"article","venue":"Journal of Pediatric Urology","topic":"Pediatric Urology and Nephrology Studies","field":"Medicine","cited_by":26,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Children's Hospital of Eastern Ontario; Hospital for Sick Children","funders":"","keywords":"Medicine; Vesicoureteral reflux; Bibliometrics; Science Citation Index; Impact factor; Citation; Index (typography); Web of science; MEDLINE; Disease; Library science; Meta-analysis; Pathology; Reflux; Computer science","score_opus":0.03211043476176292,"score_gpt":0.3084546639490122,"score_spread":0.2763442291872493,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2802617914","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99474883,0.0026332943,0.0003509901,0.001767583,0.00034721987,0.00009211966,0.000009634803,0.0000052742603,0.000045077017],"genre_scores_gemma":[0.99788576,0.00048045802,0.000183461,0.0004240197,0.0009973772,0.0000011612643,0.0000011531357,0.000007969545,0.000018636318],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9982702,0.00015540047,0.00082427624,0.00017137908,0.00035541737,0.00022329888],"domain_scores_gemma":[0.9971322,0.000604334,0.0013218691,0.00023456966,0.00059827615,0.00010876184],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0008210742,0.00015155206,0.00076645805,0.028594522,0.00007235203,0.0000031627503,0.000156443,0.00027779245,0.000038941274],"category_scores_gemma":[0.0011046823,0.000091287664,0.00031152894,0.035217572,0.00032199206,0.00007843008,0.00007571335,0.0004156899,7.7319913e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021599997,0.0005644254,0.98871356,0.000005461284,0.0028595328,0.000012432705,0.00020575509,0.00003581338,0.0013646123,0.000021898764,0.0034203983,0.00063608564],"study_design_scores_gemma":[0.002346322,0.0049685673,0.9834268,7.037109e-7,0.008393644,0.00007985249,0.0000074282557,0.00035272617,0.000073959905,0.0001264925,0.00016173595,0.00006176886],"about_ca_topic_score_codex":0.00004661841,"about_ca_topic_score_gemma":0.000011789437,"teacher_disagreement_score":0.00662305,"about_ca_system_score_codex":0.000084800464,"about_ca_system_score_gemma":0.00022695768,"threshold_uncertainty_score":0.98528934},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2804203984","doi":"10.1002/asi.24028","title":"Co‐saved, co‐tweeted, and co‐cited networks","year":2018,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Altmetrics; Citation; Computer science; Social media; Value (mathematics); World Wide Web; Information retrieval; Machine learning","score_opus":0.19676722602553115,"score_gpt":0.5112396376478187,"score_spread":0.3144724116222875,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2804203984","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.962294,0.00015981316,0.010309843,0.0141032105,0.001220654,0.00046992488,0.00003975421,0.000028273429,0.011374529],"genre_scores_gemma":[0.99872684,0.00012495997,0.00033933588,0.00042320084,0.000073953546,0.0000025683062,7.586388e-7,0.0000019367692,0.00030642547],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99409467,0.00005034599,0.00079577684,0.00013980485,0.0045649996,0.0003544072],"domain_scores_gemma":[0.98417073,0.0012557354,0.0016630946,0.00024177443,0.012534329,0.00013430374],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.03408625,0.000074811665,0.00020437688,0.025177019,0.0009253584,0.0017545598,0.001495512,0.00018034426,0.000015668309],"category_scores_gemma":[0.06442229,0.000043196607,0.000054157794,0.065256074,0.00083238777,0.0031848825,0.00017749378,0.00025585183,0.000029838462],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006642761,0.0000438294,0.55449426,0.000005447136,0.000042033607,4.1321866e-7,0.0006678327,0.00003160532,0.001990887,0.017801387,0.08468142,0.34017447],"study_design_scores_gemma":[0.0021833254,0.0009572962,0.23402531,0.000020574556,0.000021737684,0.00011073426,0.0024275423,0.030305708,0.01391759,0.02323912,0.6925508,0.00024024793],"about_ca_topic_score_codex":0.0000027331048,"about_ca_topic_score_gemma":0.0000020421483,"teacher_disagreement_score":0.6078694,"about_ca_system_score_codex":0.00024591287,"about_ca_system_score_gemma":0.00031346595,"threshold_uncertainty_score":0.9992817},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics","scholarly_communication"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2807638568","doi":"10.1109/bigcomp.2018.00065","title":"Article Impact Value for Nearby Citation Network Analysis","year":2018,"lang":"en","type":"article","venue":"","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Impact factor; Citation; Computer science; Metadata; Value (mathematics); Similarity (geometry); Citation impact; Data science; Measure (data warehouse); Information retrieval; Citation analysis; Data mining; World Wide Web; Artificial intelligence; Political science; Machine learning","score_opus":0.6134994399169142,"score_gpt":0.6385578056579951,"score_spread":0.02505836574108089,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2807638568","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.62909865,0.000075857126,0.3553285,0.0005141652,0.0002508757,0.0002120463,0.000014686233,0.000026314703,0.014478905],"genre_scores_gemma":[0.9808481,0.0000053942963,0.015468421,0.00022398605,0.00032752176,0.000009571951,0.0000042798642,0.000005445104,0.0031073112],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9944021,0.00012505894,0.00045490614,0.000492932,0.0039321966,0.00059279654],"domain_scores_gemma":[0.99070346,0.0045820973,0.00013210742,0.0005835131,0.0036602335,0.0003385882],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.019606391,0.00008786814,0.00023338803,0.03320032,0.0003416445,0.0021836604,0.0009463899,0.000064905376,0.0023592892],"category_scores_gemma":[0.020527562,0.00005238668,0.00036024893,0.3756772,0.0001237161,0.00044014092,0.000172451,0.000060892136,0.0007180465],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009410932,0.00011484323,0.6269308,0.0000010562676,0.00030448858,0.000001071871,0.00015848559,0.006998422,0.0007990551,0.013912087,0.18352419,0.16716142],"study_design_scores_gemma":[0.00024535475,0.00033074297,0.47690114,5.366019e-7,0.000041488227,4.1300277e-7,0.00008202003,0.47613522,0.00064118236,0.03315788,0.012363451,0.00010056659],"about_ca_topic_score_codex":0.0001595796,"about_ca_topic_score_gemma":0.000080089456,"teacher_disagreement_score":0.4691368,"about_ca_system_score_codex":0.000067482266,"about_ca_system_score_gemma":0.00009482328,"threshold_uncertainty_score":0.9988522},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2883757404","doi":"10.1016/j.jpsychires.2018.07.010","title":"Bibliometric analysis of manuscript characteristics that influence citations: A comparison of six major psychiatry journals","year":2018,"lang":"en","type":"article","venue":"Journal of Psychiatric Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":47,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Vancouver General Hospital","funders":"","keywords":"Citation; Rank correlation; Spearman's rank correlation coefficient; Impact factor; Univariate; Multivariate statistics; Psychology; Multivariate analysis; Statistics; Medicine; Library science; Mathematics; Computer science; Chemistry","score_opus":0.5968616088670978,"score_gpt":0.6301201752005898,"score_spread":0.033258566333492,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2883757404","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9795207,0.01046674,0.0048508947,0.001671984,0.0017979047,0.0002836881,0.00009756198,0.000005211017,0.001305363],"genre_scores_gemma":[0.9867042,0.0021328945,0.010376019,0.00004538997,0.00044976952,0.0000036306833,0.0000022143852,0.000020682939,0.00026523645],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9614793,0.0016991335,0.005244855,0.00064433325,0.02985787,0.0010744829],"domain_scores_gemma":[0.93443704,0.010218004,0.0054369783,0.001395311,0.04760624,0.00090643094],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.08591991,0.00026720666,0.0017956293,0.8672909,0.0005002534,0.001646327,0.006129359,0.0002896029,0.001351759],"category_scores_gemma":[0.07156113,0.00018634829,0.0010789467,0.93099636,0.001000055,0.0011621555,0.0006614576,0.0014897891,0.00009409264],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026624725,0.0008762339,0.930644,0.00004460936,0.0006391684,0.0000060359876,0.00030102298,0.000054982498,0.0010743319,0.0004084448,0.026820043,0.03886488],"study_design_scores_gemma":[0.0008096612,0.0019882878,0.9844057,0.00006087221,0.0003194894,0.00001889264,0.0016948208,0.0021644838,0.0005258755,0.0056808433,0.0021587138,0.00017237711],"about_ca_topic_score_codex":0.00017682479,"about_ca_topic_score_gemma":0.00006518977,"teacher_disagreement_score":0.06370548,"about_ca_system_score_codex":0.00017435463,"about_ca_system_score_gemma":0.0015753772,"threshold_uncertainty_score":0.99956113},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2885423250","doi":"10.1051/epjconf/201818605001","title":"A Bibliometric Analysis of Observatory Publications 2011-2015","year":2018,"lang":"en","type":"article","venue":"EPJ Web of Conferences","topic":"Historical Astronomy and Related Studies","field":"Physics and Astronomy","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada; Dominion Astrophysical Observatory","funders":"National Aeronautics and Space Administration","keywords":"Bibliometrics; Observatory; Measure (data warehouse); Library science; Data science; Geography; Regional science; Computer science; Astronomy; Data mining; Physics","score_opus":0.03642453352893265,"score_gpt":0.28562196512392385,"score_spread":0.2491974315949912,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2885423250","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.74460334,0.0008712589,0.004526692,0.0004478731,0.0002378176,0.000105959094,0.000112560294,0.000025424615,0.24906906],"genre_scores_gemma":[0.99819624,0.000018011358,0.00030933102,0.0000043803334,0.00008354713,0.000009951907,0.000021306689,0.0000026557882,0.0013545613],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9992133,0.000030092982,0.00031038641,0.00014496478,0.00017010298,0.00013119244],"domain_scores_gemma":[0.9989101,0.00010395877,0.00024046414,0.00019511841,0.00049480377,0.000055553777],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00014180219,0.000086772874,0.00032600254,0.014044722,0.0000806387,0.000011095745,0.00021393446,0.00003445646,0.0039228913],"category_scores_gemma":[0.000015298443,0.000068017485,0.00015848466,0.040695336,0.00028236938,0.000093436254,0.000044335222,0.00006868057,0.00004392032],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004618729,0.000148817,0.92972887,0.0000041308294,0.0015992891,2.1365246e-8,0.0001764331,0.0000064379833,0.00012384972,0.048693534,0.007922312,0.011591686],"study_design_scores_gemma":[0.0002170832,0.00013509547,0.7877921,0.000011504931,0.0007934594,1.8903764e-8,0.0004682727,0.0006595838,0.000994637,0.0009437595,0.20783645,0.00014801488],"about_ca_topic_score_codex":0.00041548326,"about_ca_topic_score_gemma":0.000025566133,"teacher_disagreement_score":0.2535929,"about_ca_system_score_codex":0.0000073022925,"about_ca_system_score_gemma":0.00031872583,"threshold_uncertainty_score":0.9971302},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2889499003","doi":"10.1503/cmaj.180154","title":"How predatory journals leak into PubMed","year":2018,"lang":"en","type":"article","venue":"Canadian Medical Association Journal","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":146,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"Ottawa Hospital","funders":"University of Ottawa","keywords":"Key (lock); Data science; Computer science; MEDLINE; World Wide Web; Computer security; Political science","score_opus":0.3334433946008996,"score_gpt":0.49615533824865493,"score_spread":0.16271194364775532,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2889499003","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.41811004,0.0046129115,0.009230076,0.49973026,0.015018801,0.00049344194,0.000093361334,0.00005080419,0.0526603],"genre_scores_gemma":[0.9742511,0.000494603,0.00024222457,0.0061596506,0.0058349804,0.0000070961364,0.0000013371813,0.0000141316705,0.012994914],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9711182,0.0007254727,0.0008256675,0.00039136317,0.02564573,0.001293546],"domain_scores_gemma":[0.9762717,0.0021838038,0.00064313784,0.0003524112,0.012872343,0.0076766363],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.081073776,0.0001390894,0.00031107126,0.03402918,0.0010814692,0.010648636,0.0027977617,0.00055010524,0.01038845],"category_scores_gemma":[0.3063317,0.00009917778,0.00020366542,0.065459564,0.00020763416,0.0013078277,0.00016406787,0.0014972622,0.0011711649],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000022139843,0.000012483178,0.10155125,5.924e-7,0.000024893747,0.000119374425,0.00012342192,2.1349014e-7,0.0000059008603,0.00010704813,0.44962513,0.44842747],"study_design_scores_gemma":[0.00037535562,0.00006154756,0.14908528,0.00001317011,0.0000033680515,0.00018016706,0.00050850585,0.0004152021,0.000030059055,0.005552176,0.84365284,0.00012230406],"about_ca_topic_score_codex":0.0007693563,"about_ca_topic_score_gemma":0.031135788,"teacher_disagreement_score":0.556141,"about_ca_system_score_codex":0.0019717785,"about_ca_system_score_gemma":0.005237541,"threshold_uncertainty_score":0.99960655},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","scholarly_communication","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2891358922","doi":"","title":"A Bibliometric Analysis of the Literature on Open Access in Scopus","year":2017,"lang":"en","type":"article","venue":"Qualitative and Quantitative Methods in Libraries","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Subject (documents); Library science; Bibliometrics; Impact factor; English language; Political science; MEDLINE; Computer science; Psychology; Law; Mathematics education","score_opus":0.8790385872218804,"score_gpt":0.7668242092888382,"score_spread":0.11221437793304223,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2891358922","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.968034,0.0047670845,0.0092907045,0.002677395,0.0003507683,0.00064251904,0.0002426237,0.0000050441004,0.013989802],"genre_scores_gemma":[0.8838767,0.00054875045,0.114401266,0.00022454862,0.000010529576,0.00004569387,0.0000043719915,0.000011548898,0.00087660603],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98608893,0.008483156,0.0011504672,0.0009280265,0.0029128147,0.00043659497],"domain_scores_gemma":[0.9232477,0.071864516,0.0013403982,0.001772143,0.0015980599,0.00017716654],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.064378716,0.00022220208,0.0010459211,0.55816,0.00050483807,0.01656952,0.008974403,0.00013363188,0.00013079698],"category_scores_gemma":[0.25758505,0.0001223615,0.00021646479,0.834519,0.0014306215,0.006781475,0.0053850645,0.0004980281,0.000005538369],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002660429,0.00022943765,0.5169576,0.00002534844,0.00019663354,0.000008607911,0.033372097,0.000056757683,0.0002577525,0.36478114,0.000514643,0.08333395],"study_design_scores_gemma":[0.00042923624,0.00017453356,0.765057,0.00009199074,0.000020013933,1.6142455e-7,0.009150799,0.0011885402,0.0014302992,0.2219257,0.00039781345,0.00013385886],"about_ca_topic_score_codex":0.00080342655,"about_ca_topic_score_gemma":0.0002613502,"teacher_disagreement_score":0.27635902,"about_ca_system_score_codex":0.000041464402,"about_ca_system_score_gemma":0.00021693026,"threshold_uncertainty_score":0.99638754},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2892809183","doi":"10.1016/j.apmr.2018.07.415","title":"Trends in Randomized Controlled Trials Of Technology-Based Interventions For Upper Limb Motor Recovery Post Stroke","year":2018,"lang":"en","type":"article","venue":"Archives of Physical Medicine and Rehabilitation","topic":"Stroke Rehabilitation and Recovery","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lawson Health Research Institute","funders":"","keywords":"Physical medicine and rehabilitation; Rehabilitation; Stroke (engine); Psychological intervention; Upper limb; Motor function; Randomized controlled trial; Physical therapy; Medicine; Surgery; Engineering","score_opus":0.02007942061728634,"score_gpt":0.35263441828369424,"score_spread":0.3325549976664079,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2892809183","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98158604,0.0013262391,0.0022754427,0.010877755,0.00019122373,0.002685964,0.000085949934,0.000023965053,0.00094744656],"genre_scores_gemma":[0.9848808,0.00006339869,0.013998111,0.000071511306,0.00022852299,0.0003660832,0.000047417852,0.000012961876,0.00033117988],"study_design_codex":"randomized_trial","study_design_gemma":"randomized_trial","domain_scores_codex":[0.9974008,0.0004624256,0.0015554525,0.00024673512,0.00019124505,0.00014330895],"domain_scores_gemma":[0.9492353,0.04965275,0.00059467135,0.0001966089,0.0002555307,0.000065133194],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0019786502,0.00015982924,0.0032972589,0.003176179,0.000030538133,0.000002270277,0.000055503,0.00007931307,0.000044302582],"category_scores_gemma":[0.05842678,0.0000924758,0.0030308478,0.0002891329,0.0019096271,0.000053437234,0.0000147128685,0.00010984911,5.0910296e-7],"study_design_candidate":"randomized_trial","study_design_consensus":"randomized_trial","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.41475993,0.0017974915,0.0023342199,0.0024692123,0.00074434705,3.531256e-7,0.0034520642,0.000009529031,0.3075603,0.009288598,0.00012647294,0.25745746],"study_design_scores_gemma":[0.75209826,0.0993245,0.05885486,0.017288469,0.0024721804,0.0000024304993,0.0065218084,0.012457051,0.008335153,0.042058527,0.0002896307,0.0002971024],"about_ca_topic_score_codex":0.000024745123,"about_ca_topic_score_gemma":0.0000076219208,"teacher_disagreement_score":0.33733836,"about_ca_system_score_codex":0.000014151513,"about_ca_system_score_gemma":0.000059712187,"threshold_uncertainty_score":0.9495045},"labels":[{"model":"gemma","categories":["metaresearch","metaepi_broad"],"domain":"methods","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2900769680","doi":"10.3171/2018.2.peds17354","title":"The 50 most cited publications in endoscopic third ventriculostomy: a bibliometric analysis","year":2018,"lang":"en","type":"article","venue":"Journal of Neurosurgery Pediatrics","topic":"Cerebrospinal fluid and hydrocephalus","field":"Neuroscience","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Hamilton General Hospital","funders":"","keywords":"Endoscopic third ventriculostomy; Medicine; Publish or perish; Neurosurgery; Hydrocephalus; Bibliometrics; Obstructive hydrocephalus; Library science; Surgery; Literature; Publishing; Computer science","score_opus":0.03188451817586877,"score_gpt":0.2908797676511414,"score_spread":0.2589952494752726,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2900769680","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99469817,0.00084015366,0.0002606821,0.0017294171,0.0014524653,0.00013224171,0.00001625267,0.000033001415,0.00083762396],"genre_scores_gemma":[0.9933714,0.004304811,0.000035070858,0.0007771509,0.0011336274,0.000005069953,0.0000012005997,0.000020134508,0.00035151423],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9966522,0.00028830374,0.0011493221,0.0003267082,0.0010498384,0.00053364865],"domain_scores_gemma":[0.99519587,0.0027014748,0.0009234922,0.00041769375,0.0005075838,0.00025391803],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.001189082,0.00020609272,0.00043837886,0.08878255,0.00045882762,0.00044674682,0.0007821045,0.00009063055,0.000059539347],"category_scores_gemma":[0.008950205,0.00014201201,0.00037714056,0.3564421,0.00016413885,0.00054458674,0.00012869015,0.00054957333,0.000049151382],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000109713284,0.00045376143,0.95427436,0.000017773782,0.00004151904,0.00017244545,0.000076035416,0.0001480325,0.013390575,0.00048635525,0.02754576,0.0032836823],"study_design_scores_gemma":[0.00061762024,0.00038781078,0.93108875,0.000008573454,0.00037101036,0.00019284697,0.000020053441,0.0010587294,0.0031961894,0.00034065385,0.062481638,0.00023612562],"about_ca_topic_score_codex":0.000012023747,"about_ca_topic_score_gemma":0.000013009628,"teacher_disagreement_score":0.26765954,"about_ca_system_score_codex":0.00009963576,"about_ca_system_score_gemma":0.0003037142,"threshold_uncertainty_score":0.9993978},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2904190351","doi":"10.1109/smartworld.2018.00311","title":"Landscapes and Emerging Trends of Virtual Reality in Recent 30 Years: A Bibliometric Analysis","year":2018,"lang":"en","type":"article","venue":"","topic":"Virtual Reality Applications and Impacts","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Virtual reality; VRML; Mixed reality; Computer science; Optical head-mounted display; Visualization; Augmented reality; Human–computer interaction; Artificial intelligence","score_opus":0.03533571695498103,"score_gpt":0.32736697482558563,"score_spread":0.2920312578706046,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2904190351","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.82971287,0.00008523646,0.1593759,0.00083891937,0.000024606768,0.000046946472,0.0000070725796,0.00004374353,0.00986471],"genre_scores_gemma":[0.997362,0.000247954,0.0022309795,0.00006592205,0.00001639868,0.0000030999506,0.0000023236648,0.0000019913612,0.0000693025],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9991474,0.00004200874,0.0002305462,0.0002323494,0.00019667031,0.00015100236],"domain_scores_gemma":[0.9993241,0.00007490705,0.00007109492,0.0003723205,0.00007853123,0.00007904133],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0004987215,0.00006181063,0.0001708776,0.029669242,0.00002645615,0.00006308992,0.0002658016,0.000034162927,0.00014091875],"category_scores_gemma":[0.00008985132,0.000053037624,0.000034905006,0.15653726,0.000039041628,0.00022524966,0.00014477648,0.000040895196,0.0000072558823],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000003836266,0.0000996772,0.0110749,0.0000026716114,0.000058528738,7.0059394e-7,0.0007212436,0.000044791577,0.00020714899,0.015048341,0.00067611533,0.97206205],"study_design_scores_gemma":[0.00024254985,0.00017951409,0.93312824,0.000004433826,0.000024800745,9.909127e-7,0.00007154661,0.059595402,0.00048111498,0.00066994625,0.0054982193,0.000103272774],"about_ca_topic_score_codex":0.00071771693,"about_ca_topic_score_gemma":0.0005450448,"teacher_disagreement_score":0.97195876,"about_ca_system_score_codex":0.000016577644,"about_ca_system_score_gemma":0.000021555848,"threshold_uncertainty_score":0.9813286},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2908202864","doi":"","title":"Bibliometric analysis of Traffic Medicine – related publications: 2008 – 2015","year":2019,"lang":"en","type":"article","venue":"IUG Journal of Natural Studies","topic":"Traffic and Road Safety","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Road traffic; Web of science; Geography; Medicine; Transport engineering; MEDLINE; Political science; Engineering","score_opus":0.01993017348047422,"score_gpt":0.29584967331143897,"score_spread":0.27591949983096475,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2908202864","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7693722,0.22741681,0.000022507431,0.0011859359,0.0010352739,0.000063396285,0.0000045549,0.00004902561,0.00085029297],"genre_scores_gemma":[0.9816293,0.017522525,0.000162785,0.000013181745,0.000066908025,6.02778e-7,0.0000039963606,0.000009675726,0.00059101917],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.9986429,0.00002710567,0.00069290097,0.00008199869,0.00040062354,0.00015447006],"domain_scores_gemma":[0.9986305,0.0002722529,0.00028124746,0.00013605645,0.0006215157,0.00005843329],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00045203982,0.00012699861,0.0006610418,0.038848385,0.000036711546,0.000006459641,0.00020671089,0.00006315854,0.0001039954],"category_scores_gemma":[0.0002577937,0.00008044373,0.00021424377,0.09014015,0.00007916195,0.00021563467,0.00002179365,0.00029617373,0.0000107222495],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008515769,0.00013665183,0.029856903,0.00023086707,0.034451704,0.00001934772,0.002780176,0.6068072,0.0010405098,0.00031707474,0.19629513,0.1279793],"study_design_scores_gemma":[0.0016847433,0.00026173316,0.9555066,0.00018980078,0.002456466,0.000060283575,0.001670432,0.01908157,0.000057566343,0.000033907556,0.018723384,0.0002735161],"about_ca_topic_score_codex":0.0000014066624,"about_ca_topic_score_gemma":0.0000053003246,"teacher_disagreement_score":0.9256497,"about_ca_system_score_codex":0.00006832804,"about_ca_system_score_gemma":0.000015342775,"threshold_uncertainty_score":0.9720454},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2910274605","doi":"10.4000/belgeo.30799","title":"L’émergence des géographies étudiantes : une littérature anglophone substantielle, une recherche francophone à bâtir","year":2018,"lang":"fr","type":"article","venue":"BELGEO","topic":"Historical Geography and Geographical Thought","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Francophone University Association","funders":"","keywords":"Humanities; French; Sociology; Philosophy","score_opus":0.0777656181305753,"score_gpt":0.3273649800011468,"score_spread":0.24959936187057147,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2910274605","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.104960635,0.80432355,0.00055311585,0.017811852,0.009133658,0.0005565585,0.00013306308,0.00039121124,0.06213635],"genre_scores_gemma":[0.25742134,0.6936474,0.0073280907,0.0009112092,0.004937831,0.000065599335,0.000025156342,0.00008951285,0.035573896],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9937245,0.0014019858,0.000832362,0.0011521233,0.00102387,0.0018651204],"domain_scores_gemma":[0.9962987,0.00085319544,0.00032122585,0.00080923113,0.0008292588,0.0008884295],"candidate_categories":["metaepi_narrow","bibliometrics","sts","research_integrity","insufficient_payload"],"consensus_categories":["sts","insufficient_payload"],"category_scores_codex":[0.00209946,0.00072161073,0.000849061,0.00049030944,0.0027190854,0.00013671491,0.0010594341,0.0016822275,0.0018048527],"category_scores_gemma":[0.0006559261,0.0007072275,0.0008493697,0.56780994,0.012657628,0.00060531753,0.00016103136,0.0015539392,0.0010675124],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019303574,0.0009027865,0.016840806,0.0001843603,0.0003420466,0.000107267035,0.03361976,0.0000019276454,0.001027005,0.03397618,0.037089232,0.8757156],"study_design_scores_gemma":[0.0005127934,0.0005576879,0.024734542,0.0003828945,0.00012760222,0.000010515031,0.00858075,0.000008282727,0.0005044735,0.04436832,0.9192932,0.0009189342],"about_ca_topic_score_codex":0.015515664,"about_ca_topic_score_gemma":0.04182912,"teacher_disagreement_score":0.882204,"about_ca_system_score_codex":0.00015386731,"about_ca_system_score_gemma":0.0002576626,"threshold_uncertainty_score":0.99971026},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2910422483","doi":"10.21083/partnership.v13i2.4066","title":"An Examination of Bibliometrics in Calls for Major Canadian Research Awards","year":2019,"lang":"en","type":"article","venue":"Partnership The Canadian Journal of Library and Information Practice and Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"Concordia University","funders":"Concordia University","keywords":"Bibliometrics; CLARITY; Nomination; Multidisciplinary approach; Library science; Political science; Sociology; Social science; Computer science; Law; Biology","score_opus":0.537772614021961,"score_gpt":0.5752178515967156,"score_spread":0.0374452375747546,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2910422483","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9154994,0.0018538585,0.000056084893,0.033546492,0.00022586847,0.00087451946,0.00006490031,0.0000021101798,0.04787676],"genre_scores_gemma":[0.9975314,0.0010162259,0.00033560087,0.0004224414,0.00006835586,0.000007750101,0.000008581502,0.000006629699,0.00060299714],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99100024,0.0021976202,0.00094373204,0.00019100757,0.004887743,0.00077966775],"domain_scores_gemma":[0.9730206,0.01875061,0.00037333646,0.0003973679,0.005884368,0.0015737242],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.12084667,0.00008641049,0.00022838329,0.19779724,0.00066530885,0.0047421344,0.0014639833,0.00017903258,0.00021900507],"category_scores_gemma":[0.054626934,0.00005495437,0.000046594654,0.15123385,0.0004553213,0.028041992,0.000110923866,0.0010309868,0.000027600421],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006349681,0.0001333261,0.23994103,0.00022924076,0.00007055163,0.000049398343,0.0144651225,0.00022453378,0.00010690443,0.24532546,0.04165731,0.45716214],"study_design_scores_gemma":[0.00091621414,0.0012029897,0.35409218,0.000047695263,0.0000054231127,0.000068967,0.02338605,0.00399949,0.00034294554,0.0068855104,0.608925,0.00012758476],"about_ca_topic_score_codex":0.02224763,"about_ca_topic_score_gemma":0.013885175,"teacher_disagreement_score":0.56726766,"about_ca_system_score_codex":0.00011862844,"about_ca_system_score_gemma":0.0060474095,"threshold_uncertainty_score":0.99958736},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W2911600539","doi":"10.1002/pra2.2018.14505501073","title":"Peer review, bibliometrics and altmetrics ‐ Do we need them all?","year":2018,"lang":"en","type":"article","venue":"Proceedings of the Association for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Altmetrics; Bibliometrics; Citation; Science Citation Index; Computer science; Impact factor; Proxy (statistics); Quality (philosophy); Social media; Data science; Toolbox; Data mining; World Wide Web; Political science","score_opus":0.2820165169104853,"score_gpt":0.4876526026311676,"score_spread":0.20563608572068232,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2911600539","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5581638,0.0045300024,0.0007363784,0.36073324,0.0015910076,0.003012206,0.00016606432,0.00019221498,0.070875056],"genre_scores_gemma":[0.9889711,0.0067793056,0.0017004773,0.0007872584,0.000035979243,0.000033289554,9.00774e-7,0.000004673861,0.0016870138],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9873976,0.000008478667,0.0008023398,0.000321856,0.011006906,0.0004628012],"domain_scores_gemma":[0.91113234,0.0014810675,0.0018440691,0.00030969683,0.08505588,0.00017694743],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.054493397,0.00011219452,0.00027458163,0.10012245,0.0006308879,0.0019620855,0.0022462853,0.00019104233,0.000012718832],"category_scores_gemma":[0.36318597,0.00006763353,0.00005593106,0.4804925,0.00084067136,0.0042193476,0.0011231335,0.00018819123,0.000033865235],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011083031,0.000036678925,0.25533092,0.000103932856,0.00002002996,1.4350629e-8,0.00047804843,1.0054887e-7,0.0014136187,0.12219577,0.22095151,0.3994583],"study_design_scores_gemma":[0.00048781314,0.0002114794,0.022469299,0.00005252361,0.000021383867,0.000008944978,0.0015051829,0.0011078948,0.01026737,0.06087028,0.9028373,0.00016055732],"about_ca_topic_score_codex":0.0000043261175,"about_ca_topic_score_gemma":3.5166107e-7,"teacher_disagreement_score":0.6818858,"about_ca_system_score_codex":0.00019881582,"about_ca_system_score_gemma":0.00018675938,"threshold_uncertainty_score":0.999074},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2915058129","doi":"10.1002/eco.2080","title":"Characteristics and development trends of ecohydrology in lakes and reservoirs: Insights from bibliometrics","year":2019,"lang":"en","type":"article","venue":"Ecohydrology","topic":"Fish Ecology and Management Studies","field":"Environmental Science","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Chinese Academy of Sciences; National Natural Science Foundation of China","keywords":"Ecohydrology; Environmental science; China; Water resources; Hydrology (agriculture); Lake ecosystem; Ecosystem; Environmental resource management; Ecology; Geography; Geology","score_opus":0.010468852971162684,"score_gpt":0.21259842529032535,"score_spread":0.20212957231916268,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2915058129","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.995537,0.00006930434,0.00000485767,0.00052928395,0.00017490586,0.0000958755,0.0000056435815,0.000019352914,0.003563764],"genre_scores_gemma":[0.99858546,0.00037158455,0.00022080695,0.00033732643,0.000010031685,0.00001569417,0.000021125887,0.000008835177,0.00042911866],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9989815,0.00006626757,0.00028422364,0.00034651437,0.00008261099,0.00023893238],"domain_scores_gemma":[0.99945223,0.0002445153,0.000110136985,0.00014873443,0.000004332912,0.00004005183],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00019479252,0.000121075725,0.00031550636,0.0015668013,0.00005057546,0.000005534192,0.00013078323,0.00014729683,0.0007405346],"category_scores_gemma":[0.000054918706,0.000114458584,0.000010714627,0.0014993631,0.0002556376,0.00010916215,0.00055228785,0.0001031372,0.00022867188],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000033562606,0.000049332473,0.9967826,0.0000114694785,0.000031971264,0.000013605029,0.000643555,0.00012540926,0.00035997515,0.00007358591,0.0005104637,0.0013644575],"study_design_scores_gemma":[0.00053530675,0.00013689042,0.98727715,0.0000032827384,0.000008093373,0.0000015854396,0.000013094674,0.0014493964,0.00007859681,0.00083251175,0.009553861,0.00011024873],"about_ca_topic_score_codex":0.000113256705,"about_ca_topic_score_gemma":0.0054726894,"teacher_disagreement_score":0.0095054805,"about_ca_system_score_codex":0.000025232544,"about_ca_system_score_gemma":0.00000418166,"threshold_uncertainty_score":0.81083345},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2915246525","doi":"10.5958/2249-5576.2018.00041.9","title":"Grid Computing Research from India: A Bibliometric Assessment of Publications during 2008–2017","year":2018,"lang":"en","type":"article","venue":"International Journal of Information Dissemination and Technology","topic":"Distributed and Parallel Computing Systems","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Society of Intestinal Research","funders":"","keywords":"Grid; Library science; Computer science; Mathematics","score_opus":0.022617767123200853,"score_gpt":0.3801915876615889,"score_spread":0.35757382053838804,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2915246525","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.58497524,0.00012333177,0.40590456,0.0041378574,0.0011406584,0.000113655675,0.000031574727,0.00007134625,0.0035017626],"genre_scores_gemma":[0.98830307,0.000059439353,0.011399199,0.00001902742,0.00016438139,0.0000022252812,0.000028806078,0.000002379381,0.000021472551],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99815315,0.000057110537,0.00077254174,0.00010267985,0.00076816516,0.00014635187],"domain_scores_gemma":[0.9940319,0.00024875248,0.000926648,0.00019419745,0.0045418944,0.00005657583],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0011483446,0.0000781026,0.00015684916,0.020903083,0.00014866627,0.00030188117,0.0011613446,0.000107675834,0.00001664966],"category_scores_gemma":[0.0007319757,0.000072104514,0.000036986832,0.010702631,0.00014530019,0.0015132222,0.00032005133,0.00026454916,0.000009623906],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004459266,0.0004875721,0.1358426,0.000082404884,0.0005307108,0.000022613045,0.005817005,0.00021678794,0.006728328,0.41395575,0.022636682,0.41363496],"study_design_scores_gemma":[0.0016202105,0.0004136966,0.89179677,0.0005440694,0.000012400857,0.0006160574,0.0013216514,0.0661519,0.0079181595,0.0075942497,0.0217375,0.00027332024],"about_ca_topic_score_codex":0.000010291682,"about_ca_topic_score_gemma":0.0000010640864,"teacher_disagreement_score":0.7559542,"about_ca_system_score_codex":0.00009107554,"about_ca_system_score_gemma":0.00013109412,"threshold_uncertainty_score":0.99019414},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2922006169","doi":"10.1139/er-2018-0105","title":"Mapping sources of atmospheric pollution: integrating spatial and cluster bibliometrics","year":2019,"lang":"en","type":"article","venue":"Environmental Reviews","topic":"Atmospheric chemistry and aerosols","field":"Earth and Planetary Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Scopus; Bibliometrics; Air quality index; Air pollution; Apportionment; Cluster (spacecraft); Environmental science; Geography; Environmental resource management; Regional science; Data science; Meteorology; Political science; Computer science; Library science; Ecology","score_opus":0.012635783097151057,"score_gpt":0.20432216616760168,"score_spread":0.19168638307045063,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2922006169","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9761868,0.019884484,0.00059090566,0.00003766715,0.00008665137,0.00021160029,0.000009229781,0.000006717368,0.0029859303],"genre_scores_gemma":[0.98935777,0.0054810606,0.004295749,0.00015032869,0.000055196197,9.1340195e-7,0.000021871168,0.0000027437716,0.0006343769],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99918205,0.000046232653,0.00028533593,0.00019180159,0.00015184187,0.00014275034],"domain_scores_gemma":[0.9995742,0.000055632267,0.00017153841,0.0001327877,5.4190787e-7,0.00006532464],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00024987597,0.00011835243,0.00023582908,0.000012404223,0.000052829055,0.000020062502,0.00010203909,0.00004402561,0.003024097],"category_scores_gemma":[0.000028324353,0.00008836233,0.00006123578,0.00064865383,0.00007799099,0.00012680382,0.000024865289,0.00008975508,0.00024346003],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000050061662,0.00000915323,0.74899954,0.00004681629,0.0000054442135,7.449582e-7,0.0001273449,0.00004020324,0.0016193022,4.1707335e-7,0.00008551515,0.2490605],"study_design_scores_gemma":[0.00040445104,0.00014328511,0.8126144,0.00016498403,0.000015456268,0.0000339511,0.00077025895,0.004893515,0.0012001711,0.000028806511,0.1794293,0.0003014762],"about_ca_topic_score_codex":0.00013793336,"about_ca_topic_score_gemma":0.000010143319,"teacher_disagreement_score":0.24875902,"about_ca_system_score_codex":0.00000533024,"about_ca_system_score_gemma":0.0000046338178,"threshold_uncertainty_score":0.99788725},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2922152001","doi":"10.1016/j.afjem.2019.01.015","title":"A cross-sectional description of open access publication costs, policies and impact in emergency medicine and critical care journals","year":2019,"lang":"en","type":"article","venue":"African Journal of Emergency Medicine","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan; McMaster University","funders":"","keywords":"Medicine; Scopus; Interquartile range; Purchasing power parity; Impact factor; Turkish; MEDLINE; Family medicine; Business; Political science; Finance; Internal medicine","score_opus":0.7233153791963524,"score_gpt":0.6748636395905908,"score_spread":0.048451739605761546,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2922152001","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9682568,0.015348908,0.0003175374,0.006705908,0.002150474,0.00026862658,0.000022582468,0.0000023677624,0.0069268486],"genre_scores_gemma":[0.99095863,0.008006932,0.00006356852,0.000031264237,0.00035518006,0.000003269693,0.0000031070815,0.0000103776165,0.00056769134],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99070376,0.00037943208,0.0029445277,0.0004150748,0.0051345713,0.0004226263],"domain_scores_gemma":[0.9838287,0.0015325489,0.0014937312,0.00037809176,0.0120903365,0.0006765513],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.02209679,0.000187106,0.0007479143,0.03263143,0.00013880995,0.00047132876,0.0025191724,0.00010230172,0.016323451],"category_scores_gemma":[0.07384476,0.00010978641,0.000110959154,0.053268965,0.00041174478,0.0037185338,0.0007860601,0.00043292355,0.0000048370634],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012513265,0.00006923231,0.97763675,0.00003265746,0.000024210116,0.0000027465737,0.0012128488,0.000010201619,0.0034780195,0.00047119678,0.012970796,0.003966195],"study_design_scores_gemma":[0.0010961547,0.0014261608,0.9847072,0.00010119556,0.000017239037,0.000053530253,0.004786564,0.00018488432,0.00003908757,0.0058233603,0.0016566033,0.000108057706],"about_ca_topic_score_codex":0.0019611383,"about_ca_topic_score_gemma":0.000098331104,"teacher_disagreement_score":0.051747974,"about_ca_system_score_codex":0.00018953891,"about_ca_system_score_gemma":0.00027194968,"threshold_uncertainty_score":0.98457575},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2923089718","doi":"10.1007/s10433-019-00499-w","title":"Defining organizational contributions to sustaining an ageing workforce: a bibliometric review","year":2019,"lang":"en","type":"review","venue":"European Journal of Ageing","topic":"Retirement, Disability, and Employment","field":"Social Sciences","cited_by":51,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Workforce; Human capital; Aging in the American workforce; Economic shortage; Human resources; Human resource management; Business; Workforce planning; Economic growth; Political science; Economics; Government (linguistics); Management","score_opus":0.2467445403644406,"score_gpt":0.47974842317780564,"score_spread":0.23300388281336504,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2923089718","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000057791658,0.9812559,0.0012107616,0.00034981282,0.0008175094,0.0007615484,0.000019949308,0.000045148805,0.015481606],"genre_scores_gemma":[0.0008835963,0.99642646,0.00056922255,0.00037773015,0.0011830482,0.0000037152145,0.00003662881,0.00009483147,0.0004247685],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99209315,0.0037700795,0.0018537062,0.00039096968,0.0012323126,0.00065978424],"domain_scores_gemma":[0.99546343,0.0006400355,0.0017302741,0.0003764014,0.0012041393,0.00058570516],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.012868837,0.0003749908,0.0015485365,0.006910994,0.0007820703,0.0005150001,0.0011531985,0.00008505445,0.0003633969],"category_scores_gemma":[0.009524741,0.0003236854,0.00059165916,0.02866101,0.00010306017,0.00053042953,0.00022911809,0.00072594854,0.00032500704],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000033468916,0.00006885166,0.00018862681,0.007155682,0.0001325388,0.00046124263,0.0016027687,0.000027062608,2.3137278e-7,0.0025090287,0.0028536269,0.984997],"study_design_scores_gemma":[0.00015338979,0.000162951,0.00018405895,0.063048,0.0006740948,0.000033678087,0.000595892,5.987458e-7,5.6773146e-8,0.000034614266,0.93477553,0.00033715105],"about_ca_topic_score_codex":0.00002961237,"about_ca_topic_score_gemma":0.000014458121,"teacher_disagreement_score":0.98465985,"about_ca_system_score_codex":0.0008510488,"about_ca_system_score_gemma":0.0012887644,"threshold_uncertainty_score":0.9999215},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2927052045","doi":"10.21083/partnership.v14i1.4764","title":"Hype or Real Threat: The Extent of Predatory Journals in Student Bibliographies","year":2019,"lang":"en","type":"article","venue":"Partnership The Canadian Journal of Library and Information Practice and Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"Brandon University","funders":"","keywords":"Publishing; Library science; Computer science; Political science; Law","score_opus":0.49515394773640964,"score_gpt":0.5635849346576557,"score_spread":0.06843098692124605,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2927052045","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.91509354,0.005948665,0.000004587576,0.039488416,0.00017760012,0.00036067865,0.0000111167265,0.0000014624507,0.03891394],"genre_scores_gemma":[0.95972604,0.03891219,0.000045308596,0.00056440686,0.000054563687,0.0000020712755,6.2641277e-7,0.0000033680394,0.00069144135],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9927843,0.0013879044,0.0009520792,0.00011138253,0.0043618088,0.0004025022],"domain_scores_gemma":[0.98457724,0.012170289,0.0006204181,0.00034218314,0.0017290722,0.00056080054],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.04963524,0.00007882779,0.00020576923,0.076146945,0.0004700686,0.004634749,0.0014045604,0.00008364795,0.00046234493],"category_scores_gemma":[0.011772314,0.000032865235,0.000062446634,0.096823476,0.00053458137,0.020681743,0.00023926751,0.0009100303,0.00002076038],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007681403,0.000071151306,0.8674404,0.0000518118,0.000090868045,0.000049362745,0.010298543,0.00011154713,0.000023757622,0.026473118,0.051008962,0.04361234],"study_design_scores_gemma":[0.00055183086,0.0005790186,0.6515591,0.00006427701,0.000007639953,0.00018671637,0.046384644,0.00023113027,0.00008483701,0.0032121723,0.2970665,0.00007212087],"about_ca_topic_score_codex":0.000697358,"about_ca_topic_score_gemma":0.0003684941,"teacher_disagreement_score":0.24605756,"about_ca_system_score_codex":0.000016274487,"about_ca_system_score_gemma":0.0014517852,"threshold_uncertainty_score":0.99655193},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2928201498","doi":"","title":"A citation network analysis of mentoring in sport","year":2018,"lang":"en","type":"article","venue":"Journal of Exercise, Movement, and Sport (SCAPPS refereed abstracts repository)","topic":"Motivation and Self-Concept in Sports","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Windsor; McGill University","funders":"","keywords":"Coaching; Citation; Centrality; Sports science; Psychology; Athletes; Point (geometry); Pedagogy; Applied psychology; Library science; Political science; Computer science","score_opus":0.01876323734305667,"score_gpt":0.28034132450122273,"score_spread":0.26157808715816605,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2928201498","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9696819,0.00043835383,0.000054384815,0.000026101463,0.0012635569,0.00023217958,0.000002197199,0.000014083334,0.028287243],"genre_scores_gemma":[0.9957162,0.00023342794,0.0004694988,0.000088866545,0.00052437466,0.000008002375,0.000014597567,0.000021981281,0.0029230607],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9965863,0.000023977123,0.0020691112,0.00033646153,0.0005991311,0.00038501486],"domain_scores_gemma":[0.99666435,0.000036099227,0.0022558807,0.00036299662,0.000455825,0.00022486209],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012263964,0.00024612297,0.00091668335,0.0009381603,0.00010080447,0.000047510453,0.0002167881,0.00019637676,0.00040490142],"category_scores_gemma":[0.000015852744,0.00021899816,0.00024671454,0.0010600127,0.00010893977,0.0002960307,0.00003036221,0.00029662906,0.0000032322878],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00062907324,0.00041510395,0.9887138,0.0000342212,0.000754723,0.00022315735,0.0037311986,0.0005394852,0.00067498157,0.00087447156,0.0011867771,0.002222997],"study_design_scores_gemma":[0.0009462216,0.00025174586,0.99135154,0.00023442873,0.00060578587,0.000015702699,0.002069958,0.000019394807,0.0005551924,0.00014825852,0.0036044042,0.0001973545],"about_ca_topic_score_codex":0.0003532924,"about_ca_topic_score_gemma":0.00008853167,"teacher_disagreement_score":0.02603429,"about_ca_system_score_codex":0.0000756295,"about_ca_system_score_gemma":0.00007017354,"threshold_uncertainty_score":0.8930486},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2942000353","doi":"10.5958/2249-5576.2019.00004.9","title":"Fog Computing Research: A Scientometric Assessment of Global Publications Output During 2012–18","year":2019,"lang":"en","type":"article","venue":"International Journal of Information Dissemination and Technology","topic":"IoT and Edge/Fog Computing","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Library science; Medicine; Computer science; Operations research; Engineering","score_opus":0.027276099924247853,"score_gpt":0.39111370329240275,"score_spread":0.3638376033681549,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2942000353","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.81966656,0.00008654785,0.16632538,0.004902376,0.0029453996,0.00014041802,0.0000016515536,0.000061868035,0.005869768],"genre_scores_gemma":[0.9876579,0.000022620701,0.0120967245,0.000024799894,0.00010342321,0.0000014496288,0.0000045414745,0.0000021763321,0.00008634234],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9980751,0.000038990653,0.00070376723,0.000109472196,0.00088726,0.00018537215],"domain_scores_gemma":[0.9955808,0.00015801148,0.0007436778,0.00017852437,0.003279353,0.000059641356],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013742117,0.00008093019,0.00015233907,0.003498448,0.000108049164,0.00027011614,0.0010543423,0.00009015325,0.000011219931],"category_scores_gemma":[0.00057428173,0.000075780175,0.00004752746,0.0031831407,0.00008648802,0.0024539353,0.0004603194,0.0002552636,0.000013120337],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016953474,0.00021892946,0.21325755,0.00007382648,0.00013476993,0.0000048000884,0.0011771385,0.00020327605,0.0018136458,0.35576785,0.005596099,0.42173517],"study_design_scores_gemma":[0.0019976576,0.0003526888,0.8573825,0.00045700464,0.00001114995,0.0007740857,0.0010513435,0.10655047,0.003352309,0.00995243,0.017805,0.00031336656],"about_ca_topic_score_codex":0.0000024376216,"about_ca_topic_score_gemma":2.6551956e-7,"teacher_disagreement_score":0.6441249,"about_ca_system_score_codex":0.00024111843,"about_ca_system_score_gemma":0.00016682268,"threshold_uncertainty_score":0.31216288},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2945263934","doi":"10.1111/petr.13455","title":"Publication trends in pediatric renal transplantation: Bibliometric analysis of literature from 1950 to 2017","year":2019,"lang":"en","type":"article","venue":"Pediatric Transplantation","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Markham Stouffville Hospital; SickKids Foundation; Hospital for Sick Children; McMaster University; University of Toronto","funders":"","keywords":"Medicine; Impact factor; Bibliometrics; Transplantation; Science Citation Index; Citation; Index (typography); Observational study; MEDLINE; Retrospective cohort study; Citation analysis; Internal medicine; Demography; Library science","score_opus":0.18926390724222622,"score_gpt":0.47469290173080453,"score_spread":0.2854289944885783,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2945263934","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9857872,0.0015787006,0.0071435627,0.00044428254,0.00045723733,0.00035225347,0.0016020981,0.00003519903,0.002599471],"genre_scores_gemma":[0.9798264,0.016306907,0.0012752938,0.00007134444,0.00023194043,0.000030172965,0.0017132189,0.000016004082,0.00052875606],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9892605,0.00034990165,0.0017268091,0.0012331295,0.006829777,0.0005998805],"domain_scores_gemma":[0.99220437,0.0035208894,0.0006372034,0.00084765285,0.0023769832,0.00041293143],"candidate_categories":["bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.007215534,0.0002789267,0.0006937652,0.7959818,0.00008879007,0.0012066893,0.0017290368,0.00028690105,0.0014211364],"category_scores_gemma":[0.0016856048,0.00023320239,0.0003831992,0.9270696,0.000023898074,0.0015549144,0.00005425439,0.00034976218,0.00038045243],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019287529,0.00017233014,0.97244394,0.000043111006,0.000044601733,0.000019965162,0.0011914026,0.0020868753,0.00019772875,0.000078440346,0.0014716482,0.022057055],"study_design_scores_gemma":[0.001106329,0.00013826034,0.9945273,0.0000063330745,0.0006152596,0.0000030370525,0.000018254874,0.0027152426,0.00018082502,0.0002011657,0.00022264039,0.00026538188],"about_ca_topic_score_codex":0.00040867866,"about_ca_topic_score_gemma":0.00028654817,"teacher_disagreement_score":0.1310878,"about_ca_system_score_codex":0.00009718281,"about_ca_system_score_gemma":0.00014595159,"threshold_uncertainty_score":0.9998301},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2949231210","doi":"10.3390/ijerph16122077","title":"Global Long-Term Care Research: A Scientometric Review","year":2019,"lang":"en","type":"review","venue":"International Journal of Environmental Research and Public Health","topic":"Geriatric Care and Nursing Homes","field":"Health Professions","cited_by":42,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Tianjin Municipal Education Commission; Tianjin University; China Postdoctoral Science Foundation","keywords":"Science Citation Index; Dementia; Citation; Long-term care; Gerontology; Web of science; Medicine; Library science; Computer science; Nursing; Meta-analysis; Pathology","score_opus":0.46957921111963535,"score_gpt":0.6277644904087086,"score_spread":0.15818527928907328,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2949231210","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00024252507,0.98989415,0.000010002782,0.0036915948,0.002273503,0.0017810712,0.00052749255,0.000007944303,0.0015717419],"genre_scores_gemma":[0.0008028403,0.99618894,0.000040982723,0.00029156823,0.001305926,0.00005031243,0.00028448328,0.000038032602,0.0009969366],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.98633295,0.0047612907,0.0018545385,0.0005163885,0.0050200573,0.0015147971],"domain_scores_gemma":[0.99418026,0.001875666,0.0011446961,0.00043971356,0.0011025219,0.0012571636],"candidate_categories":["research_integrity","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.015346168,0.0002995665,0.0015568357,0.003445739,0.00074053113,0.0001567031,0.0016161641,0.0003299794,0.0011824252],"category_scores_gemma":[0.0016972002,0.0002242853,0.00039497996,0.0024839868,0.0004737881,0.00034190837,0.0009913055,0.003248942,0.00049358304],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000020601388,0.00023324728,0.0019718476,0.027987309,0.00015517205,0.00006921716,0.0003541627,7.3480626e-9,2.986613e-8,0.00009209284,0.03057587,0.93854046],"study_design_scores_gemma":[0.00048665848,0.00071737077,0.00436167,0.047315046,0.000033987635,0.00021723457,0.0013963586,6.4736845e-8,5.853816e-9,0.00005153473,0.94528157,0.00013852748],"about_ca_topic_score_codex":0.00016042129,"about_ca_topic_score_gemma":0.00004524719,"teacher_disagreement_score":0.93840194,"about_ca_system_score_codex":0.008246266,"about_ca_system_score_gemma":0.008482737,"threshold_uncertainty_score":0.99973065},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2950927168","doi":"10.48550/arxiv.0808.2517","title":"Why it has become more difficult to predict Nobel Prize winners: a bibliometric analysis of Nominees and Winners of the Chemistry and Physics Prizes (1901-2007)","year":2008,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Citation; Hierarchy; Attribution; Field (mathematics); Fragmentation (computing); Centrality; Scientific progress; Data science; Political science; Epistemology; Psychology; Library science; Computer science; Mathematics; Statistics; Law; Philosophy; Social psychology","score_opus":0.3440724253615619,"score_gpt":0.4528800998423715,"score_spread":0.10880767448080958,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2950927168","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99273264,0.0018896002,0.0007876637,0.0024399029,0.00031585322,0.0006555095,0.0006640556,0.00002367538,0.00049109495],"genre_scores_gemma":[0.9953264,0.0026741687,0.00090471725,0.00031249,0.00014059401,0.000036957834,0.00003841215,0.000031814034,0.00053447444],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9881251,0.00019276186,0.0016858305,0.0017470462,0.0074968096,0.000752449],"domain_scores_gemma":[0.98768,0.0037008442,0.0013526309,0.0022758571,0.0043515903,0.0006390812],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0053956513,0.0005548934,0.0016846093,0.19617869,0.00029328992,0.00095914886,0.0038027347,0.0004815907,0.00011698523],"category_scores_gemma":[0.02732534,0.00035662632,0.0006464553,0.74330586,0.0013578229,0.00035097188,0.005901036,0.00078575016,0.0000061392157],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000044632754,0.00027476263,0.97291976,0.00014155178,0.0007821749,0.0000068360587,0.00078949885,0.0013570299,0.0054648477,0.0000021941448,0.010762849,0.0074538365],"study_design_scores_gemma":[0.00054834795,0.00009457612,0.984043,0.000087295215,0.0003903037,0.000005046575,0.0005064869,0.006725411,0.0037662555,0.000067447705,0.00336497,0.00040088483],"about_ca_topic_score_codex":0.00060071144,"about_ca_topic_score_gemma":0.000027275593,"teacher_disagreement_score":0.5471272,"about_ca_system_score_codex":0.00012667745,"about_ca_system_score_gemma":0.00044083752,"threshold_uncertainty_score":0.9998886},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2951515884","doi":"10.1002/asi.20987","title":"A new approach for detecting scientific specialties from raw cocitation networks","year":2008,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"Complex Systems and Time Series Analysis","field":"Economics, Econometrics and Finance","cited_by":65,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Computer science; Hierarchy; Similarity (geometry); Cluster (spacecraft); Interpretation (philosophy); Data science; Information retrieval; Theoretical computer science; Data mining; Artificial intelligence; Political science","score_opus":0.031008761395710405,"score_gpt":0.2318401739998235,"score_spread":0.2008314126041131,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2951515884","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.30887485,0.0002827444,0.68876034,0.0012308678,0.00032322365,0.00022066089,0.000016788665,0.00001259398,0.00027792188],"genre_scores_gemma":[0.95552236,0.00006551385,0.043993954,0.00017917869,0.000117361,0.000007219822,0.0000021835715,0.0000032318412,0.00010901139],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991305,0.000002446994,0.0005136212,0.00010514802,0.00008876406,0.00015950335],"domain_scores_gemma":[0.9978786,0.00005644158,0.0014877918,0.0001552064,0.00038489533,0.000037080772],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010727925,0.000060059385,0.0002571358,0.00028596123,0.00087320484,0.00015606129,0.0003550833,0.000034539535,0.0000042752995],"category_scores_gemma":[0.00035241284,0.000047923622,0.0002128551,0.0018129583,0.00087260036,0.000965997,0.00006146256,0.000088127665,0.000001464148],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003051458,0.00013122377,0.06967558,0.00013724927,0.0011195707,3.1387924e-7,0.03386844,0.019835861,0.0023032965,0.34221494,0.10946371,0.4209447],"study_design_scores_gemma":[0.0020544091,0.00047089302,0.015098901,0.0000282179,0.000073783085,0.00009097741,0.03319187,0.6019155,0.0007169732,0.04247516,0.30341852,0.00046482307],"about_ca_topic_score_codex":0.00009359526,"about_ca_topic_score_gemma":0.000003922845,"teacher_disagreement_score":0.6466475,"about_ca_system_score_codex":0.00008569341,"about_ca_system_score_gemma":0.00011656351,"threshold_uncertainty_score":0.67160743},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2952696989","doi":"10.1002/asi.20997","title":"National and international dimensions of the Triple Helix in Japan: University–industry–government versus international coauthorship relations","year":2009,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"University-Industry-Government Innovation Models","field":"Business, Management and Accounting","cited_by":153,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Institute of Informatics","keywords":"Triple helix; Government (linguistics); International relations; Political science; Library science; Value (mathematics); Regional science; Sociology; Computer science; Politics; Law","score_opus":0.023866743336898272,"score_gpt":0.2543653634252096,"score_spread":0.23049862008831132,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2952696989","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.945246,0.0000034757763,0.00045839883,0.050858155,0.00040097776,0.00014294873,0.000014316647,0.000008377137,0.0028673734],"genre_scores_gemma":[0.99791306,0.000014824554,0.0006096022,0.0012395309,0.00004488561,4.2241066e-7,9.098535e-7,0.0000013833212,0.00017537714],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988476,0.000004755106,0.0002561871,0.000068875124,0.0007249579,0.00009762246],"domain_scores_gemma":[0.99812084,0.000049423714,0.00095047726,0.000081619226,0.00078807917,0.000009561675],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007972515,0.000060061313,0.000093882474,0.00028753607,0.00025812225,0.000063829044,0.0005328443,0.00008333671,0.000007570384],"category_scores_gemma":[0.00063376804,0.000041824373,0.00006445764,0.0017326757,0.0005951077,0.002399083,0.00022530752,0.00033837804,7.023477e-7],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044253975,0.00021716885,0.14617392,0.00001651511,0.00019040013,4.90529e-7,0.0010082523,0.0026543092,0.0037274163,0.76849735,0.035855632,0.041215993],"study_design_scores_gemma":[0.0070220567,0.00022446635,0.3596581,0.00016139186,0.00014096801,0.000040281597,0.05768771,0.06901142,0.00062577584,0.013913797,0.49110052,0.00041350932],"about_ca_topic_score_codex":0.000010334142,"about_ca_topic_score_gemma":0.0000030541776,"teacher_disagreement_score":0.75458354,"about_ca_system_score_codex":0.00033516734,"about_ca_system_score_gemma":0.00014770374,"threshold_uncertainty_score":0.21926989},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2953604448","doi":"","title":"The influence of learning algorithms for Bayesian Networks on predictions: A citation analysis study case","year":2019,"lang":"en","type":"article","venue":"Research Repository (Delft University of Technology)","topic":"Bayesian Modeling and Causal Inference","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Computer science; Bayesian probability; Bayesian network; Machine learning; Artificial intelligence; Cluster analysis; Variable-order Bayesian network; Algorithm; Data mining; Bayesian inference","score_opus":0.028581890462339755,"score_gpt":0.2980044318096698,"score_spread":0.26942254134733,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2953604448","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.60411614,0.000030807278,0.39517713,0.00013197916,0.000026053644,0.00032650295,8.2644954e-7,0.00007284699,0.000117716736],"genre_scores_gemma":[0.9945458,0.000027789305,0.0049442574,0.0000011141993,0.0000071535646,0.000003969674,6.766043e-7,0.0000038192097,0.00046542377],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9985414,0.00023918478,0.0001960042,0.00036839236,0.0003899234,0.00026506613],"domain_scores_gemma":[0.9977561,0.00048990175,0.00016559879,0.000698909,0.0008385412,0.00005090793],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012263868,0.00008251773,0.0001926617,0.0008875513,0.00078249146,0.00004238086,0.0009604692,0.0001389819,6.4567627e-7],"category_scores_gemma":[0.00016410377,0.00007564655,0.00009649744,0.0024616579,0.00027095756,0.00019293273,0.00026494934,0.0004892555,0.0000015736096],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018529447,0.00052586664,0.051867932,0.000050208066,0.0010030594,0.0002571019,0.001595438,0.8856499,0.0019488381,0.028622475,0.000079868056,0.028214004],"study_design_scores_gemma":[0.0003438492,0.0014975113,0.005309861,0.00003887558,0.00006587146,0.000022762466,0.0055638985,0.9860824,0.00033233952,0.0005693847,0.00008815019,0.00008514195],"about_ca_topic_score_codex":0.0002932984,"about_ca_topic_score_gemma":0.000026632144,"teacher_disagreement_score":0.39042968,"about_ca_system_score_codex":0.00007876163,"about_ca_system_score_gemma":0.000107355074,"threshold_uncertainty_score":0.6018371},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2955631702","doi":"10.1111/cid.12810","title":"The funding sources of implantology research in the period 2008‐2017: A bibliometric analysis","year":2019,"lang":"en","type":"article","venue":"Clinical Implant Dentistry and Related Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Dentistry; Bibliometrics; Citation; Scopus; Government (linguistics); Citation analysis; Medicine; Subject (documents); Commercialization; Library science; Dental research; Web of science; Period (music); MEDLINE; Political science; Computer science","score_opus":0.7769852019218009,"score_gpt":0.6830523692882974,"score_spread":0.0939328326335035,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2955631702","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9802386,0.0100139985,0.000017385759,0.0026388979,0.0018286719,0.00063703855,0.00005770428,0.000008660878,0.0045590387],"genre_scores_gemma":[0.9862344,0.009039131,0.000019444506,0.00001837126,0.000074760486,0.000020924943,0.00001036174,0.000010085862,0.0045725033],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9670802,0.011412159,0.002659802,0.0014775399,0.015131107,0.0022392012],"domain_scores_gemma":[0.8515741,0.1417635,0.00043683185,0.0019447505,0.0037231464,0.000557624],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.2984419,0.00019108692,0.00082998304,0.39707202,0.0015513289,0.0034695526,0.006026392,0.0006421983,0.0003884048],"category_scores_gemma":[0.06210956,0.00008957701,0.00045028012,0.81475496,0.0029855734,0.00033615375,0.0021956787,0.0041103396,0.0010136596],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002522648,0.0003655972,0.90347123,0.000028610684,0.00039132053,0.0032578383,0.00029093487,0.000021501926,0.00013704941,0.002036522,0.020791277,0.068955846],"study_design_scores_gemma":[0.00092525943,0.0006739248,0.95808285,0.000032044456,0.000039583025,0.004971016,0.005665898,0.0028993671,0.000039022674,0.008317101,0.018189145,0.00016481089],"about_ca_topic_score_codex":0.00037968231,"about_ca_topic_score_gemma":0.0001791257,"teacher_disagreement_score":0.41768295,"about_ca_system_score_codex":0.00005269462,"about_ca_system_score_gemma":0.00039726286,"threshold_uncertainty_score":0.99976414},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2955873301","doi":"10.1007/s11192-019-03146-4","title":"The citation advantage of foreign language references for Chinese social science papers","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":24,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Office for Philosophy and Social Sciences; China Scholarship Council","keywords":"Citation; Internationalization; Foreign language; Citation index; Impact factor; Scientific communication; Sociology of scientific knowledge; Social science; Consumption (sociology); Artifact (error); Scientometrics; Diversity (politics); Sociology; Computer science; Library science; Political science; Law; Business","score_opus":0.34542183294758166,"score_gpt":0.5814282516665357,"score_spread":0.236006418718954,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2955873301","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9018667,0.00061490666,0.0013327581,0.00024741524,0.0014756516,0.0006967943,0.00009095738,0.000022638364,0.09365218],"genre_scores_gemma":[0.9944675,0.00007381278,0.0010178257,0.000058068163,0.00007487888,0.000019324729,0.000005186096,0.000010364544,0.0042730127],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.97952145,0.00012466194,0.00082433457,0.00090268505,0.017645406,0.000981442],"domain_scores_gemma":[0.9799633,0.011449912,0.0006208788,0.0008780792,0.0067739016,0.00031390542],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.062178716,0.00017452335,0.000344946,0.08917785,0.0011526555,0.0027499814,0.0052497867,0.00009178874,0.00016474965],"category_scores_gemma":[0.11688691,0.00009311091,0.00023739002,0.5141261,0.0011355059,0.0010036702,0.00073695445,0.00019063121,0.00012322004],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000104797145,0.00017411394,0.38074687,0.000032231805,0.000018574585,0.0000012871133,0.0019325325,0.000033459808,0.0406542,0.064988695,0.0021479782,0.5091652],"study_design_scores_gemma":[0.0017745101,0.0008786184,0.8985406,0.000012157224,0.000013497525,0.0000036860718,0.019622473,0.010949568,0.007718425,0.025862094,0.03410206,0.00052229466],"about_ca_topic_score_codex":0.00001615571,"about_ca_topic_score_gemma":0.000007617001,"teacher_disagreement_score":0.5177937,"about_ca_system_score_codex":0.00017799286,"about_ca_system_score_gemma":0.00050655234,"threshold_uncertainty_score":0.99828523},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2961163828","doi":"10.12688/f1000research.18757.1","title":"Understanding the contribution of UK public health research to clinical guidelines: a bibliometric analysis","year":2019,"lang":"en","type":"preprint","venue":"F1000Research","topic":"Clinical practice guidelines implementation","field":"Medicine","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Université du Québec à Montréal","funders":"Research Councils UK; National Cancer Institute; National Medical Research Council; British Heart Foundation; National Institutes of Health; National Health and Medical Research Council; Cancer Research UK; Canadian Institutes of Health Research; National Institute for Health and Care Research; Department of Health and Social Care; Research England; Government of the United Kingdom; Medical Research Council; Economic and Social Research Council; Sanofi; Pfizer","keywords":"Open peer review; Plant biology; Public health; Bibliometrics; Medicine; Physiology; Alternative medicine; Biology; Library science; Pathology; Computer science","score_opus":0.9567812570672606,"score_gpt":0.7309965427940688,"score_spread":0.2257847142731918,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2961163828","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20052904,0.0022428667,0.35716018,0.4268964,0.0012756626,0.010437296,0.00034957242,0.00011098639,0.0009979795],"genre_scores_gemma":[0.9865882,0.0039052696,0.0030684501,0.0032725278,0.0011039999,0.00019431733,0.001133981,0.000069655944,0.0006635929],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9754037,0.0074352836,0.006309694,0.0015458255,0.007531891,0.0017735859],"domain_scores_gemma":[0.94235444,0.033430267,0.001531588,0.0036570001,0.017580044,0.0014466444],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.15224668,0.00034399104,0.0018833493,0.10451279,0.0003732524,0.0003698198,0.0013995174,0.0006165426,0.0011130653],"category_scores_gemma":[0.24695164,0.0002500064,0.0009989891,0.20617378,0.00057532586,0.00014133586,0.0037264868,0.004930948,0.00035244337],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011448181,0.00079078256,0.6478381,0.00049048924,0.0039787446,0.000020137591,0.0003273795,0.0012537417,0.000056557594,0.0020727175,0.3067934,0.035233114],"study_design_scores_gemma":[0.0062421267,0.0050380738,0.8359742,0.0010433359,0.0017740459,0.000021070819,0.009457692,0.0530222,0.000046375488,0.006323485,0.080363944,0.00069343735],"about_ca_topic_score_codex":0.00368134,"about_ca_topic_score_gemma":0.0009754884,"teacher_disagreement_score":0.78605914,"about_ca_system_score_codex":0.002372622,"about_ca_system_score_gemma":0.009783024,"threshold_uncertainty_score":0.99999523},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2966647780","doi":"","title":"Identifying Information Retrieval Research Trends Using Author Co-citation Network","year":2019,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Health Education and Validation","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Co-citation; Scopus; Computer science; Field (mathematics); Information retrieval; Subject (documents); Library science; Data science; World Wide Web; MEDLINE; Political science; Mathematics","score_opus":0.7637289468299195,"score_gpt":0.7338980042026019,"score_spread":0.02983094262731767,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2966647780","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95994794,0.0009152611,0.0008955091,0.0012584772,0.0027189383,0.00074115157,0.0000117705695,0.000045569064,0.033465356],"genre_scores_gemma":[0.99549043,0.001314974,0.00051550433,0.00026681984,0.00067891134,0.000012785828,0.000054566557,0.00001905673,0.0016469575],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99484235,0.0016548893,0.000931872,0.00023740681,0.0017586095,0.00057487155],"domain_scores_gemma":[0.99668413,0.00070970424,0.0008804287,0.00027506673,0.0011237684,0.00032690997],"candidate_categories":["sts","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.015184237,0.00013634446,0.00033092004,0.0017116087,0.0013901646,0.003029986,0.0012737472,0.00017462623,0.015092221],"category_scores_gemma":[0.0016045114,0.00014368887,0.000090452144,0.0044010677,0.0001390322,0.00891892,0.00019008988,0.00050149637,0.00018095241],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004077839,0.00019767588,0.68212944,0.00027857162,0.000095118274,0.0000021324734,0.03926974,0.0014842212,0.010361836,0.0086900005,0.16477236,0.0923111],"study_design_scores_gemma":[0.00052670564,0.000013983293,0.7679486,0.0005038853,0.000029438455,0.0000017635443,0.007959135,0.0009788292,0.0013567447,0.008579542,0.21172884,0.00037252557],"about_ca_topic_score_codex":0.00430077,"about_ca_topic_score_gemma":0.00014976335,"teacher_disagreement_score":0.09193858,"about_ca_system_score_codex":0.00053779,"about_ca_system_score_gemma":0.0010003797,"threshold_uncertainty_score":0.9999099},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2970371491","doi":"10.1016/j.vhri.2019.04.008","title":"Review of Pharmacoeconomic Studies in Russian Cancer Research: An Outside View","year":2019,"lang":"en","type":"review","venue":"Value in Health Regional Issues","topic":"Health Systems, Economic Evaluations, Quality of Life","field":"Economics, Econometrics and Finance","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre for Excellence in Mining Innovation; University of Toronto; Toronto Public Health","funders":"","keywords":"Scope (computer science); Medicine; Quality (philosophy); Economic evaluation; Medical physics; Actuarial science; Family medicine; Pathology; Computer science; Business","score_opus":0.9485554896044527,"score_gpt":0.7168480817941256,"score_spread":0.23170740781032706,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2970371491","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000061523446,0.95517826,9.874636e-7,0.0379234,0.0011541026,0.0048727915,0.00046361788,0.000026158019,0.00031915063],"genre_scores_gemma":[0.000011723264,0.98381066,0.0004353978,0.012814567,0.00075145875,0.0014069281,0.00014978688,0.00012308205,0.00049636676],"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","domain_scores_codex":[0.9756765,0.0063502877,0.014976976,0.001554252,0.00028816433,0.0011538457],"domain_scores_gemma":[0.98697394,0.0040015127,0.0072862976,0.0012326121,0.0001507744,0.00035484807],"candidate_categories":["metaresearch","metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.072832994,0.0006296812,0.010079605,0.0022263674,0.00012769736,0.000034545956,0.0010518836,0.0003857308,0.0003594017],"category_scores_gemma":[0.0025105742,0.0006985432,0.00049993506,0.0011030548,0.0003157976,0.0005557659,0.00018376524,0.0011967963,0.0014726153],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000008536728,0.00018941505,0.00028865342,0.8856268,0.00018195853,0.0000039521237,0.0023139182,0.00007485492,4.809149e-9,0.03820949,0.03604926,0.037053153],"study_design_scores_gemma":[0.00025979147,0.00008037603,0.000048745645,0.2996893,0.000016908672,0.000006329153,0.00024779743,0.00006927364,1.6011803e-8,0.002427494,0.69679606,0.00035792537],"about_ca_topic_score_codex":0.025080146,"about_ca_topic_score_gemma":0.0009936125,"teacher_disagreement_score":0.6607468,"about_ca_system_score_codex":0.0049924375,"about_ca_system_score_gemma":0.0031475052,"threshold_uncertainty_score":0.9995466},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2971101863","doi":"10.1007/s11192-019-03215-8","title":"Cooperation, scale-invariance and complex innovation systems: a generalization","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"Innovation and Knowledge Management","field":"Business, Management and Accounting","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Generalization; Scale (ratio); Scale invariance; Mathematical economics; Mathematics; Computer science; Statistics; Geography; Cartography; Mathematical analysis","score_opus":0.03896026622934054,"score_gpt":0.2602012405683491,"score_spread":0.22124097433900855,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2971101863","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5273571,0.00014914191,0.034213327,0.0010716247,0.004629312,0.0012640674,0.0000037036425,0.00028983862,0.43102187],"genre_scores_gemma":[0.9912259,0.0000050477956,0.00042631713,0.0025272998,0.00038334928,0.000013003191,0.00013118559,0.000012870076,0.005275024],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987905,0.000008512131,0.00031413633,0.00028266755,0.00041861806,0.00018553615],"domain_scores_gemma":[0.99825656,0.000012941299,0.00019766319,0.00018784482,0.001337357,0.000007650194],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0010462706,0.00010712009,0.00012091215,0.003540655,0.00018476688,0.0009440291,0.00014312341,0.000041711806,0.00012221519],"category_scores_gemma":[0.00021870028,0.00010264906,0.000011264917,0.023049098,0.000033477267,0.0008616677,0.0001338184,0.000050465416,0.0006625837],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000003518695,0.000031994357,0.04329791,0.00016318588,0.000005201914,2.7374236e-7,0.000028546392,0.00022306676,0.0008739779,0.94648594,0.0073126974,0.0015736651],"study_design_scores_gemma":[0.0012077633,0.000017793092,0.046739355,0.000048019152,0.000020998768,0.0000015622048,0.00034734898,0.3271549,0.000041859275,0.0010794132,0.6229269,0.0004140869],"about_ca_topic_score_codex":0.00002815178,"about_ca_topic_score_gemma":0.0000075827857,"teacher_disagreement_score":0.94540656,"about_ca_system_score_codex":0.000050654624,"about_ca_system_score_gemma":0.00001586177,"threshold_uncertainty_score":0.9977165},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2974322369","doi":"10.56059/jl4d.v6i2.331","title":"Early Years of the Journal of Learning for Development: A Combination of Bibliometrics and Thematic Analysis","year":2019,"lang":"en","type":"article","venue":"Journal of Learning for Development","topic":"Online and Blended Learning","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Carl von Ossietzky Universität Oldenburg","keywords":"Bibliometrics; Thematic map; Library science; Data science; Computer science; Information retrieval; Geography; Cartography","score_opus":0.03112556393741915,"score_gpt":0.3210867008059854,"score_spread":0.2899611368685662,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2974322369","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9869976,0.0002694022,0.012126504,0.00013649238,0.00016719023,0.00020425938,3.1068404e-7,0.0000031574446,0.00009506487],"genre_scores_gemma":[0.9650347,0.00012044483,0.034080457,0.000004342716,0.000041038136,0.0000013643169,0.00000109198,0.000011886859,0.0007046273],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99750745,0.00030593248,0.0010369647,0.00008256111,0.00085935154,0.00020774745],"domain_scores_gemma":[0.99486434,0.0010371403,0.0027714863,0.000057775553,0.0011964174,0.000072827526],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0073745046,0.000097404074,0.00049648335,0.0030080136,0.0002506488,0.00004539145,0.00030518006,0.00008651785,0.000024744255],"category_scores_gemma":[0.0024335703,0.00007469575,0.00022187723,0.0039394153,0.00006743738,0.0002474577,0.00004991523,0.00035861897,6.320687e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021892469,0.00026473834,0.62959206,0.00044088936,0.0018575157,0.0000014410134,0.06771162,0.012036723,0.0018178485,0.00053658214,0.00006508635,0.28545657],"study_design_scores_gemma":[0.005086403,0.0018610096,0.7944954,0.0014584564,0.0013446109,0.00002296664,0.038950335,0.0014997423,0.005378342,0.00055153074,0.14887872,0.0004724503],"about_ca_topic_score_codex":0.000007912773,"about_ca_topic_score_gemma":0.0000123154505,"teacher_disagreement_score":0.28498414,"about_ca_system_score_codex":0.0001113601,"about_ca_system_score_gemma":0.0010346443,"threshold_uncertainty_score":0.30460045},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2979130558","doi":"10.5151/9788580391619-02","title":"Análise Bibliométrica do Comportamento do Consumidor","year":2016,"lang":"pt","type":"book-chapter","venue":"Editora Blucher eBooks","topic":"Business and Management Studies","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Ontario","funders":"","keywords":"Business; Political science; Economics","score_opus":0.09864069129335044,"score_gpt":0.34111755356827245,"score_spread":0.242476862274922,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2979130558","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00018514259,0.00558416,0.0017667976,0.0025289375,0.08503764,0.0017606437,0.0004770447,0.00029874375,0.9023609],"genre_scores_gemma":[0.08124658,0.0014741316,0.00021737823,0.00089111045,0.025973681,0.0001698375,0.00001677615,0.00025674814,0.88975376],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9886727,0.0001699682,0.0024106563,0.0026785834,0.0047504045,0.0013176989],"domain_scores_gemma":[0.990961,0.0016421966,0.002075852,0.0027878687,0.0018917365,0.0006413108],"candidate_categories":["metaepi_narrow","scholarly_communication","insufficient_payload"],"consensus_categories":["metaepi_narrow","insufficient_payload"],"category_scores_codex":[0.0024553337,0.0016684086,0.0021857303,0.0056596296,0.00095919904,0.0019754225,0.002611604,0.0010769527,0.015720632],"category_scores_gemma":[0.00048774658,0.0011696378,0.0008773922,0.0008504649,0.0017584759,0.00048766236,0.0024343478,0.00085196114,0.010088342],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017684193,0.00011617457,0.0007829473,0.00011028127,0.0007826467,0.0003215516,0.0004928891,5.7125635e-7,0.00007012982,0.04319145,0.71825576,0.23569873],"study_design_scores_gemma":[0.0016674028,0.00014560575,0.00030901824,0.0008244498,0.00041444585,0.0000064580863,0.000359746,0.000010545703,0.000028387905,0.020837957,0.9738871,0.001508912],"about_ca_topic_score_codex":0.00004543831,"about_ca_topic_score_gemma":0.00001104365,"teacher_disagreement_score":0.2556313,"about_ca_system_score_codex":0.00020743163,"about_ca_system_score_gemma":0.00030914065,"threshold_uncertainty_score":0.9996063},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2979198997","doi":"10.1093/reseval/rvz024","title":"Do we need a book citation index for research evaluation?","year":2019,"lang":"en","type":"article","venue":"Research Evaluation","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Citation; Index (typography); Citation index; Argument (complex analysis); Sociology; Library science; Social science; Computer science; World Wide Web; Medicine","score_opus":0.9412224849470798,"score_gpt":0.7642154891255654,"score_spread":0.17700699582151447,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2979198997","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8609644,0.017830778,0.0075164405,0.02813098,0.0018485655,0.024961956,0.000092803086,0.0000820624,0.05857202],"genre_scores_gemma":[0.98429847,0.0006454031,0.00056053075,0.000059414757,0.00035502226,0.0018004513,0.00006227393,0.00003878331,0.012179659],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.84470934,0.014333956,0.0016577301,0.0025044307,0.13411953,0.0026750066],"domain_scores_gemma":[0.8360946,0.043446597,0.00031195904,0.0022528623,0.11715275,0.00074118667],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.5927599,0.00021778983,0.00041102836,0.11170551,0.001184962,0.005648504,0.0031656425,0.00034628352,0.011677729],"category_scores_gemma":[0.3318739,0.00017155371,0.00022843033,0.20163214,0.00038743435,0.002057452,0.00091896945,0.0011711334,0.0066332226],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003589586,0.00029700398,0.017754134,0.000040106166,0.000038025602,9.966508e-7,0.00094120635,0.0018385501,0.0054839216,0.0057954006,0.20874706,0.75870466],"study_design_scores_gemma":[0.002722943,0.00077199814,0.040202692,0.000049637852,0.000010697236,0.0000016623088,0.003318897,0.61771464,0.00075727975,0.23427902,0.09995902,0.00021151458],"about_ca_topic_score_codex":0.00012367575,"about_ca_topic_score_gemma":0.000038456892,"teacher_disagreement_score":0.7584931,"about_ca_system_score_codex":0.0021027925,"about_ca_system_score_gemma":0.0045259246,"threshold_uncertainty_score":0.99538374},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W2980868182","doi":"10.1051/matecconf/201929403014","title":"Railway transportation of dangerous goods: a bibliometric aspect","year":2019,"lang":"en","type":"article","venue":"MATEC Web of Conferences","topic":"Economic and Technological Systems Analysis","field":"Business, Management and Accounting","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Fundamental Research Funds for the Central Universities; Social Sciences and Humanities Research Council of Canada; National Research Council Canada; Institut d'Estudis Catalans; Alberta Innovates; Centre National pour la Recherche Scientifique et Technique; Mitacs; Bulgarian National Science Fund; Dwight D. Eisenhower Army Medical Center; City, University of London; Northwestern University; Ministry of Infrastructure, Transport and Networks; Generalitat Valenciana; Natural Sciences and Engineering Research Council of Canada; Rutgers, The State University of New Jersey; Office of the Assistant Secretary for Research and Technology; National University Rail Center; National Natural Science Foundation of China; Northwestern University Transportation Center; U.S. Department of Transportation","keywords":"Telematics; Scopus; Transport engineering; Citation; Dangerous goods; Data science; Field (mathematics); Empirical research; Thematic map; Computer science; Business; Engineering; World Wide Web; Political science; Geography; Telecommunications","score_opus":0.021033364441517113,"score_gpt":0.2141361653291731,"score_spread":0.19310280088765597,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2980868182","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9171125,0.00014702644,0.00012496315,0.00008997447,0.00008134434,0.000118814765,0.000005809294,0.000055657438,0.082263865],"genre_scores_gemma":[0.9996383,0.000029288389,0.00005039038,0.000026203192,0.000040433242,0.0000056626845,0.000013663442,0.000005038301,0.00019098414],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99922,0.000003957009,0.00037092317,0.00015449179,0.00013931353,0.00011131089],"domain_scores_gemma":[0.9992844,0.000033045362,0.00042945138,0.0001487445,0.00009938472,0.0000049578507],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00021890225,0.000090754285,0.00034207653,0.004984636,0.000017386643,0.000049308634,0.00022786012,0.00007271088,0.0018514994],"category_scores_gemma":[0.000024955554,0.0000694899,0.00009233833,0.007368247,0.00004858124,0.00027531246,0.000017237398,0.000044521268,0.00013914287],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000124619855,0.000052029052,0.80699074,0.00032068082,0.000072521936,8.8933626e-7,0.000009188506,0.00001781831,0.0020573966,0.1776831,0.0001623579,0.012620799],"study_design_scores_gemma":[0.00108569,0.00010714235,0.91196436,0.0002178649,0.00020523669,5.184097e-7,0.00124106,0.0022089703,0.0035105587,0.013916081,0.06510591,0.00043658412],"about_ca_topic_score_codex":0.0021645918,"about_ca_topic_score_gemma":0.0002500206,"teacher_disagreement_score":0.16376701,"about_ca_system_score_codex":0.0000047867766,"about_ca_system_score_gemma":0.000029695106,"threshold_uncertainty_score":0.9990609},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2984651096","doi":"10.1080/09588221.2019.1681465","title":"An examination of the studies on foreign language teaching in pre-school education: a bibliometric mapping analysis","year":2019,"lang":"en","type":"article","venue":"Computer Assisted Language Learning","topic":"Digital literacy in education","field":"Computer Science","cited_by":98,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Foreign language; Citation; Psycholinguistics; Citation analysis; Applied linguistics; Language education; Mathematics education; Information and Communications Technology; Educational technology; Computer science; Field (mathematics); Psychology; Pedagogy; Library science; Linguistics; World Wide Web; Cognition","score_opus":0.014841125603407493,"score_gpt":0.3143350200127951,"score_spread":0.2994938944093876,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2984651096","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.962738,0.00037060422,0.031862058,0.00004812648,0.00048704888,0.00027981293,9.868721e-7,0.00010722609,0.0041061314],"genre_scores_gemma":[0.98122245,0.00000131246,0.01790286,0.00021620594,0.00012109121,0.000022958187,0.000026789055,0.000012933033,0.000473423],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99781346,0.00062302,0.00044384346,0.00047058953,0.00043128783,0.00021781238],"domain_scores_gemma":[0.99816406,0.00040569165,0.0004013331,0.0008630939,0.000104439176,0.00006140777],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0010982741,0.00017192782,0.0002994731,0.010900652,0.00009514178,0.00032349725,0.000934427,0.000057505145,0.000010769578],"category_scores_gemma":[0.00036796785,0.00013827274,0.00012595204,0.02500297,0.000022256625,0.0010338429,0.00023300694,0.00041177016,0.000013513413],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000003895831,0.00041299252,0.14731745,0.00010838394,0.00017398687,0.0000040462037,0.09039904,0.028182765,0.0009806391,0.0025804425,0.000025539583,0.72981083],"study_design_scores_gemma":[0.00017028648,0.00010430139,0.8559509,0.000199134,0.000026479163,0.000005519863,0.008730582,0.1343261,0.00022431364,0.00003217583,0.000063798616,0.00016643574],"about_ca_topic_score_codex":0.000052913085,"about_ca_topic_score_gemma":0.0000147729015,"teacher_disagreement_score":0.72964436,"about_ca_system_score_codex":0.00018436422,"about_ca_system_score_gemma":0.00008092297,"threshold_uncertainty_score":0.99572104},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2986816625","doi":"10.1007/s11192-019-03285-8","title":"The national system of researchers in Mexico: implications of publication incentives for researchers in social sciences","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"Science, Technology, and Education in Latin America","field":"Social Sciences","cited_by":20,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Université du Québec à Montréal","funders":"Consejo Nacional de Ciencia y Tecnología","keywords":"Incentive; Promotion (chess); SNi; Revocation; Political science; Sociology; Engineering; Economics; Law; Chemistry","score_opus":0.3660182259990773,"score_gpt":0.5331802589625843,"score_spread":0.16716203296350696,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2986816625","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7515376,0.00035475555,0.00035794833,0.035104934,0.00068740017,0.001973698,0.0000483727,0.000052248724,0.20988305],"genre_scores_gemma":[0.99831176,0.0000593767,0.00095791183,0.000031816904,0.000030696883,0.00010366473,0.0000025702732,0.0000032109092,0.0004989891],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9968878,0.00027848504,0.00037394706,0.0003029628,0.0016509026,0.00050590915],"domain_scores_gemma":[0.99574393,0.0023533013,0.0002998389,0.0001498462,0.0013923392,0.000060763603],"candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.021252371,0.000053977117,0.00013138658,0.0037229126,0.0007373274,0.00012717505,0.0013171823,0.00008916166,0.000020294525],"category_scores_gemma":[0.016369717,0.000044841614,0.000046731202,0.037386656,0.0034948688,0.0003790136,0.00010082518,0.00013758149,0.0000059923586],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010625704,0.00010102023,0.3793327,0.000025492049,0.0000034659859,8.078279e-9,0.006991916,0.000049884366,0.00036967103,0.6034181,0.0026782127,0.0070189144],"study_design_scores_gemma":[0.0006145745,0.00014101536,0.81205875,0.00004978304,0.0000027103836,2.0854695e-7,0.12980972,0.0019397804,0.00065529323,0.018448452,0.036105484,0.0001742262],"about_ca_topic_score_codex":0.00047745154,"about_ca_topic_score_gemma":0.00022879629,"teacher_disagreement_score":0.58496964,"about_ca_system_score_codex":0.0007141673,"about_ca_system_score_gemma":0.0016424361,"threshold_uncertainty_score":0.99921703},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2987353874","doi":"10.5539/elt.v12n12p59","title":"Study of Corpus’ Influences in EAP Research (2009-2018): A Bibliometric Analysis in CiteSpace","year":2019,"lang":"en","type":"article","venue":"English Language Teaching","topic":"Second Language Learning and Teaching","field":"Arts and Humanities","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Division of Undergraduate Education; Fundamental Research Funds for the Central Universities; China University of Geosciences; China University of Geosciences, Beijing","keywords":"English for academic purposes; Citation; Genre analysis; Corpus linguistics; Psychology; Literacy; Linguistics; Library science; Sociology; Computer science; Mathematics education; Pedagogy; Natural language processing","score_opus":0.04484412301575874,"score_gpt":0.3328953624065002,"score_spread":0.28805123939074145,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2987353874","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9635747,0.0018850454,0.0000017640504,0.0000134746,0.00023232093,0.0003661329,0.000007519164,0.0000913193,0.033827744],"genre_scores_gemma":[0.996765,0.0000060800094,0.00004591717,0.000036976322,0.00024088148,0.000019475427,0.000014749683,0.000027872935,0.0028430219],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9964448,0.0013613644,0.00052641105,0.0004639944,0.00067371794,0.00052966963],"domain_scores_gemma":[0.9979489,0.0012059513,0.00017453426,0.00052060076,0.00008597515,0.000064054635],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0055971514,0.00019262345,0.0005351458,0.07010227,0.00016274172,0.0002812111,0.0004615375,0.00007588743,0.0012957542],"category_scores_gemma":[0.0023647828,0.0001714392,0.000103307,0.023446776,0.00007414759,0.00046498395,0.0001440834,0.0017265774,0.00003390266],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001530215,0.0003437874,0.3977681,0.000026501282,0.00008422355,0.0000686991,0.5922461,0.00059966213,0.000085984604,0.00086368783,0.000022202343,0.0078757005],"study_design_scores_gemma":[0.0011777596,0.0003322217,0.081323914,0.00010732689,0.000051136783,8.1329694e-7,0.91461754,0.00035515285,0.000023428931,0.000015620368,0.0017360366,0.00025904065],"about_ca_topic_score_codex":0.054966945,"about_ca_topic_score_gemma":0.046546683,"teacher_disagreement_score":0.32237142,"about_ca_system_score_codex":0.00009979627,"about_ca_system_score_gemma":0.000031366984,"threshold_uncertainty_score":0.9996172},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2987668259","doi":"10.1007/s11192-019-03291-w","title":"Evaluating medical conferences: the emerging need for a quality metric","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":11,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Alberta Hip and Knee Clinic; University of Calgary","funders":"","keywords":"Impact factor; Scope (computer science); Quality (philosophy); Metric (unit); Publishing; Observational study; Computer science; Library science; Ranking (information retrieval); Political science; Medicine; Business; Information retrieval","score_opus":0.8632070423881478,"score_gpt":0.7154768984215607,"score_spread":0.14773014396658712,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2987668259","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9001105,0.0022637183,0.061688304,0.0062394417,0.0057880296,0.0021358018,0.00008964359,0.000100133155,0.021584466],"genre_scores_gemma":[0.9903953,0.00010061093,0.0031072923,0.00080921926,0.00020031589,0.000067092806,0.000009476813,0.000020811,0.005289852],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.94184977,0.0010309861,0.001922253,0.0017154092,0.051814627,0.0016669576],"domain_scores_gemma":[0.85030746,0.11738067,0.001814633,0.004292361,0.023896141,0.0023087128],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.25881523,0.00029724822,0.000694419,0.1489679,0.0008757854,0.005164391,0.008702159,0.00026744057,0.0037286726],"category_scores_gemma":[0.5399934,0.00016397025,0.0004924886,0.63241637,0.00042524864,0.0007880301,0.0017442633,0.0006269276,0.0013563843],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000057222805,0.0002451529,0.288818,0.000030487234,0.000047020792,0.0000025957752,0.00042357654,0.0004342368,0.00041838916,0.023080721,0.013487978,0.6729546],"study_design_scores_gemma":[0.004200752,0.0011559937,0.2884158,0.000039336097,0.000038644233,0.000016942202,0.0052769696,0.4901171,0.00079091394,0.03399798,0.17495292,0.0009966604],"about_ca_topic_score_codex":0.00022195427,"about_ca_topic_score_gemma":0.000024696876,"teacher_disagreement_score":0.67195797,"about_ca_system_score_codex":0.0002707261,"about_ca_system_score_gemma":0.0017491643,"threshold_uncertainty_score":0.9994212},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2989484692","doi":"10.1007/s11192-019-03301-x","title":"Do articles in open access journals have more frequent altmetric activity than articles in subscription-based journals? An investigation of the research output of Finnish universities","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":105,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Turun Yliopisto","keywords":"Altmetrics; Discipline; Scientometrics; Bibliometrics; Social media; Library science; Computer science; Sociology; Social science; World Wide Web","score_opus":0.838152280754456,"score_gpt":0.6455667087672443,"score_spread":0.1925855719872117,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2989484692","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9955313,0.0008126932,0.00020764513,0.0013037829,0.0004433122,0.0009106401,0.00006028094,0.000006311641,0.0007239901],"genre_scores_gemma":[0.99892265,0.00014185451,0.00034326222,0.00006284908,0.000023143297,0.000008554824,0.0000015016543,0.000013897187,0.00048227905],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.97376555,0.0024317822,0.0014986485,0.0010111089,0.020212078,0.0010808457],"domain_scores_gemma":[0.9783849,0.008512352,0.0012215439,0.0017518249,0.009483315,0.00064602785],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.108231656,0.0002111957,0.0006495731,0.26909086,0.0003236766,0.007054362,0.011431104,0.00019942832,0.00043864132],"category_scores_gemma":[0.043196917,0.0001450322,0.00019400302,0.61527854,0.00093990914,0.0059272307,0.0033438117,0.00092968275,0.000024694027],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000714087,0.0004190861,0.9661016,0.000020500667,0.000007404975,0.0000059402514,0.0014834232,0.003067941,0.007809614,0.00016598376,0.0003029797,0.020544104],"study_design_scores_gemma":[0.0010863772,0.00021378273,0.93484783,0.00011935488,0.0000043483183,0.0000016582044,0.008116978,0.019445434,0.029963762,0.005993111,0.00004977768,0.00015761389],"about_ca_topic_score_codex":0.004475062,"about_ca_topic_score_gemma":0.0009897391,"teacher_disagreement_score":0.34618768,"about_ca_system_score_codex":0.00096834963,"about_ca_system_score_gemma":0.0019846612,"threshold_uncertainty_score":0.9939764},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2996045466","doi":"10.5433/1679-0359.2020v41n1p225","title":"Half a century of research on cattle foot and claw diseases: a scientometric analysis","year":2019,"lang":"en","type":"article","venue":"Semina Ciências Agrárias","topic":"Plant and fungal interactions","field":"Agricultural and Biological Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Razi Vaccine and Serum Research Institute; Conselho Nacional de Desenvolvimento Científico e Tecnológico","keywords":"Hoof; Claw; Web of science; Subject (documents); Scopus; Veterinary medicine; Library science; Foot (prosody); Bibliometrics; Medicine; Geography; Meta-analysis; Pathology; Political science; Biology; MEDLINE; Art; Computer science; Anatomy","score_opus":0.044725089654667725,"score_gpt":0.3195191335717688,"score_spread":0.2747940439171011,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2996045466","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98952395,0.000507434,3.514791e-7,0.00028743927,0.00017538015,0.00019329159,0.00021536136,0.0000240277,0.009072742],"genre_scores_gemma":[0.9978569,0.00013492227,0.00000986208,0.000048043694,0.000110833964,0.000008716279,0.00010173771,8.380942e-7,0.0017281091],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9985264,0.000104482,0.00018730869,0.0003590874,0.0005027936,0.00031991018],"domain_scores_gemma":[0.99841666,0.0011019065,0.000078240446,0.000105889936,0.000141917,0.00015537777],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00046001733,0.000107513006,0.00024715834,0.0004155998,0.00016441593,0.000070151495,0.00023194647,0.000077517165,0.00048460954],"category_scores_gemma":[0.00019780637,0.000043490603,0.00014233527,0.006238183,0.00009696903,0.00010496002,0.000105218154,0.00018320719,0.00012660967],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006366404,0.002606075,0.5657055,0.000114751085,0.00080123055,0.000023003995,0.0011659546,0.000056278597,0.34391055,0.00953849,0.026385045,0.04905653],"study_design_scores_gemma":[0.00023244762,0.0006841231,0.958652,0.00007512564,0.0001525484,0.0000048392662,0.0017514403,0.0001726926,0.0025496916,0.00020272867,0.035304576,0.00021774841],"about_ca_topic_score_codex":0.00066668453,"about_ca_topic_score_gemma":0.00071019557,"teacher_disagreement_score":0.39294657,"about_ca_system_score_codex":0.00004243685,"about_ca_system_score_gemma":0.000012178758,"threshold_uncertainty_score":0.5306135},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2998095279","doi":"10.1002/asi.24330","title":"The citation advantage of promoted articles in a cross‐publisher distribution platform: A 12‐month randomized controlled trial","year":2019,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"MaRS; Toronto Rehabilitation Institute; University of Toronto","funders":"Ontario Centres of Excellence","keywords":"Randomized controlled trial; Medicine; Confidence interval; Citation; Internal medicine; Computer science; Library science","score_opus":0.09548717587147706,"score_gpt":0.45004287221649125,"score_spread":0.35455569634501416,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2998095279","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9911189,0.000056896908,0.0008493295,0.004145204,0.0009129448,0.0020541043,0.00000962853,0.000004952545,0.000848041],"genre_scores_gemma":[0.99948174,0.000055200115,0.000083814884,0.000029372311,0.000012332804,0.00003659941,9.852926e-7,0.0000012903955,0.00029868336],"study_design_codex":"randomized_trial","study_design_gemma":"randomized_trial","domain_scores_codex":[0.9922283,0.00015747636,0.0018738199,0.00011199606,0.005340166,0.00028823712],"domain_scores_gemma":[0.97818434,0.006354505,0.0038949961,0.0002496931,0.011263565,0.00005292018],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.09832134,0.000067898196,0.00049723533,0.011350602,0.0003253522,0.0015753064,0.0012856465,0.00012783134,0.0000046875152],"category_scores_gemma":[0.28985515,0.000029876946,0.00017989897,0.04213507,0.00041260885,0.004635259,0.00018188053,0.00021521168,0.0000041365993],"study_design_candidate":"randomized_trial","study_design_consensus":"randomized_trial","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.45014042,0.00033894688,0.21037234,0.000023733413,0.00017131513,3.267245e-7,0.0022472846,0.00040816172,0.0022821594,0.1349822,0.0039769807,0.19505613],"study_design_scores_gemma":[0.83602047,0.0006552982,0.034031503,0.000023562608,0.000026932483,0.0000063119123,0.0028874131,0.03190398,0.001673925,0.0845325,0.008126157,0.000111932124],"about_ca_topic_score_codex":0.0000052007194,"about_ca_topic_score_gemma":0.000007914343,"teacher_disagreement_score":0.38588005,"about_ca_system_score_codex":0.0003172617,"about_ca_system_score_gemma":0.00051974005,"threshold_uncertainty_score":0.9998549},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"randomized_trial","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"randomized_trial","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3008004346","doi":"10.1101/2020.02.14.20023119","title":"Characteristics and Impact of Librarian Co-authored Systematic Reviews: A Bibliometric Analysis","year":2020,"lang":"en","type":"preprint","venue":"medRxiv","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"MEDLINE; Library science; Web of science; Bibliometrics; Cochrane Library; Collection development; Inclusion (mineral); Citation analysis; Grey literature; Citation; Medicine; Medical education; Computer science; Political science; Sociology; Social science","score_opus":0.7135928510555525,"score_gpt":0.5326164287326238,"score_spread":0.18097642232292865,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3008004346","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9211296,0.036425527,0.03278062,0.00037293872,0.00044326155,0.0062719434,0.00075154647,0.000024958073,0.0017996323],"genre_scores_gemma":[0.99222,0.0036724617,0.0026799422,0.00006929043,0.00012046295,0.0001948131,0.00012079304,0.00003447223,0.0008877704],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.93242264,0.03014882,0.02760113,0.002341028,0.0070736553,0.00041271007],"domain_scores_gemma":[0.94810396,0.009502121,0.031487253,0.0086960085,0.0015508021,0.00065984234],"candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","metaepi_broad","bibliometrics","insufficient_payload"],"category_scores_codex":[0.14948043,0.0009083776,0.024972754,0.08678115,0.0000890074,0.002420678,0.0039789537,0.00037831048,0.0055373763],"category_scores_gemma":[0.20031436,0.00040818556,0.011225328,0.20734778,0.00011891454,0.00022578021,0.0010458434,0.0005785175,0.0011709797],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010905373,0.00008237202,0.92828685,0.045944043,0.010546628,0.000023595225,0.00066152925,0.000037115256,0.00008642276,0.00013791399,0.011216283,0.0029663467],"study_design_scores_gemma":[0.00013428621,0.000091487724,0.94440114,0.0029488693,0.018512988,0.000009752656,0.00008453478,0.030117773,0.000009463926,0.0013555586,0.0017598955,0.00057421764],"about_ca_topic_score_codex":0.000036076988,"about_ca_topic_score_gemma":0.000003554918,"teacher_disagreement_score":0.12056663,"about_ca_system_score_codex":0.00005124881,"about_ca_system_score_gemma":0.00020545839,"threshold_uncertainty_score":0.999837},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3012083313","doi":"10.3389/fpsyt.2020.00182","title":"The Arab Region's Contribution to Global Mental Health Research (2009–2018): A Bibliometric Analysis","year":2020,"lang":"en","type":"review","venue":"Frontiers in Psychiatry","topic":"Mental Health Treatment and Access","field":"Psychology","cited_by":74,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Impact; McMaster University","funders":"","keywords":"Mental health; Middle East; Per capita; Latin Americans; Economic growth; Bibliometrics; Population; Publication; Political science; Geography; Medicine; Library science; Environmental health; Psychiatry","score_opus":0.10178412994823664,"score_gpt":0.4854889908538,"score_spread":0.3837048609055633,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3012083313","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000017023049,0.9554495,0.0010496303,0.0100303115,0.027326073,0.0042060604,0.00060127344,0.00006729892,0.0012528028],"genre_scores_gemma":[0.00005586935,0.9946618,0.0012697455,0.0007832319,0.0007151371,0.00088377873,0.0006593635,0.000049586124,0.00092147855],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9921075,0.0023831592,0.0015147725,0.0012982797,0.0009916001,0.0017046835],"domain_scores_gemma":[0.9971878,0.00020357587,0.00066950574,0.0011135354,0.00009752504,0.000728045],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0026403994,0.0005867099,0.002343807,0.037137486,0.0011003583,0.00021527836,0.0013756185,0.00049217034,0.000071969116],"category_scores_gemma":[0.00009991803,0.00042210592,0.0008121946,0.2687443,0.00017098935,0.000097486976,0.0002304117,0.0011199497,0.00049290643],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000104653875,0.00017249191,0.0050867037,0.0005217001,0.0007724696,0.000012627772,0.000038265967,7.489074e-8,8.967885e-11,0.0011523025,0.47759745,0.51454127],"study_design_scores_gemma":[0.0007802282,0.00040570355,0.003307063,0.0013702026,0.00048611898,0.000022485714,0.00035879508,0.000005274717,2.1691806e-9,0.00067500176,0.992275,0.0003141054],"about_ca_topic_score_codex":0.00058195315,"about_ca_topic_score_gemma":0.00046964054,"teacher_disagreement_score":0.5146776,"about_ca_system_score_codex":0.003214654,"about_ca_system_score_gemma":0.0007222631,"threshold_uncertainty_score":0.9998231},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3014203931","doi":"10.1016/j.jclepro.2020.121433","title":"A bibliometric analysis of process safety research in China: Understanding safety research progress as a basis for making China’s chemical industry more sustainable","year":2020,"lang":"en","type":"article","venue":"Journal of Cleaner Production","topic":"Risk and Safety Analysis","field":"Decision Sciences","cited_by":78,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University","funders":"National Key Research and Development Program of China; China Scholarship Council; Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China; Canada Research Chairs","keywords":"China; Process (computing); Sustainable development; Sustainability; Process safety; Business; Work in process; Engineering; Political science; Computer science; Marketing","score_opus":0.32039705632194077,"score_gpt":0.5196480824640306,"score_spread":0.19925102614208984,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3014203931","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93451875,0.00065221125,0.002657295,0.060785685,0.00010379201,0.00065830664,0.000011177034,0.000012476469,0.00060031156],"genre_scores_gemma":[0.9980719,0.00037446432,0.00064698234,0.000023657425,0.00052228704,0.000014648982,0.0000045700654,0.000025621788,0.00031584036],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.98836726,0.0013702936,0.0023496668,0.00085978856,0.0061820955,0.0008709243],"domain_scores_gemma":[0.99123436,0.0014338688,0.001256221,0.0005797195,0.0051895278,0.0003063221],"candidate_categories":["metaresearch","bibliometrics","research_integrity"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.041102465,0.00021560736,0.0011526382,0.10648109,0.00057326927,0.00036134876,0.0012892504,0.00036678446,0.00015025427],"category_scores_gemma":[0.029150803,0.00016270948,0.0005655728,0.35085562,0.0005455956,0.0013708926,0.0002860553,0.002417597,0.0000031059428],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.028483164,0.003108722,0.29662165,0.00095258036,0.003731144,0.00040048987,0.05103545,0.32031268,0.0012577985,0.004140698,0.009231633,0.280724],"study_design_scores_gemma":[0.0038502763,0.003542077,0.43157026,0.00091940205,0.0017472535,0.0001867953,0.3829671,0.067225836,0.008901374,0.09052077,0.007530911,0.001037944],"about_ca_topic_score_codex":0.00011238673,"about_ca_topic_score_gemma":0.000025280106,"teacher_disagreement_score":0.33193165,"about_ca_system_score_codex":0.000985107,"about_ca_system_score_gemma":0.0008525269,"threshold_uncertainty_score":0.9998839},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3016661053","doi":"10.1007/s11192-020-03462-0","title":"Telescopic and panoramic views of library and information science research 2011–2018: a comparison of four weighting schemes for author co-citation analysis","year":2020,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":10,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Citation; Field (mathematics); Context (archaeology); Weighting; Information retrieval; Data science; Zoom; Citation analysis; Set (abstract data type); Citation impact; Co-citation; Rank (graph theory); World Wide Web; Mathematics; History","score_opus":0.7748089364909767,"score_gpt":0.6271083223531914,"score_spread":0.14770061413778535,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3016661053","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9662152,0.0015286604,0.029025653,0.0011532757,0.00014467938,0.00066496735,0.00013270107,0.000018176517,0.0011167134],"genre_scores_gemma":[0.98202115,0.00028347675,0.017491654,0.000055091306,0.000025044677,0.000011165055,0.000013667436,0.0000064441547,0.00009232618],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9846341,0.00026173753,0.0014539373,0.00077130715,0.012197578,0.0006813453],"domain_scores_gemma":[0.9849724,0.007680626,0.00090838154,0.0005375562,0.0051824884,0.000718504],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.037481986,0.0001479669,0.000641403,0.2297348,0.00055749627,0.002475807,0.0019164417,0.00011470327,0.00006722835],"category_scores_gemma":[0.07969864,0.00011206927,0.00012736057,0.5852225,0.0014210263,0.0068551707,0.0010478362,0.0002689579,0.000018987314],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000590702,0.000093784074,0.7657769,0.00016751145,0.00004017806,3.185602e-7,0.002180106,0.000059356393,0.0076214527,0.0056959307,0.0024081531,0.21589728],"study_design_scores_gemma":[0.000993167,0.0011231806,0.57573444,0.000026610289,0.00006947147,0.0000010738822,0.004363029,0.34944925,0.034261897,0.0018709338,0.03181486,0.00029211232],"about_ca_topic_score_codex":0.000014878793,"about_ca_topic_score_gemma":0.0000012590331,"teacher_disagreement_score":0.3554877,"about_ca_system_score_codex":0.000047820555,"about_ca_system_score_gemma":0.0005050422,"threshold_uncertainty_score":0.9985597},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W3020379737","doi":"10.3390/jrfm13050085","title":"Current Research Trends on Interrelationships of Eco-Innovation and Internationalisation: A Bibliometric Analysis","year":2020,"lang":"en","type":"article","venue":"Journal of risk and financial management","topic":"Environmental Sustainability in Business","field":"Business, Management and Accounting","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Internationalization; Bibliometrics; Field (mathematics); Web of science; Knowledge management; Data science; Business; Computer science; Political science; World Wide Web; MEDLINE","score_opus":0.056208557419264,"score_gpt":0.31152234030295006,"score_spread":0.25531378288368606,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3020379737","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9855131,0.0003491177,0.012035275,0.00132659,0.00013975309,0.00010535324,0.0000042928673,0.0000057672482,0.00052074675],"genre_scores_gemma":[0.9987897,0.00045622227,0.00021418708,0.00013561443,0.00037164253,0.00000363475,0.000009751595,0.0000060181774,0.00001318623],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99862134,0.000034057586,0.0005274211,0.00016134405,0.00054329535,0.00011254099],"domain_scores_gemma":[0.9989228,0.00009751105,0.0005218648,0.00008569748,0.00035613443,0.000015948735],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0012809387,0.000094444586,0.00021095821,0.037039936,0.000118052885,0.00011581838,0.00014383886,0.000032419546,0.000054147476],"category_scores_gemma":[0.0006511489,0.00008401775,0.0000674176,0.05968121,0.00008968497,0.0005510156,0.00019756933,0.00024280415,0.0000037523712],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015238473,0.0001420744,0.50783855,0.00017927142,0.00006772811,0.000005807418,0.00019301099,0.00059774605,0.0000029902133,0.045021437,0.0006913073,0.44510767],"study_design_scores_gemma":[0.00049926626,0.000056290402,0.95856154,0.000041454816,0.0002298009,4.439616e-7,0.00025129275,0.0011618322,0.0000032885944,0.005279306,0.03384099,0.00007446403],"about_ca_topic_score_codex":0.000034841123,"about_ca_topic_score_gemma":0.000008744065,"teacher_disagreement_score":0.450723,"about_ca_system_score_codex":0.00005894963,"about_ca_system_score_gemma":0.0000071391737,"threshold_uncertainty_score":0.9738744},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3022074721","doi":"10.1186/s41256-020-00147-2","title":"Where is the policy? A bibliometric analysis of the state of policy research on medical tourism","year":2020,"lang":"en","type":"review","venue":"Global Health Research and Policy","topic":"Global Healthcare and Medical Tourism","field":"Health Professions","cited_by":61,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Medical tourism; Bibliometrics; Tourism; Public policy; Health policy; Scope (computer science); Public health; Corporate governance; Public relations; Discipline; Policy analysis; Publication; Political science; Sociology; Public administration; Medicine; Social science; Computer science; Economics; Library science; Management","score_opus":0.45511118710062276,"score_gpt":0.6875691708307479,"score_spread":0.2324579837301251,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3022074721","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00012258085,0.7717208,0.0000029787084,0.20824525,0.00021634577,0.0037535934,0.0038461923,0.000029447916,0.012062809],"genre_scores_gemma":[0.00060535903,0.98103595,0.000011715447,0.010739534,0.004141367,0.00033050656,0.000056981677,0.00005662752,0.003021948],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.96197975,0.020533448,0.0032642614,0.00110734,0.00868478,0.004430407],"domain_scores_gemma":[0.978489,0.010985879,0.0013631033,0.0022816788,0.0022105237,0.004669848],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","research_integrity"],"consensus_categories":["bibliometrics","research_integrity"],"category_scores_codex":[0.022989763,0.0006334556,0.0039991057,0.055984627,0.002509532,0.00004615742,0.0032381054,0.0013226352,0.00025049635],"category_scores_gemma":[0.024494532,0.00033192403,0.001018275,0.42034328,0.0024381918,0.00006573789,0.002854361,0.009242684,0.00022329042],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000066385044,0.00013866351,0.000063531756,0.024105003,0.00051325257,0.000016312164,0.0007938152,1.5432074e-7,5.856136e-9,0.017227478,0.24805275,0.70902264],"study_design_scores_gemma":[0.00044645087,0.0010788458,0.0024928798,0.01532329,0.00013565627,0.000010051506,0.00040578106,0.000017105718,6.770615e-8,0.0073098782,0.9726144,0.0001656082],"about_ca_topic_score_codex":0.3023602,"about_ca_topic_score_gemma":0.008026741,"teacher_disagreement_score":0.72456163,"about_ca_system_score_codex":0.004157732,"about_ca_system_score_gemma":0.1128296,"threshold_uncertainty_score":0.99997383},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3025061567","doi":"10.1155/2020/5357917","title":"Bibliometric Analysis of the Results of Cardio‐Oncology Research","year":2020,"lang":"en","type":"article","venue":"Evidence-based Complementary and Alternative Medicine","topic":"Chemotherapy-induced cardiotoxicity and mitigation","field":"Medicine","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Cardiotoxicity; Medicine; Oncology; Internal medicine; Web of science; Chemotherapy; Meta-analysis","score_opus":0.33516347265213287,"score_gpt":0.4550581634523884,"score_spread":0.11989469080025555,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3025061567","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95225656,0.0021865286,0.00063978817,0.043361768,0.00009632827,0.0005926383,0.00009909822,0.000010017863,0.00075729314],"genre_scores_gemma":[0.9960013,0.00089141325,0.00019800074,0.0024823302,0.0003272892,0.000012059083,0.000059207872,0.0000076321885,0.000020772191],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9971244,0.00061164546,0.00063112244,0.00033963163,0.0010932343,0.0001999646],"domain_scores_gemma":[0.9968189,0.0018349534,0.00029881962,0.0003315539,0.00055081764,0.00016493935],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0022420837,0.00013254647,0.000764708,0.01068643,0.00009100066,0.0000039693173,0.0002361756,0.000041524512,0.00020250761],"category_scores_gemma":[0.001286744,0.00008469273,0.0002064894,0.057677437,0.0007354597,0.000087427296,0.00009889779,0.00031341403,7.2312304e-7],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005420382,0.0002850536,0.1609565,0.0006806722,0.0057335086,0.000050673407,0.007747978,0.0006902038,0.7311466,0.00050753856,0.0068226205,0.07995829],"study_design_scores_gemma":[0.0091343615,0.005964414,0.34293157,0.0014297952,0.002603352,0.0000069536813,0.0027685429,0.0052208835,0.6213272,0.00015753419,0.008305547,0.000149824],"about_ca_topic_score_codex":0.0015384468,"about_ca_topic_score_gemma":0.000042281656,"teacher_disagreement_score":0.18197507,"about_ca_system_score_codex":0.00009819028,"about_ca_system_score_gemma":0.00016771244,"threshold_uncertainty_score":0.9623521},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3025142841","doi":"10.1007/s42398-020-00105-1","title":"Two decades of eco-efficiency research: a bibliometric analysis","year":2020,"lang":"en","type":"article","venue":"Environmental Sustainability","topic":"Environmental Impact and Sustainability","field":"Environmental Science","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Global Affairs Canada","keywords":"Data envelopment analysis; Sustainability; Resource efficiency; Eco-efficiency; Resource (disambiguation); Environmental economics; Bibliometrics; Bibliographic coupling; Computer science; Management science; Economics; Mathematics; Library science","score_opus":0.03503633689611895,"score_gpt":0.3363465299615412,"score_spread":0.3013101930654222,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3025142841","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9917705,0.00020264242,0.0010683729,0.0009213572,0.000023524177,0.0010496046,0.00005350448,0.00006218301,0.0048482916],"genre_scores_gemma":[0.998679,0.00004836919,0.0006504062,0.00019343085,0.00003535751,0.000055118584,0.000027371929,0.00003194261,0.0002790008],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.994359,0.00075757894,0.00081530743,0.0011931296,0.001744445,0.0011305239],"domain_scores_gemma":[0.9976624,0.00035568728,0.00021301983,0.000985841,0.000020592428,0.0007624126],"candidate_categories":["metaepi_narrow","bibliometrics","sts","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00243874,0.0003830494,0.0006235625,0.005821938,0.00037719458,0.00005249606,0.00090805546,0.0001521546,0.010764894],"category_scores_gemma":[0.0011426839,0.00036296668,0.00048722074,0.07996321,0.0027811865,0.00058341457,0.001573031,0.0005692094,0.0002974569],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011929666,0.0010017186,0.9792144,0.000056212903,0.00009633425,0.000021198488,0.0016722573,0.006783556,0.0025294926,0.0000850297,0.00019819156,0.0082223285],"study_design_scores_gemma":[0.00062454375,0.0005133704,0.9804516,0.000001545249,0.00015555014,0.000001935852,0.006127177,0.0023915977,0.004721856,0.0019359505,0.0026993856,0.0003755325],"about_ca_topic_score_codex":0.0010777689,"about_ca_topic_score_gemma":0.00003606596,"teacher_disagreement_score":0.074141264,"about_ca_system_score_codex":0.0025700978,"about_ca_system_score_gemma":0.00004787122,"threshold_uncertainty_score":0.99993265},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3026389550","doi":"10.1017/9781108946636.010","title":"Ensuring Correctness or Promoting Consistency? Tracking Policy Priorities in Investment Arbitration through Large-Scale Citation Analysis","year":2022,"lang":"en","type":"article","venue":"Cambridge University Press eBooks","topic":"International Arbitration and Investment Law","field":"Business, Management and Accounting","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Discretion; Correctness; Consistency (knowledge bases); Mandate; Arbitration; Law and economics; Political science; Investment (military); Law; Treaty; Economics; Business; Computer science","score_opus":0.03030955508320623,"score_gpt":0.23132789298967762,"score_spread":0.2010183379064714,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3026389550","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.37522027,0.000015392186,0.0013165659,0.0001710921,0.00025210355,0.0004889763,0.00003367166,0.00014076586,0.6223612],"genre_scores_gemma":[0.9812232,0.0000030440506,0.00021114577,0.005458266,0.00020941134,0.000016097138,0.00017831287,0.000016406628,0.0126840705],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986821,0.00006973285,0.00027639535,0.0003262046,0.00037882695,0.00026673955],"domain_scores_gemma":[0.9993562,0.00003869154,0.00026932973,0.00015441122,0.00016530455,0.000016026073],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003210842,0.00016161967,0.00021137175,0.0006840921,0.0006425949,0.00019552316,0.00022583893,0.000039395054,0.000025627372],"category_scores_gemma":[0.000048998452,0.00018507497,0.000106467865,0.00044368298,0.000061460465,0.0010315884,0.00022543562,0.00020337944,0.000003432008],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012870974,0.0001174302,0.0026376774,0.00008063402,0.000098466335,0.00005405205,0.0012512145,0.0013929228,0.00033321758,0.9933549,0.00046597238,0.00008475774],"study_design_scores_gemma":[0.00521427,0.00010390696,0.02541472,0.00013613598,0.00077487313,0.000018086397,0.024136502,0.19877838,0.0022713477,0.00031299898,0.74151057,0.0013281994],"about_ca_topic_score_codex":0.0025782345,"about_ca_topic_score_gemma":0.0003442483,"teacher_disagreement_score":0.99304193,"about_ca_system_score_codex":0.00035867913,"about_ca_system_score_gemma":0.00011652094,"threshold_uncertainty_score":0.7547138},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3032390629","doi":"10.1111/jonm.13038","title":"Trends in high‐impact papers in nursing research published from 2008 to 2018: A web of science‐based bibliometric analysis","year":2020,"lang":"en","type":"article","venue":"Journal of Nursing Management","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Subject (documents); Nursing research; Bibliometrics; Citation; Citation impact; Impact factor; Web of science; Citation analysis; Library science; Medicine; Medical education; MEDLINE; Nursing; Political science; Computer science","score_opus":0.5132470776187895,"score_gpt":0.6032668532435289,"score_spread":0.0900197756247394,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3032390629","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9668023,0.00065089524,0.0017143592,0.012337232,0.00054238696,0.00023634112,0.000025275662,0.000006835512,0.017684361],"genre_scores_gemma":[0.99416244,0.00010396824,0.005412936,0.000117590935,0.00008159356,0.0000025169225,0.0000027305427,0.000012395989,0.00010385383],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9739723,0.0006936864,0.0020592306,0.00092062616,0.021174593,0.0011795706],"domain_scores_gemma":[0.9917018,0.0022836358,0.00078162446,0.00080963934,0.0032219107,0.0012014145],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.048432235,0.0002107357,0.00095079484,0.90734196,0.00018937106,0.002900207,0.0044828705,0.00009674045,0.0009770882],"category_scores_gemma":[0.017747695,0.00015325677,0.00044785521,0.9666238,0.0007102143,0.0013400102,0.00036703344,0.0007301089,0.000020043262],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044426246,0.0012856075,0.16408585,0.000005623093,0.00010713327,0.00015610825,0.0007295209,0.020396797,0.0022721088,0.00014833143,0.035304114,0.7750645],"study_design_scores_gemma":[0.0014931663,0.00080853753,0.9636149,0.00022459675,0.00006622921,0.0000012803753,0.0015708624,0.029479155,0.0002192311,0.0015293375,0.00082839426,0.00016430242],"about_ca_topic_score_codex":0.00021402125,"about_ca_topic_score_gemma":0.0000196144,"teacher_disagreement_score":0.7995291,"about_ca_system_score_codex":0.0021070484,"about_ca_system_score_gemma":0.00060798525,"threshold_uncertainty_score":0.99993616},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3036351727","doi":"","title":"Estudio bibliométrico de los trabajos especiales de grado sobre cirugía presentados de la Facultad de Odontología de La Universidad de Los Andes, 2005-2017","year":2019,"lang":"es","type":"article","venue":"","topic":"Health and Medical Education","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Specialty; Oral and maxillofacial surgery; Presentation (obstetrics); Oral surgery; Descriptive statistics; Medicine; Humanities; Psychology; Dentistry; Surgery; Family medicine; Art; Mathematics; Statistics","score_opus":0.04400685537421869,"score_gpt":0.46513745587933275,"score_spread":0.4211306005051141,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3036351727","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9095338,0.0026395172,0.009226425,0.014888875,0.00085728284,0.0026711342,0.00024735415,0.00032963694,0.059606005],"genre_scores_gemma":[0.95087624,0.015528964,0.0058887787,0.0077776713,0.002085599,0.00019124817,0.0000931008,0.00012304478,0.01743533],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.988212,0.005692809,0.0009818841,0.0008098215,0.00065736467,0.0036461286],"domain_scores_gemma":[0.9899078,0.0051594805,0.00050518615,0.00071220216,0.00025055712,0.0034647647],"candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["research_integrity"],"category_scores_codex":[0.0059859385,0.00061516714,0.0008900918,0.0019178371,0.0011651153,0.0001603448,0.00089010934,0.0021750776,0.0022476132],"category_scores_gemma":[0.0036632617,0.0005836718,0.00027960652,0.002187542,0.0005782214,0.00040547046,0.00021368016,0.0027661158,0.0004492366],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034097987,0.001308577,0.60006666,0.002511002,0.00014500019,0.000200131,0.009775782,0.00003599024,0.001552043,0.0036151982,0.36438793,0.016060736],"study_design_scores_gemma":[0.004780627,0.00026950723,0.6227603,0.0013279958,0.00034985197,0.000249397,0.01808743,0.0016572333,0.000702855,0.0011543967,0.347941,0.00071941066],"about_ca_topic_score_codex":0.016714398,"about_ca_topic_score_gemma":0.002249228,"teacher_disagreement_score":0.04217067,"about_ca_system_score_codex":0.0035949296,"about_ca_system_score_gemma":0.026791548,"threshold_uncertainty_score":0.99966145},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W3037147029","doi":"10.1016/j.athoracsur.2020.05.055","title":"Twitter Activity Is Associated With a Higher H-Index in Academic Thoracic Surgery","year":2020,"lang":"en","type":"letter","venue":"The Annals of Thoracic Surgery","topic":"Social Media in Health Education","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Joseph’s Healthcare Hamilton; McMaster University","funders":"","keywords":"Scopus; Medicine; Social media; Citation; Index (typography); Citation index; Library science; MEDLINE; World Wide Web; Computer science; Law","score_opus":0.3936658835327579,"score_gpt":0.47883222303046813,"score_spread":0.08516633949771024,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3037147029","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.33311722,0.0006817332,0.0000023120188,0.6605734,0.0041732728,0.0008969958,0.00004340588,0.0001185339,0.00039311274],"genre_scores_gemma":[0.545474,0.0018930454,0.000005018552,0.4418372,0.009007068,0.00068869436,0.00010354894,0.00014328837,0.0008481507],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98828197,0.005982636,0.0011291489,0.00078882155,0.0022886938,0.0015287349],"domain_scores_gemma":[0.94545287,0.05101904,0.0020412358,0.00070160243,0.00052251935,0.0002627132],"candidate_categories":["metaresearch","metaepi_narrow","research_integrity"],"consensus_categories":["research_integrity"],"category_scores_codex":[0.0089136185,0.00058100035,0.0018050667,0.0007286794,0.00046141786,0.00009328294,0.0009023759,0.0019026761,0.00054990436],"category_scores_gemma":[0.013706022,0.0004741788,0.0005792226,0.0031035375,0.0010003452,0.00046710978,0.000114880764,0.0046390803,0.00009253094],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024572285,0.00006559704,0.06774927,0.00031777433,0.00014731952,0.00010999069,0.008982002,6.539215e-7,0.000003449047,0.0000019702009,0.90218896,0.020187307],"study_design_scores_gemma":[0.00005079114,0.000035631205,0.09388533,0.0034639314,0.00020618836,0.0000031749987,0.0011131798,0.0000116604115,0.0007466664,0.0021416512,0.89696187,0.001379933],"about_ca_topic_score_codex":0.006977092,"about_ca_topic_score_gemma":0.0004134741,"teacher_disagreement_score":0.21873623,"about_ca_system_score_codex":0.00035263758,"about_ca_system_score_gemma":0.003232186,"threshold_uncertainty_score":0.999771},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3037387538","doi":"10.1016/j.bbadis.2020.165880","title":"Bibliometric analysis of personalized humanized mouse and Drosophila models for effective combinational therapy in cancer patients","year":2020,"lang":"en","type":"review","venue":"Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease","topic":"Hippo pathway signaling and YAP/TAZ","field":"Biochemistry, Genetics and Molecular Biology","cited_by":8,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Drosophila melanogaster; Cancer; Drosophila (subgenus); Computational biology; Biology; Personalized medicine; Productivity; Cancer treatment; Bioinformatics; Gene; Genetics","score_opus":0.027504766822015653,"score_gpt":0.3206505186213922,"score_spread":0.2931457517993766,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3037387538","genre_codex":"empirical","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5850982,0.4096,0.00006881747,0.00004075919,0.00002581758,0.0013496188,0.0038017232,0.000009039732,0.0000060372704],"genre_scores_gemma":[0.47810754,0.5179567,0.00007599479,0.00007621653,0.000028200882,0.0004678083,0.0032091362,0.00007108529,0.000007340831],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9969982,0.00043393706,0.0007857225,0.0009911057,0.000479756,0.0003113124],"domain_scores_gemma":[0.9978716,0.00014532293,0.00081689167,0.0004945426,0.00040736052,0.0002643222],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.00021960413,0.00062229997,0.00192921,0.009235784,0.000057066973,0.000037769907,0.0004589559,0.00030964994,0.0000062729046],"category_scores_gemma":[0.00021256074,0.00057659805,0.0015839054,0.014897693,0.00022589728,0.000014279497,0.00016757504,0.00016811634,4.0443487e-7],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019458462,0.004591591,0.000058458252,0.009996818,0.015732734,0.000007839358,0.00024023178,0.00014413653,0.84497017,0.00048107083,0.0000711741,0.121759936],"study_design_scores_gemma":[0.029901143,0.008465694,0.008925626,0.0076815956,0.05098992,0.0000013861179,0.00006017973,0.005099321,0.85909915,0.0011539973,0.021434134,0.0071878573],"about_ca_topic_score_codex":0.000031767002,"about_ca_topic_score_gemma":0.000002082066,"teacher_disagreement_score":0.11457208,"about_ca_system_score_codex":0.0000493434,"about_ca_system_score_gemma":0.00043025194,"threshold_uncertainty_score":0.99966854},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3048736466","doi":"10.1007/s40815-020-00924-8","title":"Citation Analysis of Fuzzy Set Theory Journals: Bibliometric Insights About Authors and Research Areas","year":2020,"lang":"en","type":"article","venue":"International Journal of Fuzzy Systems","topic":"Multi-Criteria Decision Making","field":"Decision Sciences","cited_by":26,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Consejo Nacional de Ciencia y Tecnología","keywords":"Computer science; Citation; Field (mathematics); Data science; Set (abstract data type); Fuzzy set; Citation analysis; Management science; Computational intelligence; Bibliometrics; Fuzzy logic; Operations research; Data mining; Mathematics; Artificial intelligence; World Wide Web","score_opus":0.4355136552884411,"score_gpt":0.5309238991282088,"score_spread":0.0954102438397677,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3048736466","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9632767,0.009142435,0.02051967,0.0017113322,0.002762357,0.0002057238,0.00010084913,0.000012251097,0.002268725],"genre_scores_gemma":[0.9977948,0.00042381007,0.0006956737,0.00012500906,0.0007667317,0.0000031970515,0.0000045457014,0.000017995875,0.00016823277],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9826294,0.0028933447,0.0033608703,0.0004723681,0.010356924,0.00028705323],"domain_scores_gemma":[0.97396106,0.009579346,0.0029265122,0.0003728681,0.01265283,0.0005073757],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.025614832,0.00020782727,0.00097285776,0.08665667,0.00015147716,0.0018482828,0.0023042082,0.00016419032,0.00026588712],"category_scores_gemma":[0.036635038,0.00014641126,0.0004643291,0.060791302,0.00019523843,0.0013072697,0.00034916337,0.0006361431,0.00004994022],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005115158,0.000754159,0.33379608,0.0001554545,0.016375596,0.0019930804,0.054323655,0.05574229,0.039402984,0.05638716,0.062154144,0.37380025],"study_design_scores_gemma":[0.003863587,0.00088054175,0.74882656,0.0013074976,0.00082644715,0.0007474228,0.029101023,0.08891533,0.0007342357,0.08965958,0.034402166,0.0007356017],"about_ca_topic_score_codex":0.000048415364,"about_ca_topic_score_gemma":0.000008948619,"teacher_disagreement_score":0.41503048,"about_ca_system_score_codex":0.00015150361,"about_ca_system_score_gemma":0.000227533,"threshold_uncertainty_score":0.9991879},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3049588163","doi":"10.1177/0961000620948573","title":"On the evolution of library and information science doctoral dissertation topics in China","year":2020,"lang":"en","type":"article","venue":"Journal of Librarianship and Information Science","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Fundamental Research Funds for the Provincial Universities of Zhejiang; Natural Science Foundation of Zhejiang Province; Social Sciences and Humanities Research Council of Canada","keywords":"China; Information science; Library science; Documentation science; Sociology; Political science; Computer science","score_opus":0.21682604463760077,"score_gpt":0.44240869001076066,"score_spread":0.2255826453731599,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3049588163","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97363466,0.00006137271,0.0027065647,0.013578157,0.00020150107,0.00012934869,0.00000557864,0.0000043382825,0.009678481],"genre_scores_gemma":[0.99855894,0.00013225774,0.0005347018,0.00073588645,0.000027290735,7.3381597e-7,5.324057e-7,0.0000010798584,0.000008607074],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9941429,0.0000740171,0.0010122275,0.00012048386,0.004434973,0.00021539396],"domain_scores_gemma":[0.9971583,0.0007971892,0.00081016007,0.00017086828,0.0007644904,0.00029901112],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics","scholarly_communication"],"category_scores_codex":[0.009736734,0.000075671865,0.0001589739,0.018810244,0.00028739724,0.0035283186,0.001438503,0.000040864215,0.000042789863],"category_scores_gemma":[0.020988913,0.000042199215,0.000033225213,0.078432925,0.0007135719,0.11895049,0.0002830769,0.00024525568,0.000006886946],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014863629,0.00002527948,0.17567104,0.000021950229,0.000001992457,6.7458797e-7,0.007509119,0.00034292164,0.00084593135,0.74995285,0.00044452873,0.06503508],"study_design_scores_gemma":[0.0003262855,0.0003096793,0.95591444,0.00002341796,0.0000013684424,0.0000071791737,0.0033187352,0.027153395,0.0017971674,0.008826379,0.0022569408,0.00006501446],"about_ca_topic_score_codex":0.000005970669,"about_ca_topic_score_gemma":1.5178644e-7,"teacher_disagreement_score":0.7802434,"about_ca_system_score_codex":0.00004028895,"about_ca_system_score_gemma":0.0009314887,"threshold_uncertainty_score":0.99750614},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W3076693518","doi":"10.1097/md.0000000000021828","title":"A bibliometric analysis of income and cardiovascular disease","year":2020,"lang":"en","type":"review","venue":"Medicine","topic":"Health disparities and outcomes","field":"Social Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Science Citation Index; Index (typography); Bibliometrics; Disease; Environmental health; Public health; Obesity; Gerontology; Population; Citation; Pathology; Library science","score_opus":0.09230446196376621,"score_gpt":0.4231286600761334,"score_spread":0.3308241981123672,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3076693518","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000013755411,0.9971443,0.000024854813,0.0011794352,0.00018202954,0.00032385625,0.000043506057,0.000019812289,0.0010684519],"genre_scores_gemma":[0.0002524954,0.9989787,0.000010869273,0.00020952524,0.00037720686,0.000019142031,0.000028233726,0.000010519978,0.000113343594],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978079,0.00037329888,0.0005047274,0.00027387144,0.0007964052,0.00024377066],"domain_scores_gemma":[0.9982596,0.0006034234,0.0001641773,0.00030503597,0.00006469996,0.0006030687],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0014754675,0.00015777089,0.0028812767,0.04518435,0.00011498485,0.000012204914,0.00022582579,0.00010730998,0.00030070383],"category_scores_gemma":[0.0023028478,0.000107262495,0.0007454735,0.21306811,0.0002913831,0.000035112305,0.00006205217,0.00014474313,0.000004582606],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[8.781729e-7,0.0000045660777,0.0031923307,0.012792083,0.0025128236,0.000024685203,0.00032245254,2.1935615e-7,7.2473354e-11,0.00058232347,0.0003467939,0.98022085],"study_design_scores_gemma":[0.000073493895,0.000010069657,0.014460812,0.0025875368,0.021061141,1.6772218e-7,0.00014698961,0.000003550623,2.1993436e-10,0.000013848374,0.96155405,0.00008834854],"about_ca_topic_score_codex":0.008007607,"about_ca_topic_score_gemma":0.00011654228,"teacher_disagreement_score":0.9801325,"about_ca_system_score_codex":0.000056595196,"about_ca_system_score_gemma":0.0004301969,"threshold_uncertainty_score":0.99859816},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3084104727","doi":"10.2478/jdis-2020-0038","title":"Scientometric Analysis of Research Output from Brazil in Response to the Zika Crisis Using e-Lattes","year":2020,"lang":"en","type":"article","venue":"Journal of Data and Information Science","topic":"Big Data and Business Intelligence","field":"Business, Management and Accounting","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"","keywords":"Originality; Zika virus; Dimension (graph theory); Value (mathematics); Sociology; Library science; Public relations; Political science; Medicine; Social science; Computer science; Qualitative research","score_opus":0.3078280822000569,"score_gpt":0.444984785050678,"score_spread":0.1371567028506211,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3084104727","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9788035,0.00008962727,0.012009953,0.008676639,0.00011941546,0.000075356525,0.000078493584,0.0000030170636,0.00014398216],"genre_scores_gemma":[0.99689835,0.000043668533,0.0007626056,0.0021726298,0.00010468611,2.4276298e-7,0.000015249455,0.0000013990602,0.0000011851866],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.997762,0.00002601808,0.00058464974,0.00013720887,0.0013166533,0.00017344301],"domain_scores_gemma":[0.9978387,0.00021889688,0.00042679452,0.00035762062,0.0011218591,0.00003617992],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.00862939,0.000059477792,0.00018220744,0.0057948725,0.00021696728,0.0009001504,0.0017773269,0.000020996722,0.00005264674],"category_scores_gemma":[0.0054925107,0.000038172628,0.000025159658,0.033208776,0.00022769049,0.024833314,0.0010873596,0.00015475212,0.000022256501],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031290695,0.00029708917,0.54094815,0.0004589046,0.0003814061,0.000017459583,0.021086883,0.039634477,0.025710352,0.0029325313,0.070717156,0.29468653],"study_design_scores_gemma":[0.00025753528,0.00003345347,0.44850358,0.000088467124,0.00013177881,0.0000022555275,0.0057852184,0.4100939,0.00024177165,0.000066440385,0.13466805,0.00012755003],"about_ca_topic_score_codex":0.00042544986,"about_ca_topic_score_gemma":0.00002287253,"teacher_disagreement_score":0.3704594,"about_ca_system_score_codex":0.000024751394,"about_ca_system_score_gemma":0.00013397906,"threshold_uncertainty_score":0.98880583},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3084269305","doi":"10.3233/jifs-189199","title":"Leading countries in computer science: A bibliometric overview","year":2020,"lang":"en","type":"article","venue":"Journal of Intelligent & Fuzzy Systems","topic":"Economic Growth and Development","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"China; Web of science; Field (mathematics); Ranking (information retrieval); Computer science; Library science; Data science; Regional science; Political science; Geography; Information retrieval; Mathematics; MEDLINE; Law","score_opus":0.05322433609975351,"score_gpt":0.2831886703868556,"score_spread":0.22996433428710206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3084269305","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24333496,0.01714026,0.7300215,0.0025712035,0.00511114,0.0003455724,0.0000015640309,0.000049401595,0.0014244252],"genre_scores_gemma":[0.99054915,0.0009350129,0.0071540787,0.0009713779,0.00036321557,0.0000023621512,1.3244728e-7,0.000007240238,0.000017415909],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9974277,0.00006710949,0.0011510147,0.00028241094,0.00069415744,0.0003775955],"domain_scores_gemma":[0.99832493,0.00014431795,0.00057076203,0.00020576578,0.00039768548,0.00035653333],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.002286577,0.00015468756,0.00047526517,0.010729357,0.00008177612,0.0007183488,0.0014933223,0.000046786772,0.000008818884],"category_scores_gemma":[0.00013100058,0.00012573141,0.00010408865,0.021610402,0.000070748436,0.001178142,0.00026911998,0.00023128802,0.00018173976],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021299883,0.00066330004,0.36833218,0.0026627902,0.00044377721,0.001823917,0.034382,0.052308757,0.0015668507,0.20035322,0.040225055,0.29702517],"study_design_scores_gemma":[0.0048257904,0.0046884534,0.086547874,0.0056168386,0.000066791596,0.0033938328,0.0020842666,0.6241859,0.019121774,0.005208556,0.24105132,0.0032085893],"about_ca_topic_score_codex":0.00001839393,"about_ca_topic_score_gemma":0.0000013707692,"teacher_disagreement_score":0.7472142,"about_ca_system_score_codex":0.0003204821,"about_ca_system_score_gemma":0.0006037546,"threshold_uncertainty_score":0.99918574},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3085347427","doi":"10.5114/pedm.2020.98165","title":"Pediatric type 1 diabetes research in the 21st century: a scientometric review","year":2020,"lang":"en","type":"review","venue":"Pediatric Endocrinology Diabetes and Metabolism","topic":"Diabetes and associated disorders","field":"Biochemistry, Genetics and Molecular Biology","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Pediatric endocrinology; Diabetes mellitus; Type 2 diabetes; Medicine; Internal medicine; Endocrinology","score_opus":0.04295512984565002,"score_gpt":0.34477967322457603,"score_spread":0.301824543378926,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3085347427","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0044055833,0.9917464,6.259673e-8,0.00039469707,0.0007979433,0.0015184336,0.00012577834,0.0000308077,0.000980258],"genre_scores_gemma":[0.00066483923,0.9950563,0.00003042628,0.0009879691,0.0021709795,0.00047375285,0.00046562645,0.00009707882,0.000053058775],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9933199,0.0016415346,0.000995859,0.0014000755,0.0006596408,0.0019829446],"domain_scores_gemma":[0.997604,0.00057628256,0.0004766959,0.00082635804,0.00024265784,0.0002740089],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0019411923,0.0007721087,0.0021917922,0.0015957139,0.0002784961,0.00013735268,0.0014145654,0.0005509455,0.000026972515],"category_scores_gemma":[0.0016113226,0.0005498032,0.0005389511,0.008393428,0.00025227354,0.000016548342,0.0008610031,0.0014198517,0.00010928644],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000031032719,0.00028009972,0.0072384034,0.025016155,0.00027371824,0.000022361399,0.000042882017,1.8997429e-7,0.0000051440006,0.0005665611,0.017597357,0.94895405],"study_design_scores_gemma":[0.0003564146,0.0001607636,0.00057318906,0.0005027883,0.0017726553,0.0000050980248,0.00006039662,5.224369e-7,0.000001650334,0.00005408511,0.995945,0.0005674739],"about_ca_topic_score_codex":0.000008343141,"about_ca_topic_score_gemma":0.0000029708071,"teacher_disagreement_score":0.9783476,"about_ca_system_score_codex":0.000035208388,"about_ca_system_score_gemma":0.0004987969,"threshold_uncertainty_score":0.99969536},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3087642349","doi":"10.3390/jrfm13090220","title":"Mapping the Scientific Research on Mass Customization Domain: A Critical Review and Bibliometric Analysis","year":2020,"lang":"en","type":"review","venue":"Journal of risk and financial management","topic":"Product Development and Customization","field":"Business, Management and Accounting","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Mass customization; Data science; Personalization; Domain (mathematical analysis); Computer science; Field (mathematics); Analytics; Identification (biology); Visualization; Domain analysis; Bibliometrics; Software; World Wide Web; Software development; Data mining","score_opus":0.08367487445526799,"score_gpt":0.33141014385628553,"score_spread":0.24773526940101753,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3087642349","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00001879353,0.9914822,0.0054466245,0.0010270671,0.00045890518,0.00073403784,0.000005010863,0.000013277315,0.00081407896],"genre_scores_gemma":[0.00028840193,0.99784046,0.0004162837,0.00036890645,0.0009212658,0.000027392232,0.000021416354,0.000022946686,0.000092931405],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9971372,0.00018512185,0.00095657224,0.00046186877,0.00094098336,0.00031824884],"domain_scores_gemma":[0.9979972,0.00025287288,0.0008806398,0.00026507044,0.000564698,0.000039472416],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0062013483,0.00029273704,0.0012524421,0.046638593,0.0008160313,0.001092172,0.00045500963,0.000107255786,0.000031333286],"category_scores_gemma":[0.0014221702,0.0001865908,0.00031198148,0.1261257,0.00018798049,0.0005168249,0.000355224,0.0006759331,0.000044280318],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012521754,0.000039120736,0.00008372828,0.017234277,0.0001987686,0.000040630122,0.000035086814,0.000002386026,1.5360998e-8,0.0052414252,0.014115484,0.96299654],"study_design_scores_gemma":[0.00019919628,0.000014594311,0.0008586782,0.0057487357,0.004529223,0.000004646845,0.000075891396,0.00002542563,1.3116435e-8,0.001460777,0.9868809,0.00020191332],"about_ca_topic_score_codex":0.000004369756,"about_ca_topic_score_gemma":0.0000023628156,"teacher_disagreement_score":0.97276545,"about_ca_system_score_codex":0.00007834771,"about_ca_system_score_gemma":0.00007620609,"threshold_uncertainty_score":0.9999448},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3089473186","doi":"10.1097/md.0000000000022623","title":"Bibliometric analysis of the 100 most cited articles on cervical cancer radiotherapy","year":2020,"lang":"en","type":"article","venue":"Medicine","topic":"Endometrial and Cervical Cancer Treatments","field":"Medicine","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Cervical cancer; Radiation therapy; Scopus; Radiation oncology; MEDLINE; Cancer; Bibliometrics; Web of science; Brachytherapy; Clinical Oncology; Family medicine; Internal medicine; Oncology; Meta-analysis; Library science","score_opus":0.06120082266863873,"score_gpt":0.34938254915741435,"score_spread":0.2881817264887756,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3089473186","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9256967,0.0046936674,0.00011721688,0.06735163,0.00016609255,0.0003021717,0.000028770943,0.000038826416,0.0016049257],"genre_scores_gemma":[0.98171425,0.0008095356,0.000047291356,0.016648248,0.0005377929,0.000018803685,0.0000156667,0.000014813186,0.0001935726],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.998398,0.00005951518,0.00035214672,0.00025386113,0.0007427851,0.00019371214],"domain_scores_gemma":[0.9989504,0.00024980673,0.00011689124,0.00029758806,0.00012414475,0.00026112606],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00013252684,0.00014642144,0.00066545146,0.015921198,0.000045409735,0.000004242682,0.00015441672,0.000059912472,0.0047092047],"category_scores_gemma":[0.00048731922,0.00007197242,0.0001890462,0.28804222,0.0001367951,0.000020659321,0.00002632949,0.00015687429,0.0000063034277],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019940606,0.00045349041,0.81463623,0.000109816014,0.0043756096,0.000028198112,0.0008369505,0.00028979738,0.013101936,0.00014301167,0.0066822646,0.15734862],"study_design_scores_gemma":[0.0064039524,0.0015453593,0.9705854,0.000103088234,0.003105072,0.0000014374716,0.00006616135,0.0031633358,0.0053467457,0.000017241624,0.009574808,0.00008738167],"about_ca_topic_score_codex":0.0008491514,"about_ca_topic_score_gemma":0.000029557234,"teacher_disagreement_score":0.272121,"about_ca_system_score_codex":0.00009024437,"about_ca_system_score_gemma":0.000044870558,"threshold_uncertainty_score":0.9962006},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W3089570069","doi":"10.33448/rsd-v9i10.8624","title":"Lifestyle risk behaviors of university students: a bibliometric analysis","year":2020,"lang":"en","type":"article","venue":"Research Society and Development","topic":"Health and Lifestyle Studies","field":"Health Professions","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"Coordenação de Aperfeiçoamento de Pessoal de Nível Superior","keywords":"SciELO; Web of science; Obesity; Population; Institution; Gerontology; Psychology; Environmental health; MEDLINE; Medicine; Social science; Political science; Sociology","score_opus":0.1918170919197717,"score_gpt":0.5044827073691471,"score_spread":0.31266561544937543,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3089570069","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9964957,0.0006401838,0.00045104243,0.0012875815,0.0000314868,0.00047039773,0.000034982677,0.00003544456,0.0005531775],"genre_scores_gemma":[0.98756874,0.008328616,0.0029903664,0.00036587694,0.000069388545,0.000018676026,0.000013664023,0.000008268707,0.0006364285],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9966419,0.001075021,0.00038075208,0.00030355866,0.0009891925,0.000609588],"domain_scores_gemma":[0.9973297,0.0013641936,0.0001310203,0.00013112488,0.0005096056,0.0005343372],"candidate_categories":["bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.0048816935,0.00009924179,0.0003404116,0.004824257,0.0018001755,0.000008896778,0.0002344604,0.00013298953,0.00015494895],"category_scores_gemma":[0.00081137236,0.0000884073,0.0000983787,0.06596803,0.00016878314,0.00007310495,0.00064712943,0.0008095152,0.00006664663],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004936902,0.00007796938,0.95913464,0.00028951047,0.00045066222,0.0000033345545,0.02799484,0.0000018275838,0.0000075837906,0.000026701164,0.0086768875,0.0032866525],"study_design_scores_gemma":[0.0006228346,0.0000761285,0.8590529,0.000026638116,0.000063740656,3.5585145e-8,0.037003696,0.0000292326,0.000009889601,0.0000029094053,0.10303328,0.00007868845],"about_ca_topic_score_codex":0.00064639497,"about_ca_topic_score_gemma":0.00009181671,"teacher_disagreement_score":0.100081734,"about_ca_system_score_codex":0.00017079872,"about_ca_system_score_gemma":0.00096803607,"threshold_uncertainty_score":0.9994993},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3091797487","doi":"10.21272/mmi.2020.3-21","title":"The Law Aspects in Health Management: A Bibliometric Analysis of Issues on the Injury, Damage and Harm in Criminal Law","year":2020,"lang":"en","type":"article","venue":"Marketing and Management of Innovations","topic":"Legal, Health, Environmental and COVID-19 Challenges","field":"Medicine","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Harm; Scopus; Criminal law; Political science; Mistake; Drug control; Law; Business; Criminology; Psychology; MEDLINE","score_opus":0.04455892307830201,"score_gpt":0.33086891414412944,"score_spread":0.28630999106582744,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3091797487","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8093931,0.0013475636,0.000024319741,0.14934218,0.000021391885,0.00076887663,0.000011712459,0.000016783579,0.03907409],"genre_scores_gemma":[0.9855869,0.009067402,0.00026290352,0.0048733805,0.000009396698,0.00003111218,0.0000090856865,0.0000071415457,0.00015269328],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.998758,0.00013245837,0.0004531761,0.00021836981,0.00026792084,0.0001700493],"domain_scores_gemma":[0.99938345,0.00019776316,0.00016419115,0.0001960718,0.000019203782,0.00003932342],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.001694506,0.000100608275,0.00024086941,0.004103399,0.00015086189,0.000023770319,0.00009584232,0.000021449607,0.000015661812],"category_scores_gemma":[0.00004202722,0.00006794341,0.000028014181,0.02232124,0.00014857516,0.000043794927,0.00014586152,0.00013986648,5.3601946e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00039818935,0.0003571457,0.01213087,0.004496809,0.00057779194,0.000019457699,0.0014168455,0.000027651206,0.000039102368,0.95323545,0.0007576463,0.026543064],"study_design_scores_gemma":[0.0010826448,0.00043145745,0.95899534,0.00065291615,0.00031130933,4.0820848e-7,0.011074626,0.0010089232,0.000056305194,0.0008570629,0.025408203,0.00012082386],"about_ca_topic_score_codex":0.0005603091,"about_ca_topic_score_gemma":0.00022657902,"teacher_disagreement_score":0.9523784,"about_ca_system_score_codex":0.000036935544,"about_ca_system_score_gemma":0.0000073402916,"threshold_uncertainty_score":0.9984598},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3093133284","doi":"10.5539/ijef.v12n11p1","title":"Governance and Performance in Insurance Companies: A Bibliometric Analysis and A Meta-Analysis","year":2020,"lang":"en","type":"article","venue":"International Journal of Economics and Finance","topic":"Insurance and Financial Risk Management","field":"Economics, Econometrics and Finance","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Corporate governance; Accounting; Meta-analysis; Point (geometry); Quantitative analysis (chemistry); Business; Insurance industry; Empirical research; Actuarial science; Finance; Statistics","score_opus":0.04692001812354021,"score_gpt":0.24029358453952596,"score_spread":0.19337356641598574,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3093133284","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9755288,0.01940649,0.0012164119,0.0028963261,0.0001014553,0.000082726096,0.0002456539,0.0000027386964,0.000519424],"genre_scores_gemma":[0.91194123,0.08621774,0.0009072156,0.0008006606,0.000056513694,0.0000060284397,0.000002705403,0.000008920295,0.000058963196],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99833196,0.000011719199,0.0010350529,0.0003815215,0.000058520734,0.000181245],"domain_scores_gemma":[0.9985639,0.00006362469,0.0010859153,0.00012336808,0.00009911999,0.000064099135],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005363098,0.00018239292,0.0012416563,0.0076540546,0.0000469337,0.00021338878,0.00031370798,0.000062639636,0.000023849067],"category_scores_gemma":[0.000075655706,0.00018995743,0.0003608827,0.009602693,0.00008898621,0.0008677745,0.000120394005,0.00018282753,0.000004550288],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020841531,0.00008652913,0.8039316,0.000030331466,0.03716482,0.00003261187,0.00073550735,0.017233506,6.5975917e-7,0.13309379,0.000030137282,0.007452077],"study_design_scores_gemma":[0.0006641095,0.00011626934,0.95078063,0.0000029168946,0.003580198,0.000007840469,0.00002365172,0.031464066,0.000006333357,0.002453551,0.010681215,0.00021919327],"about_ca_topic_score_codex":0.00016770096,"about_ca_topic_score_gemma":0.00011471709,"teacher_disagreement_score":0.14684904,"about_ca_system_score_codex":0.000049599083,"about_ca_system_score_gemma":0.000015444155,"threshold_uncertainty_score":0.77462393},"labels":[{"model":"gemma","categories":["bibliometrics","metaepi_broad"],"domain":null,"study_design":"meta_analysis","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"meta_analysis","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W3093745611","doi":"","title":"A bibliometrics analysis of Canadian electrical and computer engineering institutions based on IEEE journal publications (1996-2006)","year":2012,"lang":"en","type":"article","venue":"","topic":"Second Language Learning and Teaching","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Calgary","funders":"","keywords":"Computer science; Bibliometrics; Data science; Library science","score_opus":0.05696400049386227,"score_gpt":0.2533807623015143,"score_spread":0.19641676180765202,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3093745611","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.90881795,0.00042761155,0.03457611,0.00064163475,0.0005257388,0.00008088766,0.000050535895,0.00009447078,0.05478508],"genre_scores_gemma":[0.9977976,0.0000054832126,0.0009903398,0.00043336613,0.000346166,0.0000018400568,0.000017414392,0.0000054450047,0.00040233237],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9994441,0.000026740452,0.00014238813,0.0000676655,0.00011043315,0.00020862477],"domain_scores_gemma":[0.9994253,0.00015272711,0.000044234203,0.00009084276,0.00007175358,0.00021513994],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0003006163,0.00006542211,0.00012187325,0.02590407,0.00028540546,0.00021064562,0.00006608297,0.000028537657,0.0009736689],"category_scores_gemma":[0.00010140334,0.00005350517,0.00006695271,0.004549094,0.000028593891,0.0001868857,0.0000042163665,0.00021651435,0.0000050459316],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000063467983,0.00042026714,0.12270932,0.00002468579,0.0013905776,0.000006712533,0.025441628,0.044956516,0.00003254834,0.7569997,0.01479555,0.033216182],"study_design_scores_gemma":[0.00031081834,0.000106328705,0.098813884,0.000015798185,0.00044702852,0.000014158868,0.0006618316,0.42503038,0.00001780314,0.0000055832684,0.4742942,0.0002821865],"about_ca_topic_score_codex":0.029749526,"about_ca_topic_score_gemma":0.031229801,"teacher_disagreement_score":0.75699407,"about_ca_system_score_codex":0.00004179588,"about_ca_system_score_gemma":0.00007020424,"threshold_uncertainty_score":0.99993956},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W3093804023","doi":"10.1108/lht-06-2020-0131","title":"A bibliometric analysis and science mapping of scientific publications of Alzahra University during 1986–2019","year":2020,"lang":"en","type":"article","venue":"Library Hi Tech","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":47,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Scopus; Originality; Ranking (information retrieval); Library science; Scientific literature; Data science; Computer science; Medicine; Information retrieval; MEDLINE; Social science; Sociology; Political science","score_opus":0.33417652347373894,"score_gpt":0.43836533492749663,"score_spread":0.10418881145375769,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3093804023","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99144864,0.00048242568,0.0026773633,0.0017180298,0.00006115975,0.00015430702,0.00012780596,0.000048643862,0.0032816043],"genre_scores_gemma":[0.99369335,0.00017082132,0.0045051114,0.00001891785,0.000011365487,6.011137e-7,0.000006532494,0.0000053058884,0.0015879646],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99263674,0.00009555966,0.0005785228,0.00090407685,0.005382686,0.00040242798],"domain_scores_gemma":[0.9955205,0.0011651944,0.0004267458,0.00085782644,0.0014828937,0.000546855],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.005263601,0.000105270745,0.0003729937,0.79421437,0.0004605848,0.0016960652,0.003536855,0.00006479401,0.0004450536],"category_scores_gemma":[0.009287046,0.00008845324,0.00016340589,0.97978556,0.0012393005,0.003964498,0.00236629,0.0001603687,0.00002296134],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011141001,0.000073577474,0.93105024,0.00002131326,0.000041158746,0.0000021591409,0.00017983855,0.000018508266,0.04628126,0.0016258422,0.0015244617,0.019170487],"study_design_scores_gemma":[0.00018724122,0.00004408168,0.9617679,0.000004778127,0.000021092686,0.0000010479138,0.00037133088,0.004327313,0.02830207,0.00042138115,0.004450779,0.000100989346],"about_ca_topic_score_codex":0.000038533442,"about_ca_topic_score_gemma":9.123255e-7,"teacher_disagreement_score":0.18557116,"about_ca_system_score_codex":0.000024173856,"about_ca_system_score_gemma":0.00045813707,"threshold_uncertainty_score":0.9993403},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W3094383562","doi":"10.1177/1178633720962935","title":"Bibliometric Analysis of Early COVID-19 Research: The Top 50 Cited Papers","year":2020,"lang":"en","type":"review","venue":"Infectious Diseases Research and Treatment","topic":"COVID-19 Clinical Research Studies","field":"Medicine","cited_by":62,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; McGill University","funders":"","keywords":"Coronavirus disease 2019 (COVID-19); Scopus; Web of science; Pandemic; Bibliometrics; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Library science; Citation; 2019-20 coronavirus outbreak; MEDLINE; Medicine; Political science; Computer science; Pathology","score_opus":0.3381437134218773,"score_gpt":0.5765897513072069,"score_spread":0.2384460378853296,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3094383562","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016989079,0.98922324,0.0000060725015,0.0019941945,0.00004097008,0.004751209,0.0012949544,0.000107198786,0.00088327576],"genre_scores_gemma":[0.030096667,0.96660995,0.0000037990217,0.00013344057,0.00022939543,0.0017547115,0.00034046534,0.000065570166,0.0007660212],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9892765,0.0030217753,0.0010603736,0.0014021584,0.0038379333,0.0014012385],"domain_scores_gemma":[0.92902756,0.06474802,0.00025581496,0.0015977104,0.0013740086,0.0029968875],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0029263138,0.0006110547,0.00356553,0.108156465,0.0010242289,0.00026319147,0.0005331004,0.0003140233,0.0005845801],"category_scores_gemma":[0.08261615,0.00033129816,0.0018593527,0.2739238,0.0025382747,0.00007165097,0.0010140209,0.001526522,0.00010499143],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00048861455,0.0020567433,0.060792826,0.011449888,0.024523187,0.000613165,0.00014814839,0.0000039439237,3.3627916e-7,0.00015578,0.0008951985,0.8988722],"study_design_scores_gemma":[0.0015729861,0.0066337828,0.039883606,0.0009062972,0.010319577,0.0000035584696,0.000057662957,0.000016450052,1.7960512e-7,0.00019773001,0.94017756,0.00023060205],"about_ca_topic_score_codex":0.0051100748,"about_ca_topic_score_gemma":0.00021509564,"teacher_disagreement_score":0.93928236,"about_ca_system_score_codex":0.0027001828,"about_ca_system_score_gemma":0.0061260513,"threshold_uncertainty_score":0.99991393},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3095259702","doi":"10.3390/jrfm13110264","title":"Challenges and Trends in Sustainable Corporate Finance: A Bibliometric Systematic Review","year":2020,"lang":"en","type":"article","venue":"Journal of risk and financial management","topic":"Sustainable Supply Chain Management","field":"Business, Management and Accounting","cited_by":68,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Stakeholder; Corporate finance; Scopus; Sustainability; Business; Corporate sustainability; Stakeholder engagement; Corporate social responsibility; Sustainable development; Competitive advantage; Accounting; Finance; Economics; Political science; Public relations; Marketing; Management","score_opus":0.02511664582110678,"score_gpt":0.22813115605387194,"score_spread":0.20301451023276515,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3095259702","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06427904,0.890828,0.0042225556,0.015686193,0.0005661429,0.003944526,0.0000044135363,0.00011436027,0.02035474],"genre_scores_gemma":[0.70387954,0.29225928,0.00026226885,0.0027955864,0.0003588514,0.000045522072,0.0000021535964,0.000027714266,0.00036910365],"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978995,0.000053844375,0.0008776168,0.0003166584,0.000441164,0.00041125016],"domain_scores_gemma":[0.99815345,0.0000560202,0.0013168022,0.0001878315,0.00024714076,0.000038752012],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.002089577,0.0002639125,0.00087043777,0.011840825,0.000111563706,0.00020363485,0.0003018386,0.000054634893,0.000016605669],"category_scores_gemma":[0.0006974782,0.00022216006,0.00010899276,0.016184822,0.000047531714,0.0010138856,0.00045698334,0.000253632,0.000008529476],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018986195,0.00018979257,0.0027786924,0.52281886,0.00008789528,0.002766162,0.0005239741,0.00011323074,3.5359153e-7,0.14351213,0.009053453,0.31796557],"study_design_scores_gemma":[0.010109607,0.00072176335,0.39665008,0.071597695,0.004601655,0.00007503261,0.0232074,0.002346513,0.0000016574357,0.04991242,0.43862706,0.0021491204],"about_ca_topic_score_codex":0.000042216416,"about_ca_topic_score_gemma":0.000013883313,"teacher_disagreement_score":0.63960046,"about_ca_system_score_codex":0.00006184382,"about_ca_system_score_gemma":0.000012280201,"threshold_uncertainty_score":0.99935913},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3097660013","doi":"10.1139/cjfr-2020-0311","title":"Paradigm shifts in forestry and forest research: a bibliometric analysis","year":2020,"lang":"en","type":"article","venue":"Canadian Journal of Forest Research","topic":"Forest Management and Policy","field":"Environmental Science","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"University of British Columbia","funders":"","keywords":"Scope (computer science); Forestry; Climate change; Forest management; Geography; Computer science; Ecology","score_opus":0.13220011271372203,"score_gpt":0.3649253343340905,"score_spread":0.2327252216203685,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3097660013","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9761311,0.0006515865,0.000020801213,0.00841848,0.000038461498,0.0002781249,0.000010276953,0.000004517122,0.014446682],"genre_scores_gemma":[0.99876416,0.00022075175,0.0001360068,0.00017881897,0.0001745021,0.000009797203,0.0000050050603,0.000022608347,0.00048835867],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9958602,0.0004505195,0.0005192555,0.00036563489,0.00144801,0.0013563329],"domain_scores_gemma":[0.99678355,0.0004755508,0.00010571023,0.0003143279,0.00008318798,0.0022376426],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0046155844,0.00016475546,0.00035659753,0.058268916,0.0003602736,0.0003727892,0.0010493734,0.00013156327,0.0020222066],"category_scores_gemma":[0.0015256052,0.00014821091,0.00013768168,0.1540226,0.0010536596,0.0005635065,0.00026198154,0.0013008476,0.0004445996],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000044524062,0.00001985929,0.96683466,0.000022227343,0.00006674282,0.0006877406,0.0008433982,0.0028417213,0.0000043188,0.0025245727,0.023914218,0.0021959874],"study_design_scores_gemma":[0.00047032585,0.00041637986,0.9492857,0.000026693706,0.000026688074,0.000009880919,0.00011498669,0.0020420172,0.0000027853407,0.0054302327,0.042036515,0.0001377808],"about_ca_topic_score_codex":0.07709766,"about_ca_topic_score_gemma":0.5226211,"teacher_disagreement_score":0.4455234,"about_ca_system_score_codex":0.00045671692,"about_ca_system_score_gemma":0.00047686172,"threshold_uncertainty_score":0.9988901},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3099479995","doi":"10.1016/j.jtcms.2020.11.001","title":"Key elements that determine the efficacy of cupping therapy: A bibliometric analysis and review of clinical studies","year":2020,"lang":"en","type":"article","venue":"Journal of Traditional Chinese Medical Sciences","topic":"Acupuncture Treatment Research Studies","field":"Medicine","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Natural Science Foundation of China","keywords":"Cupping therapy; Medicine; Clinical trial; Intensive care medicine; Alternative medicine; Acupuncture; Internal medicine; Pathology","score_opus":0.3655884592953913,"score_gpt":0.5058160740352701,"score_spread":0.14022761473987877,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3099479995","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8428864,0.102811046,0.00007277887,0.05383933,0.000079443525,0.00020875785,0.000009195458,0.0000029726502,0.00009008946],"genre_scores_gemma":[0.818391,0.17859249,0.0009869415,0.0016560075,0.00036276883,0.000002902903,0.0000023457105,0.000002510329,0.0000029862495],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99504536,0.00025459714,0.0012045071,0.00017512927,0.0031671873,0.0001532249],"domain_scores_gemma":[0.99210715,0.006164253,0.00089479744,0.00009439665,0.0004049935,0.00033441794],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.005084617,0.00015495195,0.0013179926,0.0035281566,0.00010831108,0.000010047861,0.0003299764,0.000035967023,0.00025877345],"category_scores_gemma":[0.022222772,0.000051356852,0.0005604567,0.027997926,0.0014924669,0.00013582713,0.000080989135,0.00029608826,4.1957847e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023975407,0.0005951699,0.9499437,0.0010950988,0.004620025,0.000082287945,0.00031788737,0.0000018137625,0.000050050385,0.00003901142,0.0023253814,0.040689822],"study_design_scores_gemma":[0.0017555638,0.0023537984,0.9919827,0.0017829506,0.0009316679,0.000095751486,0.00010695668,0.00015014413,0.000036867394,0.00023375092,0.0005192261,0.00005063414],"about_ca_topic_score_codex":0.0000051227407,"about_ca_topic_score_gemma":0.000002400553,"teacher_disagreement_score":0.075781435,"about_ca_system_score_codex":0.000018215465,"about_ca_system_score_gemma":0.00033218405,"threshold_uncertainty_score":0.99266243},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3106155822","doi":"10.3389/frma.2020.595370","title":"Intellectual Structure of Coronavirus Research: A Perspective From an Author Cocitation Analysis","year":2020,"lang":"en","type":"article","venue":"Frontiers in Research Metrics and Analytics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta; University of Manitoba","funders":"","keywords":"Outbreak; Coronavirus; Identification (biology); Public health; Perspective (graphical); Transmission (telecommunications); Bibliometrics; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Coronavirus disease 2019 (COVID-19); Geography; Medicine; Virology; Disease; Biology; Library science; Infectious disease (medical specialty); Computer science; Ecology; Pathology","score_opus":0.8795814717877448,"score_gpt":0.6823888219904412,"score_spread":0.19719264979730367,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3106155822","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.90371364,0.006336481,0.084076606,0.0033640692,0.0002491822,0.00054072874,0.0007193145,0.000016421169,0.0009835608],"genre_scores_gemma":[0.9866938,0.0013491884,0.01150338,0.00006093083,0.000111011715,0.0000041473477,0.000035755682,0.000018654933,0.00022312297],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9766464,0.002986395,0.0011176058,0.0015184601,0.016639195,0.0010919196],"domain_scores_gemma":[0.96382105,0.018770793,0.00027479528,0.00097563566,0.014824614,0.0013331069],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.035941552,0.00021356458,0.00092683366,0.14512457,0.00041179624,0.0018260726,0.0027489616,0.00030681232,0.00033915494],"category_scores_gemma":[0.25201342,0.00016951824,0.00021409181,0.49961665,0.0012497125,0.00062126265,0.0011753195,0.001862464,0.000016140442],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00066430727,0.00058742723,0.7659837,0.00003155335,0.0009401109,0.00009888388,0.020392997,0.0019899826,0.002327941,0.0071810293,0.040648352,0.15915374],"study_design_scores_gemma":[0.0008294172,0.0012707245,0.108997375,0.000011494142,0.00008749596,5.2915055e-7,0.08093219,0.7016794,0.0012795653,0.099200994,0.005393969,0.00031685317],"about_ca_topic_score_codex":0.0023613914,"about_ca_topic_score_gemma":0.00033759518,"teacher_disagreement_score":0.6996894,"about_ca_system_score_codex":0.00050336117,"about_ca_system_score_gemma":0.0005741062,"threshold_uncertainty_score":0.9992101},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3107444849","doi":"10.2196/19391","title":"Scientific Publication Patterns of Mobile Technologies and Apps for Posttraumatic Stress Disorder Treatment: Bibliometric Co-Word Analysis","year":2020,"lang":"en","type":"article","venue":"JMIR mhealth and uhealth","topic":"Digital Mental Health Interventions","field":"Psychology","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Analytics; Psychology; Library science; Computer science; Data science","score_opus":0.11185176299854939,"score_gpt":0.45576507447534625,"score_spread":0.3439133114767969,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3107444849","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9860243,0.004421327,0.00070255925,0.0044181305,0.00014703252,0.0020066793,0.0019364432,0.0001511,0.00019242518],"genre_scores_gemma":[0.9973966,0.00030185113,0.00019938839,0.00019455109,0.000017388951,0.001109654,0.0005337582,0.000018913837,0.00022789043],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99792814,0.000078135185,0.00075695646,0.0005664322,0.0001825546,0.00048780098],"domain_scores_gemma":[0.99852574,0.00019936248,0.0004532429,0.00034198552,0.00013625163,0.00034342043],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00031687506,0.00018142436,0.00047272985,0.0071581653,0.00027891522,0.00012135728,0.00016263998,0.00011760391,0.00006700406],"category_scores_gemma":[0.00006490657,0.00015908138,0.00010794254,0.014066855,0.00016863775,0.00020882142,0.00004154416,0.00010100634,0.00001231426],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017484708,0.0015218758,0.19593301,0.0042386176,0.0002059834,7.4527225e-7,0.0026162062,0.0000022344796,0.0000034693903,0.003765177,0.00088417466,0.79065365],"study_design_scores_gemma":[0.0045595383,0.01163408,0.9458572,0.00020612436,0.00084327895,0.000010468348,0.018270716,0.0012606996,0.00020709084,0.0011528399,0.0154782515,0.0005196751],"about_ca_topic_score_codex":0.0004441045,"about_ca_topic_score_gemma":0.00025018025,"teacher_disagreement_score":0.79013395,"about_ca_system_score_codex":0.000075404154,"about_ca_system_score_gemma":0.00008088438,"threshold_uncertainty_score":0.67586577},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3107982530","doi":"10.3389/fpsyg.2020.550716","title":"The 100 Top-Cited Studies on Neuropsychology: A Bibliometric Analysis","year":2020,"lang":"en","type":"article","venue":"Frontiers in Psychology","topic":"Cancer-related cognitive impairment studies","field":"Medicine","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Neuropsychology; Web of science; Impact factor; Bibliometrics; Psychology; Clinical neuropsychology; Library science; Meta-analysis; Psychiatry; Medicine; Political science; Computer science; Cognition","score_opus":0.056041809598530196,"score_gpt":0.3918320793970088,"score_spread":0.3357902697984786,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3107982530","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.79805136,0.028686546,0.00592444,0.14175986,0.008647249,0.0013994527,0.000026159505,0.0002820055,0.015222918],"genre_scores_gemma":[0.9395916,0.013657945,0.0010120436,0.044713207,0.00039822437,0.0001594192,0.000014643451,0.000045119847,0.00040778363],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99770486,0.0002581132,0.00053394807,0.00071465515,0.0002884249,0.00049998687],"domain_scores_gemma":[0.99873996,0.0003067917,0.0001759862,0.00045481336,0.00017471985,0.00014772295],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00038969552,0.000258148,0.0007798785,0.02759308,0.00016160274,0.000016489472,0.00024144459,0.00014967652,0.000046012956],"category_scores_gemma":[0.0014338507,0.00018375184,0.00023747627,0.13240506,0.0005061914,0.000039552047,0.000077235425,0.0006270523,0.00008443241],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0028418193,0.00028072085,0.4833801,0.000022199445,0.005440825,0.00023222696,0.0012096479,0.000029749714,0.0000768603,0.000031304036,0.3991537,0.10730083],"study_design_scores_gemma":[0.004915955,0.0027693978,0.9071261,0.000033209126,0.0013108369,0.00002456331,0.001903836,0.00040170844,0.000038680548,0.00070708565,0.08050411,0.0002644835],"about_ca_topic_score_codex":0.0000046749365,"about_ca_topic_score_gemma":0.0000071566624,"teacher_disagreement_score":0.42374602,"about_ca_system_score_codex":0.00009696926,"about_ca_system_score_gemma":0.000022127313,"threshold_uncertainty_score":0.9834283},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3111126926","doi":"10.1007/s11229-020-02915-6","title":"Eight journals over eight decades: a computational topic-modeling approach to contemporary philosophy of science","year":2020,"lang":"en","type":"article","venue":"Synthese","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":20,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Canada Research Chairs; Canada Foundation for Innovation","keywords":"Philosophy of science; Philosophy of language; Philosophy of computer science; Epistemology; History and philosophy of science; Empiricism; Philosophy of biology; Field (mathematics); Object (grammar); Western philosophy; Metaphysics; Social science; Sociology; Philosophy; Mathematics; Linguistics","score_opus":0.17967260698535828,"score_gpt":0.40149572600388855,"score_spread":0.22182311901853027,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3111126926","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.22954717,0.000816112,0.5639122,0.016526984,0.00023345844,0.00046331182,0.000030802195,0.00006787362,0.1884021],"genre_scores_gemma":[0.95703113,0.000006266332,0.041660808,0.0008478362,0.00037896115,0.000006120716,0.000002941711,0.000006562337,0.000059380218],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99784577,0.00023240868,0.00036732166,0.00031771205,0.0010274062,0.00020940838],"domain_scores_gemma":[0.9987287,0.0003672732,0.00012945273,0.000098604585,0.00035320804,0.00032275423],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001516164,0.000102235346,0.00025745266,0.00023142822,0.00042156572,0.00009118524,0.00049324566,0.000045133205,0.00016933485],"category_scores_gemma":[0.0010091689,0.00008945991,0.00013211904,0.0014647653,0.00044418714,0.00033167299,0.000099036704,0.00011540546,0.000018997916],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018553158,0.00062436616,0.009603049,0.00012930411,0.00028101032,0.000015601161,0.09922926,0.2772897,0.0027108626,0.5792299,0.0014943965,0.029207034],"study_design_scores_gemma":[0.0005983724,0.00010247431,0.0035892257,0.00018915467,0.00006829432,0.0000031028962,0.006936365,0.764739,0.0005867271,0.21017869,0.012356806,0.00065182877],"about_ca_topic_score_codex":0.00019808306,"about_ca_topic_score_gemma":0.000003904394,"teacher_disagreement_score":0.7274839,"about_ca_system_score_codex":0.00005049388,"about_ca_system_score_gemma":0.0005379089,"threshold_uncertainty_score":0.3648069},"labels":[{"model":"gemma","categories":["bibliometrics","sts"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3113578509","doi":"10.29289/25945394202020200052","title":"Complex fibroadenoma: bibliometric literature review and presentation of a clinical case","year":2020,"lang":"en","type":"article","venue":"Mastology","topic":"Breast Lesions and Carcinomas","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Discovery Air (Canada)","funders":"","keywords":"Fibroadenoma; Medicine; Web of science; Pathological; Population; Breast cancer; Pathology; Presentation (obstetrics); Apocrine; Stromal cell; Cancer; Internal medicine; Radiology; Meta-analysis","score_opus":0.12490220636511605,"score_gpt":0.40521307861756567,"score_spread":0.2803108722524496,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3113578509","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.82718223,0.13700047,0.00018732007,0.03378033,0.00006193674,0.00045291852,0.000030719395,0.000027895434,0.0012761673],"genre_scores_gemma":[0.9895577,0.0047053644,0.0011524577,0.00438315,0.000092758586,0.0000049405494,0.000029404611,0.000006389519,0.00006787279],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.999221,0.00010440331,0.0003402134,0.00017632729,0.00006827614,0.00008975817],"domain_scores_gemma":[0.9994106,0.00013010872,0.00009525318,0.00012402073,0.00009374202,0.00014626472],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00013464429,0.00006612129,0.00044837638,0.0010866896,0.000017534892,0.0000039993806,0.00003191865,0.00008641966,0.00020208737],"category_scores_gemma":[0.00037899945,0.000051135306,0.00006526077,0.006878022,0.0000824307,0.000027960003,0.00004820246,0.00015677662,0.000009146944],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00091529236,0.00057296985,0.4487996,0.022984866,0.00031279243,0.019004703,0.0008852041,7.099808e-7,0.0023292652,0.0015855951,0.20188592,0.30072308],"study_design_scores_gemma":[0.001672246,0.00088053837,0.9442954,0.0003247989,0.00037734312,0.026425395,0.000059019738,0.00018687561,0.000017382625,0.00001833993,0.02566681,0.00007586946],"about_ca_topic_score_codex":0.000008640448,"about_ca_topic_score_gemma":0.0000010621715,"teacher_disagreement_score":0.49549577,"about_ca_system_score_codex":0.000002393827,"about_ca_system_score_gemma":0.000025661646,"threshold_uncertainty_score":0.33046615},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W3115533035","doi":"10.1016/j.joi.2020.101124","title":"Exploring the interdisciplinarity patterns of highly cited papers","year":2020,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":56,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"National Office for Philosophy and Social Sciences","keywords":"Computer science; Data science; Bibliometrics; Information retrieval; World Wide Web","score_opus":0.7915538673463653,"score_gpt":0.540354756721656,"score_spread":0.25119911062470934,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3115533035","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9832191,0.00021952298,0.007797399,0.0042846175,0.0007978045,0.00009901529,0.00002178753,0.000005640619,0.0035551523],"genre_scores_gemma":[0.99818474,0.0006638176,0.00051278423,0.00035665493,0.00018904897,0.0000010789245,4.5068538e-7,0.000006738394,0.000084659405],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9885271,0.00011164795,0.0017717608,0.0001793483,0.009060611,0.0003494958],"domain_scores_gemma":[0.9864179,0.0062540085,0.0015167717,0.00041082143,0.0048870007,0.00051350374],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.014953543,0.00012721922,0.0004413208,0.04018199,0.00014953327,0.00083033653,0.0033085796,0.00005683511,0.00016279289],"category_scores_gemma":[0.0823548,0.00006357809,0.0003950083,0.19905947,0.00009759679,0.001463883,0.0009447991,0.0006211129,0.00003357273],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018131682,0.00013200495,0.36896878,0.000035729536,0.00009770814,0.0000447281,0.0032025396,0.0005698271,0.0009917931,0.0008397754,0.007894688,0.6170411],"study_design_scores_gemma":[0.0012123319,0.0017710455,0.89816916,0.000045394376,0.000032070868,0.000051922692,0.009519179,0.0028481889,0.006095413,0.00079033535,0.07922098,0.00024399467],"about_ca_topic_score_codex":0.000014560274,"about_ca_topic_score_gemma":0.0000022514587,"teacher_disagreement_score":0.6167971,"about_ca_system_score_codex":0.00008873425,"about_ca_system_score_gemma":0.00021101936,"threshold_uncertainty_score":0.97069675},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3118676337","doi":"","title":"Bibliometric Analysis of Worldwide Coronavirus Research based on Web of Science between 1970 and February 2020","year":2020,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Centrality; Betweenness centrality; Web of science; Bibliometrics; Social network analysis; Webometrics; Data science; Closeness; Library science; Citation; China; Geography; Coronavirus; Coronavirus disease 2019 (COVID-19); Pandemic; Descriptive statistics; Computer science; World Wide Web; Political science; Statistics; MEDLINE; Social media; Medicine; Mathematics","score_opus":0.7796599056086541,"score_gpt":0.6744480922062391,"score_spread":0.10521181340241492,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3118676337","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9885074,0.006947482,0.00043690656,0.0016133116,0.00005655094,0.0005389332,0.00016761993,0.000026352847,0.0017054324],"genre_scores_gemma":[0.9930521,0.0052587045,0.0011157305,0.00043449653,0.000069681926,0.000019569723,0.000004152504,0.000025177245,0.000020369647],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99355984,0.0010045005,0.0016487847,0.00077066233,0.002431292,0.0005849301],"domain_scores_gemma":[0.97801083,0.01838508,0.0013799705,0.00064563093,0.0010770233,0.000501482],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.012234143,0.00031125342,0.0021561156,0.05272433,0.00038827374,0.00031638003,0.00354975,0.00011184594,0.0025199507],"category_scores_gemma":[0.03514511,0.00024349158,0.0003506104,0.22961348,0.0019904242,0.0007911102,0.0026026615,0.0006792442,0.0000050278936],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022450944,0.00022367443,0.96075565,0.00028101308,0.00062052044,0.00001448038,0.00010299579,0.0003392964,0.01674668,0.00017634891,0.010600679,0.009914186],"study_design_scores_gemma":[0.0004066617,0.00007788795,0.9810546,0.00024470536,0.00051431695,2.549133e-7,0.00006521794,0.0019995288,0.007207921,0.00767883,0.00051638414,0.0002336682],"about_ca_topic_score_codex":0.0018716827,"about_ca_topic_score_gemma":0.00008046452,"teacher_disagreement_score":0.17688915,"about_ca_system_score_codex":0.00015438476,"about_ca_system_score_gemma":0.0004003432,"threshold_uncertainty_score":0.99839187},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3119244696","doi":"10.3390/rel12010049","title":"Post-Islamism and Intellectual Production: A Bibliometric Analysis of the Evolution of Contemporary Islamic Thought","year":2021,"lang":"en","type":"article","venue":"Religions","topic":"Islamic Finance and Banking Studies","field":"Business, Management and Accounting","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Islam; Context (archaeology); Empowerment; Sociology; Restructuring; Reading (process); Social science; Relation (database); Epistemology; Political science; Law; History; Philosophy; Theology","score_opus":0.014653070539708074,"score_gpt":0.22746183873266665,"score_spread":0.21280876819295858,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3119244696","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98743683,0.005105396,0.00017584625,0.0017560057,0.00027127995,0.0001042153,0.000007341214,0.0000217371,0.0051213563],"genre_scores_gemma":[0.998233,0.00061594084,0.000036462956,0.00012782501,0.00014261369,0.0000050510866,0.000011438889,0.0000069208218,0.0008207465],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9992159,0.000014841567,0.0002696147,0.00019953008,0.00019222582,0.000107909276],"domain_scores_gemma":[0.9989372,0.00007916452,0.00022118278,0.000260876,0.00050054246,0.0000010287783],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00017834794,0.00009203779,0.000260133,0.010183645,0.000149096,0.00002753527,0.00010928639,0.000042110772,0.000014584827],"category_scores_gemma":[0.00078347285,0.00006726421,0.00015713426,0.08315989,0.00013271198,0.00021343373,0.00018959865,0.00007828253,0.000003729447],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012279308,0.0008139387,0.39388844,0.0011832439,0.0043331366,0.000016336824,0.002478434,0.0004878679,0.054827888,0.38467002,0.13895328,0.018224629],"study_design_scores_gemma":[0.0004229722,0.000024550145,0.97241634,0.00025183396,0.001772055,0.0000072399607,0.0011751807,0.0010705767,0.0023293986,0.0120632965,0.00819669,0.00026986978],"about_ca_topic_score_codex":0.0006827366,"about_ca_topic_score_gemma":0.00019620782,"teacher_disagreement_score":0.57852787,"about_ca_system_score_codex":0.000021212792,"about_ca_system_score_gemma":0.00006333237,"threshold_uncertainty_score":0.936328},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3119634921","doi":"10.18438/eblip29764","title":"Dissertations and Theses in Top Nursing Publications: A Bibliometric Study","year":2020,"lang":"en","type":"article","venue":"Evidence Based Library and Information Practice","topic":"Health Sciences Research and Education","field":"Health Professions","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Citation; Nursing research; Scholarship; Bibliometrics; Nursing literature; Citation analysis; Scholarly communication; Library science; Medicine; Nursing; Alternative medicine; Publishing; Political science; Computer science","score_opus":0.168932986573074,"score_gpt":0.5012158202472208,"score_spread":0.3322828336741468,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3119634921","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.45614937,0.000688457,0.0009628145,0.53233373,0.00012888317,0.001699564,0.00000618526,0.0001066132,0.007924405],"genre_scores_gemma":[0.9524394,0.0014797451,0.0017354663,0.04393684,0.00007031968,0.00024666506,0.000021048018,0.000004113298,0.00006638786],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9980486,0.00073196116,0.0005133537,0.00015488506,0.00029507038,0.0002561024],"domain_scores_gemma":[0.9932035,0.0059704143,0.00023824633,0.00013394684,0.00011872627,0.0003351316],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.001306785,0.00007411756,0.00010009871,0.0044800704,0.0005991326,0.00026722148,0.00014357852,0.000051748673,0.00030598964],"category_scores_gemma":[0.017273167,0.00006320661,0.000008231058,0.022291565,0.00005082587,0.14648475,0.000046814228,0.00038461245,0.00008024672],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00058102066,0.0004029028,0.7703345,0.00072181073,0.000008518356,0.0000014944245,0.06065277,0.000044641394,0.000025070662,0.06520568,0.0152077805,0.086813845],"study_design_scores_gemma":[0.00035359515,0.0002692767,0.8143222,0.0002908961,0.0000063314687,9.3957357e-7,0.088081434,0.0060476772,0.00000920035,0.000074238946,0.090461604,0.000082657294],"about_ca_topic_score_codex":0.000062114945,"about_ca_topic_score_gemma":6.281628e-7,"teacher_disagreement_score":0.49629006,"about_ca_system_score_codex":0.000020312138,"about_ca_system_score_gemma":0.0009516745,"threshold_uncertainty_score":0.9984901},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3120593931","doi":"10.1101/2021.01.11.21249624","title":"Emergence and Evolution of Big Data Science in HIV Research: Bibliometric Analysis of Federally Sponsored Studies 2000-2019","year":2021,"lang":"en","type":"preprint","venue":"medRxiv","topic":"Data-Driven Disease Surveillance","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Institute of Allergy and Infectious Diseases; National Institutes of Health","keywords":"Big data; Bibliometrics; Human immunodeficiency virus (HIV); China; Library science; Metadata; Data science; Political science; Medicine; Computer science; Family medicine; Data mining; World Wide Web","score_opus":0.24897965493622373,"score_gpt":0.43769117898587756,"score_spread":0.18871152404965383,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3120593931","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9636514,0.033637676,0.00015351536,0.00023651418,0.00030151493,0.00042645747,0.0013812623,0.000020081567,0.00019158916],"genre_scores_gemma":[0.9824844,0.015542933,0.0010811249,0.000008344847,0.00004718463,0.000017730476,0.0006687792,0.000016104972,0.00013338264],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9946912,0.0005631875,0.0009260078,0.0013411774,0.0020228273,0.0004555884],"domain_scores_gemma":[0.9935903,0.0005719842,0.0004155714,0.0030389875,0.0021780136,0.00020516914],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.008930435,0.0002306928,0.0013371848,0.10881316,0.00008527557,0.000044816232,0.0012706032,0.0001192746,0.00006855635],"category_scores_gemma":[0.016596474,0.00021231174,0.000121747245,0.22667614,0.0014608502,0.00025529275,0.0052354126,0.0005282138,0.0000034734912],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016788719,0.0002591514,0.9874648,0.0009396534,0.0008538004,0.00006767342,0.00033024926,0.00016486786,0.006032962,0.000018521421,0.00076261105,0.00293785],"study_design_scores_gemma":[0.00044625168,0.000049798324,0.99242824,0.0006092199,0.0005403941,0.0000022387796,0.00085652474,0.0044370685,0.00036139344,0.000038905597,0.00007783807,0.00015215193],"about_ca_topic_score_codex":0.0010227237,"about_ca_topic_score_gemma":0.0010307311,"teacher_disagreement_score":0.11786298,"about_ca_system_score_codex":0.00021796071,"about_ca_system_score_gemma":0.002313071,"threshold_uncertainty_score":0.9916872},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3121153159","doi":"10.1177/2325967120969902","title":"A Bibliometric Analysis of the Top Cited Articles in Sports and Exercise Medicine","year":2021,"lang":"en","type":"article","venue":"Orthopaedic Journal of Sports Medicine","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":40,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Canadian Sport Centre Pacific; International Collaboration On Repair Discoveries; Simon Fraser University; University of British Columbia","funders":"","keywords":"Sports medicine; Medicine; Web of science; Scopus; Alternative medicine; Bibliometrics; MEDLINE; Citation; Narrative review; Sports science; Family medicine; Medical education; Meta-analysis; Physical therapy; Library science; Internal medicine; Pathology; Political science","score_opus":0.1919176015695751,"score_gpt":0.46394098333202166,"score_spread":0.27202338176244656,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3121153159","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9630334,0.029499887,0.00027927273,0.005589844,0.00072444684,0.00012128947,0.0000047751396,0.0000034377106,0.0007436717],"genre_scores_gemma":[0.9860546,0.01292175,0.00016921511,0.00022278987,0.0001457185,8.529257e-7,0.0000020757661,0.000009167778,0.0004738251],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98064065,0.00020548735,0.0027430458,0.00046292058,0.0154930195,0.00045485675],"domain_scores_gemma":[0.98809385,0.0027342392,0.001838665,0.00085259095,0.0058618034,0.00061887444],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.036704514,0.00019725414,0.0014673942,0.6319511,0.00009201457,0.00009692895,0.0010732047,0.00011822597,0.00263296],"category_scores_gemma":[0.046115216,0.000096232405,0.00033259668,0.90381885,0.00074604235,0.00033543466,0.00036136777,0.0005038519,8.276193e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000039793445,0.00008468944,0.69179386,0.00000806559,0.000059471273,0.00044569222,0.00033822472,0.000055130502,0.00016756645,0.00004836072,0.0013859724,0.3055732],"study_design_scores_gemma":[0.0014144108,0.00011880908,0.9911288,0.00046680245,0.00046987095,0.00010916325,0.0018178029,0.0005100251,0.00012777494,0.0020932518,0.0016511041,0.00009217933],"about_ca_topic_score_codex":0.00008197735,"about_ca_topic_score_gemma":0.000024687857,"teacher_disagreement_score":0.30548102,"about_ca_system_score_codex":0.000057253324,"about_ca_system_score_gemma":0.00037912434,"threshold_uncertainty_score":0.9982788},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3122632720","doi":"10.2174/1874303x02013010039","title":"Top 100 Cited Articles on Peritoneal Dialysis: A Bibliometric Analysis","year":2020,"lang":"en","type":"article","venue":"The Open Urology & Nephrology Journal","topic":"Dialysis and Renal Disease Management","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Peritoneal dialysis; Medicine; Bibliometrics; Nephrology; Internal medicine; Subject (documents); Dialysis; Family medicine; Library science","score_opus":0.03965231964253754,"score_gpt":0.30965837000541063,"score_spread":0.27000605036287306,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3122632720","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.86483836,0.00078132295,0.00059623376,0.12808347,0.00011885581,0.00039416112,0.0000058504525,0.000034215824,0.0051475065],"genre_scores_gemma":[0.91453475,0.00032150286,0.00015902604,0.08407537,0.0005926556,0.00002767505,0.000017968958,0.000021114782,0.00024992213],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99706197,0.0008406276,0.0006211893,0.0004625478,0.0004520842,0.0005615845],"domain_scores_gemma":[0.99820644,0.00025917887,0.00031938314,0.00049826805,0.00017034938,0.00054639357],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0013876358,0.00025854676,0.0009052321,0.017401963,0.0004432165,0.00025335315,0.00095046934,0.00015829034,0.003911414],"category_scores_gemma":[0.00031992325,0.00015252712,0.0006210755,0.06395128,0.00031356886,0.00012947188,0.00047914617,0.0007548936,0.00048779655],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.041436967,0.005475121,0.30969858,0.00006513085,0.113861,0.008837484,0.0074007073,0.025257839,0.015364597,0.0065555368,0.26012838,0.20591867],"study_design_scores_gemma":[0.0057186545,0.0036670512,0.84374267,0.0000078984385,0.029293507,0.00045798218,0.00045253686,0.015210553,0.00017942501,0.00088315853,0.09999906,0.0003875132],"about_ca_topic_score_codex":0.000043951815,"about_ca_topic_score_gemma":0.00001683703,"teacher_disagreement_score":0.5340441,"about_ca_system_score_codex":0.000036361373,"about_ca_system_score_gemma":0.00011566119,"threshold_uncertainty_score":0.99699914},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3123335187","doi":"10.18329/09757597/2020/13209","title":"Global Machine-learning Research: a scientometric assessment of global literature during 2009–18","year":2020,"lang":"en","type":"article","venue":"World Digital Libraries - An international journal","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Subject (documents); Science Citation Index; Index (typography); Citation index; Citation impact; Artificial intelligence; Library science; Per capita; Zhàng; China; Political science; Geography; Computer science; Sociology; Demography; World Wide Web","score_opus":0.2428365320297471,"score_gpt":0.4890182123725702,"score_spread":0.24618168034282312,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3123335187","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9508476,0.0009984473,0.0006063645,0.010971142,0.0010888538,0.00019109079,0.0002783027,0.00007154654,0.03494665],"genre_scores_gemma":[0.9959067,0.00007743353,0.0012558494,0.0004931084,0.0015998427,0.0000034700686,0.00014511206,0.0000125101105,0.0005059511],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9971733,0.000102365506,0.0006150148,0.0002893834,0.0014649393,0.00035497878],"domain_scores_gemma":[0.9978366,0.000109311586,0.00022216096,0.0001322499,0.0010978874,0.000601784],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.00035790208,0.000145981,0.0002198435,0.00055830635,0.00025780004,0.0018399564,0.00047471753,0.00006679672,0.00049521524],"category_scores_gemma":[0.001183164,0.00012896411,0.00013602801,0.003749401,0.0001922025,0.0025729644,0.00018621876,0.0006957932,0.000022853717],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005214108,0.00026829203,0.95972085,0.00004950083,0.00009759442,0.00018958053,0.0006148138,0.00017530228,0.000050266746,0.017309152,0.0017858613,0.019217357],"study_design_scores_gemma":[0.00082555,0.0020069685,0.9049552,0.0010749734,0.000048336293,0.0017118879,0.0032269808,0.009626926,0.00090089743,0.01920853,0.055921484,0.0004922635],"about_ca_topic_score_codex":0.00004272738,"about_ca_topic_score_gemma":0.000012714419,"teacher_disagreement_score":0.054765664,"about_ca_system_score_codex":0.00046740932,"about_ca_system_score_gemma":0.0009230689,"threshold_uncertainty_score":0.99919623},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3124667861","doi":"10.3386/w25742","title":"In-Text Patent Citations: A User’s Guide","year":2019,"lang":"en","type":"report","venue":"National Bureau of Economic Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Office of Behavioral and Social Sciences Research; High Energy Physics; NIH Office of the Director; RIKEN; National Cancer Institute; National Institutes of Health; Eidgenössische Technische Hochschule Zürich; National Jewish Health; IFP Energies Nouvelles; National Institute on Aging; University of Washington","keywords":"Information retrieval; Computer science; World Wide Web","score_opus":0.939468718043896,"score_gpt":0.7311586833684702,"score_spread":0.20831003467542575,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3124667861","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0080128405,0.0023848964,0.00002731809,0.003209434,0.0023212207,0.0012330678,0.0003487979,0.000008960748,0.98245347],"genre_scores_gemma":[0.55058706,0.0076888492,0.0007377835,0.00010776629,0.0010985122,0.00024222836,0.00030282352,0.000080898695,0.4391541],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9570763,0.00084934145,0.0029954878,0.0016489298,0.036335472,0.0010944846],"domain_scores_gemma":[0.94042945,0.022199089,0.00096190476,0.0013273002,0.03469286,0.00038941734],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.14821918,0.0002810675,0.0009703525,0.16583428,0.00016400823,0.0017492283,0.0047069294,0.0006368581,0.0061715115],"category_scores_gemma":[0.16114505,0.00023341057,0.0004130079,0.06624294,0.0004188858,0.0008736836,0.0014937721,0.0014397416,0.0065668384],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000026607113,0.00018278175,0.015465781,0.000050258805,0.000059048856,0.000013936275,0.000039519397,0.0006567341,0.000060816812,0.0712654,0.9034777,0.008701405],"study_design_scores_gemma":[0.00067914976,0.00017585994,0.018558655,0.000078950456,0.000003484979,0.000015638454,0.000180987,0.0019816677,0.00007665264,0.33091083,0.64704573,0.00029237557],"about_ca_topic_score_codex":0.004675769,"about_ca_topic_score_gemma":0.00044871814,"teacher_disagreement_score":0.5432994,"about_ca_system_score_codex":0.005323423,"about_ca_system_score_gemma":0.018227195,"threshold_uncertainty_score":0.99928707},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W3124735532","doi":"10.5539/jms.v11n1p64","title":"People, Planet, and Profit: A Bibliometric Analysis of Triple Bottom Line Theory","year":2021,"lang":"en","type":"article","venue":"Journal of Management and Sustainability","topic":"Sustainable Supply Chain Management","field":"Business, Management and Accounting","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Triple bottom line; Sustainability; Scarcity; Profit (economics); Operations research; Regional science; Environmental economics; Management science; Economics; Computer science; Sociology; Engineering; Neoclassical economics; Microeconomics; Ecology","score_opus":0.008315496690591931,"score_gpt":0.23717890094129726,"score_spread":0.22886340425070534,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3124735532","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9846794,0.003501983,0.0046078097,0.0020789371,0.00024253898,0.00071126164,0.0000061046135,0.000033033637,0.0041389517],"genre_scores_gemma":[0.9965632,0.00064343953,0.0004733732,0.00035531845,0.00018244173,0.000011912396,0.000028054968,0.000019490155,0.001722795],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9974541,0.00012265217,0.00094293925,0.0003924278,0.0006718081,0.0004160858],"domain_scores_gemma":[0.99708056,0.00027595696,0.00080757064,0.0004224768,0.001364045,0.000049410806],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0041180174,0.00026283454,0.000811319,0.035271972,0.00015380005,0.0003337951,0.00028996813,0.00008556721,0.0001977398],"category_scores_gemma":[0.0011230812,0.0002286406,0.00021739054,0.057491187,0.00013097805,0.00092016516,0.0006874323,0.00025349905,0.0000010466481],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006371989,0.00047823775,0.8357912,0.0062520974,0.0035734226,0.0006112556,0.00031179792,0.0008580663,0.000003630102,0.07426888,0.0028904085,0.07432386],"study_design_scores_gemma":[0.001835095,0.00008161525,0.9046197,0.000033682594,0.004469078,0.000008759033,0.023830788,0.0038416854,0.0000075422217,0.030514179,0.030447721,0.0003101853],"about_ca_topic_score_codex":0.00014459607,"about_ca_topic_score_gemma":0.00006654483,"teacher_disagreement_score":0.07401367,"about_ca_system_score_codex":0.00012917486,"about_ca_system_score_gemma":0.00008523014,"threshold_uncertainty_score":0.9756624},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3124916602","doi":"10.1007/s11831-021-09545-2","title":"Integration of BIM and GIS for Construction Automation, a Systematic Literature Review (SLR) Combining Bibliometric and Qualitative Analysis","year":2021,"lang":"en","type":"article","venue":"Archives of Computational Methods in Engineering","topic":"BIM and Construction Integration","field":"Engineering","cited_by":71,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"École de Technologie Supérieure","funders":"","keywords":"Building information modeling; Automation; Computer science; Systems engineering; Geographic information system; Robot; Construction engineering; Data science; Engineering management; Engineering; Artificial intelligence; Operations management; Remote sensing","score_opus":0.020745270271691228,"score_gpt":0.3570799097212898,"score_spread":0.33633463944959857,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3124916602","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013854812,0.031901944,0.9537891,0.000027927048,0.000090610614,0.0002325552,0.000024062374,0.000034968176,0.000044010703],"genre_scores_gemma":[0.22084686,0.0030250545,0.7760086,0.0000049691953,0.0000067824585,0.000045467623,0.000052413227,0.000008329405,0.0000015323984],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988799,0.00019677203,0.0006053554,0.0001353549,0.000109130706,0.00007347235],"domain_scores_gemma":[0.9974486,0.0021339357,0.00013831377,0.000076231976,0.00017464148,0.0000282345],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005380032,0.00011391267,0.00045750898,0.0056404737,0.00001945641,0.00002267104,0.00003715817,0.000041414893,0.0000019880556],"category_scores_gemma":[0.00075866276,0.00011477937,0.00007116684,0.009006096,0.00004710604,0.00015505603,0.000012885812,0.00009151249,2.7963505e-8],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000015365958,0.000034535973,0.00030048875,0.17043495,0.0011210915,0.0000013379911,0.011592801,0.45255762,0.04261892,0.14395824,0.000005327972,0.17735931],"study_design_scores_gemma":[0.00027223345,0.000027919963,0.0054944255,0.015710525,0.00024652664,0.000034060187,0.0008536051,0.95729905,0.004603999,0.015324173,0.0000037043487,0.00012979661],"about_ca_topic_score_codex":0.0000011796805,"about_ca_topic_score_gemma":0.0000012252981,"teacher_disagreement_score":0.50474143,"about_ca_system_score_codex":0.000011889016,"about_ca_system_score_gemma":0.000015754222,"threshold_uncertainty_score":0.50329363},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3126045849","doi":"","title":"The Retraction Penalty: Evidence from the Web of Science","year":2013,"lang":"en","type":"article","venue":"","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Kellogg's (Canada)","funders":"","keywords":"Citation; Web of science; Computer science; Library science; Information retrieval; MEDLINE; Political science; Law","score_opus":0.04295528310462751,"score_gpt":0.33237270555352094,"score_spread":0.28941742244889346,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3126045849","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6832528,0.00059970503,0.00030535233,0.112715326,0.0012056332,0.00033853686,0.000002146122,0.000040868614,0.20153964],"genre_scores_gemma":[0.9940343,0.001090989,0.00015128577,0.0005269457,0.00024710695,0.000005419057,1.2321252e-7,0.0000012718424,0.0039425925],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987127,0.00019336272,0.00012831029,0.00011156125,0.00066316046,0.00019089978],"domain_scores_gemma":[0.99674374,0.0026981398,0.000084676576,0.00016854532,0.0002592993,0.000045586283],"candidate_categories":["sts","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0044293883,0.000037490936,0.000046387955,0.000012926049,0.0016421818,0.0001393721,0.0008297914,0.00015285265,0.0011600613],"category_scores_gemma":[0.0043422426,0.000017168442,0.000027828231,0.0003643725,0.0023210773,0.0007352976,0.000048983453,0.0006127229,0.00024515603],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021187738,0.000025840134,0.040213116,0.000003221097,0.000019592086,3.8158748e-7,0.05908993,0.0000047752546,0.032965764,0.41088766,0.3760312,0.08073733],"study_design_scores_gemma":[0.000113901515,0.00005080496,0.17463812,0.00017851457,0.00002863334,0.0000013598515,0.14284788,0.003082852,0.0123550985,0.105513535,0.5609294,0.0002598484],"about_ca_topic_score_codex":0.08508394,"about_ca_topic_score_gemma":0.0055072885,"teacher_disagreement_score":0.31078148,"about_ca_system_score_codex":0.000044874665,"about_ca_system_score_gemma":0.0005191687,"threshold_uncertainty_score":0.999753},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3127643406","doi":"10.36680/j.itcon.2021.002","title":"Open access to Construction IT research articles – developments over the past 25 years","year":2021,"lang":"en","type":"article","venue":"Journal of Information Technology in Construction","topic":"Intellectual Capital and Performance Analysis","field":"Business, Management and Accounting","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Dozen; Publishing; Quarter (Canadian coin); Quality (philosophy); Business model; Engineering; Library science; Political science; Engineering ethics; World Wide Web; Business; History; Computer science; Marketing","score_opus":0.04679358624725741,"score_gpt":0.331948027033694,"score_spread":0.2851544407864366,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3127643406","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9812782,0.000015936825,0.0011565706,0.011977004,0.0004844512,0.0001477159,7.3914106e-7,0.000014457458,0.004924937],"genre_scores_gemma":[0.9963303,0.0000522851,0.0019514498,0.0014475086,0.00017169816,0.000008377558,0.0000050255117,0.0000037681843,0.000029570148],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99875605,0.000018027924,0.00064000563,0.00007387797,0.00033400825,0.00017801132],"domain_scores_gemma":[0.99829465,0.000038344024,0.0004371708,0.00013499756,0.0010849204,0.000009945001],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.001028179,0.00007248887,0.00014875374,0.0019605362,0.0002048711,0.0011056158,0.00073302136,0.00010010699,0.00037830908],"category_scores_gemma":[0.000621676,0.00005666784,0.00003812394,0.0033035798,0.00015070799,0.007022476,0.0006826519,0.0003924564,0.00019798522],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017227825,0.000048310896,0.30752665,0.000053457377,0.00010279289,0.00001689064,0.00077386387,0.00035591185,0.00042108938,0.050109703,0.02842238,0.61199665],"study_design_scores_gemma":[0.002751309,0.00011320683,0.34814826,0.0005139946,0.0000994246,0.0011247678,0.05961755,0.0021065169,0.012876425,0.11561924,0.4564519,0.00057741744],"about_ca_topic_score_codex":0.000063816165,"about_ca_topic_score_gemma":0.00009472696,"teacher_disagreement_score":0.61141926,"about_ca_system_score_codex":0.00010644224,"about_ca_system_score_gemma":0.00012476585,"threshold_uncertainty_score":0.99993134},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W3133906725","doi":"10.1111/jdv.17196","title":"Global publication productivity in dermatology: a bibliometric description of the past and estimation of the future","year":2021,"lang":"en","type":"article","venue":"Journal of the European Academy of Dermatology and Venereology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Atopic dermatitis; Scopus; Dermatology; Bibliometrics; Family medicine; MEDLINE; Library science; Political science","score_opus":0.18854739366658751,"score_gpt":0.4408241137055947,"score_spread":0.2522767200390072,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3133906725","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9497151,0.004348104,0.00031616414,0.044609733,0.00046661892,0.00011170469,0.000008364759,0.000001327458,0.00042289638],"genre_scores_gemma":[0.99894136,0.00053465227,0.00025262308,0.00017588204,0.000048202743,5.698565e-7,2.3834228e-7,0.0000037792713,0.000042690554],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9939072,0.003345445,0.0011083689,0.00023770152,0.0011893887,0.0002118808],"domain_scores_gemma":[0.99597055,0.0008736603,0.0017556859,0.00031643306,0.0010170969,0.00006658363],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.005946456,0.00009943155,0.0004254645,0.020071656,0.00012849914,0.000059216734,0.0014133914,0.00016898691,0.000008312001],"category_scores_gemma":[0.012723174,0.00004783409,0.00013847933,0.14069498,0.00087538274,0.0003571017,0.0009674225,0.00048709815,8.377576e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002948451,0.00008991735,0.97270566,0.000029536795,0.000022161761,0.0000045564025,0.00007600342,0.00007453083,0.0010791975,0.0019894708,0.0023421457,0.021557303],"study_design_scores_gemma":[0.0004585654,0.000012044308,0.98257375,0.000023944975,0.000023918188,0.003558827,0.000100523626,0.00046624226,0.0032850897,0.007862933,0.0015931033,0.000041067167],"about_ca_topic_score_codex":0.000004417326,"about_ca_topic_score_gemma":0.000009193428,"teacher_disagreement_score":0.12062333,"about_ca_system_score_codex":0.000027980628,"about_ca_system_score_gemma":0.00015322962,"threshold_uncertainty_score":0.9955931},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W3133935200","doi":"10.1097/gox.0000000000003439","title":"Assessment of the Academic Productivity of Plastic Surgeons in Saudi Arabia Using the h-index","year":2021,"lang":"en","type":"article","venue":"Plastic & Reconstructive Surgery Global Open","topic":"Diversity and Career in Medicine","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Index (typography); Medicine; Scopus; Bivariate analysis; Cross-sectional study; Rank (graph theory); Board certification; Family medicine; Psychology; Medical education; Statistics; Mathematics; MEDLINE; Residency training; Political science; Computer science","score_opus":0.053249784834869966,"score_gpt":0.3358156937246997,"score_spread":0.28256590888982974,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3133935200","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98772955,0.00008470821,0.00027642402,0.0004149478,0.0030782484,0.00037944494,0.00013018862,0.000006801841,0.007899659],"genre_scores_gemma":[0.9996912,0.000043949665,0.00009897243,0.00004089329,0.000066405664,0.0000065615504,0.0000027034853,0.0000044345315,0.00004490757],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99683845,0.0010526757,0.00046628882,0.00034776307,0.0009283019,0.00036650852],"domain_scores_gemma":[0.9944865,0.004330819,0.00043304855,0.0002954064,0.00036914874,0.00008507441],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030538826,0.00014346938,0.0005271321,0.000059843253,0.00032839624,0.000052305022,0.00087791437,0.00012385892,0.00020942646],"category_scores_gemma":[0.0069105634,0.00010358965,0.0001755525,0.0014720013,0.0015583803,0.0003310132,0.0006312041,0.00038980553,0.0000012791197],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000034986748,0.00006241145,0.9902723,0.000025687685,0.000077630175,0.000008080987,0.0004320277,0.0003840581,0.0000941391,0.0065174075,0.00020676812,0.0018844919],"study_design_scores_gemma":[0.00023146388,0.0000091397005,0.9799285,0.00040281762,0.00008800673,0.000016623422,0.015818072,0.00014267398,0.00015402696,0.0028032742,0.00027830334,0.00012707704],"about_ca_topic_score_codex":0.005080581,"about_ca_topic_score_gemma":0.005170909,"teacher_disagreement_score":0.015386044,"about_ca_system_score_codex":0.0003708599,"about_ca_system_score_gemma":0.002938383,"threshold_uncertainty_score":0.82730895},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3134007699","doi":"10.1177/1745691620979806","title":"Estimating the Prevalence of Transparency and Reproducibility-Related Research Practices in Psychology (2014–2017)","year":2021,"lang":"en","type":"article","venue":"Perspectives on Psychological Science","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":184,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Fonds de Recherche du Québec - Santé; Stiftung Charité; National Institute on Alcohol Abuse and Alcoholism; Einstein Stiftung Berlin; Laura and John Arnold Foundation","keywords":"Transparency (behavior); Credibility; Psychology; Confidence interval; Medicine; Clinical psychology; Social psychology; Political science; Law","score_opus":0.8303977243332353,"score_gpt":0.6710300330829925,"score_spread":0.15936769125024286,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3134007699","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9479204,0.0042287945,0.0005178048,0.010948379,0.0004474581,0.00059193675,0.0000051767024,0.0000056778904,0.03533441],"genre_scores_gemma":[0.99237984,0.00043336552,0.0056874314,0.00014189516,0.000028439239,0.000032788543,1.4626129e-7,0.0000034585216,0.00129265],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.96469176,0.018396266,0.0043328553,0.00643433,0.005604085,0.0005406974],"domain_scores_gemma":[0.97022873,0.012215503,0.003299846,0.011454862,0.002620866,0.00018018717],"candidate_categories":["metaresearch","sts","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.37310612,0.00019305604,0.0011843108,0.00043033017,0.00041812626,0.00043455858,0.0037427142,0.00010118701,0.004156955],"category_scores_gemma":[0.39923227,0.00008170154,0.0003017064,0.007155955,0.00319495,0.00040163263,0.00024000733,0.0007256275,0.00030588463],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002685632,0.0054618744,0.63569605,0.00033003074,0.00013092114,0.00019108596,0.06401878,0.00069033285,0.03464501,0.088226944,0.011906525,0.15843387],"study_design_scores_gemma":[0.00019573171,0.00029817765,0.91227764,0.000058045116,0.000021479544,0.00009668678,0.0058525642,0.002069968,0.00007956713,0.078402504,0.00052278687,0.00012486528],"about_ca_topic_score_codex":0.000015431038,"about_ca_topic_score_gemma":0.000013073965,"teacher_disagreement_score":0.27658156,"about_ca_system_score_codex":0.000042357668,"about_ca_system_score_gemma":0.00009722513,"threshold_uncertainty_score":0.9995178},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3137057920","doi":"10.3390/pathogens10030356","title":"Publication Trends in Neglected Tropical Diseases of Latin America and the Caribbean: A Bibliometric Analysis","year":2021,"lang":"en","type":"article","venue":"Pathogens","topic":"Parasites and Host Interactions","field":"Immunology and Microbiology","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Brock University","funders":"","keywords":"Neglected tropical diseases; Latin Americans; Tropical disease; Dengue fever; Malaria; Global health; Geography; Millennium Development Goals; Tropical medicine; Socioeconomics; Demography; Developing country; Economic growth; Political science; Medicine; Disease; Health care; Virology; Economics","score_opus":0.015680360263248854,"score_gpt":0.2825741433193035,"score_spread":0.2668937830560546,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3137057920","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.989837,0.007149655,0.00036610375,0.0013318624,0.000060138682,0.000047664587,0.00029513912,0.000015857217,0.0008966094],"genre_scores_gemma":[0.9983124,0.00023997515,0.0000627906,0.00008341867,0.0000070521887,0.000017004691,0.00055305375,0.0000030130077,0.00072125235],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9992944,0.0002143025,0.00020039211,0.00015046814,0.00002398097,0.00011644792],"domain_scores_gemma":[0.9993292,0.0002796399,0.00008563856,0.0001806279,0.000110839785,0.0000140626535],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.000016396314,0.00005956131,0.00022660487,0.010042439,0.000055851353,0.000016458165,0.00005908292,0.000051490086,0.0006734166],"category_scores_gemma":[0.0003436936,0.000042296695,0.00011910554,0.054788314,0.00012384268,0.000036224523,0.00003667271,0.00009799059,0.0000147368],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015541533,0.0003334337,0.8166532,0.0000054066422,0.0009477819,0.0000056928607,0.0005486373,0.00002355304,0.12459962,0.0015310118,0.0012156114,0.053980645],"study_design_scores_gemma":[0.0005972844,0.000019681427,0.9932413,0.0000019280058,0.00036716263,0.0000064432884,0.00010925629,0.00015929632,0.00060490373,0.000021933129,0.00482676,0.00004402846],"about_ca_topic_score_codex":0.0001702147,"about_ca_topic_score_gemma":0.00019174343,"teacher_disagreement_score":0.17658813,"about_ca_system_score_codex":0.000011347319,"about_ca_system_score_gemma":0.000019423553,"threshold_uncertainty_score":0.9653026},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3157841043","doi":"10.21203/rs.3.rs-443411/v1","title":"A Bibliometric Analysis of the Cannabis and Cannabinoid Research Literature","year":2021,"lang":"en","type":"preprint","venue":"Research Square","topic":"Cannabis and Cannabinoid Research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McMaster University","funders":"University of Toronto; Universidad Complutense de Madrid; University of Washington; King's College London; Institut National de la Santé et de la Recherche Médicale; Virginia Commonwealth University; McMaster University; National Institutes of Health; U.S. Department of Health and Human Services","keywords":"Cannabis; Cannabinoid; Psychology; Medicine; Psychiatry","score_opus":0.06259364514902867,"score_gpt":0.42987273507312274,"score_spread":0.3672790899240941,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3157841043","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9085815,0.051969294,0.000004935695,0.032777052,0.00022287935,0.0028705755,0.00065336365,0.000039146104,0.0028812748],"genre_scores_gemma":[0.9482285,0.008328618,0.000071402836,0.00009648422,0.0003711934,0.0008413246,0.00045954372,0.00010445501,0.041498493],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.983194,0.0032754166,0.0009221963,0.0018019788,0.008807082,0.0019993354],"domain_scores_gemma":[0.97914034,0.00027097802,0.00020032897,0.0036715388,0.01569834,0.0010184915],"candidate_categories":["metaepi_narrow","bibliometrics","scholarly_communication","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.011434418,0.00051358563,0.0016691589,0.20720981,0.0007580361,0.0010401602,0.0013040333,0.0011616034,0.00064552785],"category_scores_gemma":[0.008009317,0.0003555447,0.0012727043,0.48438472,0.001992079,0.0001292475,0.007083691,0.0096108755,0.000006378806],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010299385,0.0021218683,0.30431217,0.027328879,0.008450774,0.0014823168,0.010583977,0.00028642282,0.010078953,0.0010138389,0.5935963,0.039714616],"study_design_scores_gemma":[0.0007892633,0.00052921596,0.9427935,0.0031746249,0.00078321574,0.00006118824,0.0027584482,0.0011109936,0.0028276267,0.00023214285,0.04456663,0.00037317045],"about_ca_topic_score_codex":0.009404716,"about_ca_topic_score_gemma":0.0020329524,"teacher_disagreement_score":0.6384813,"about_ca_system_score_codex":0.0009994531,"about_ca_system_score_gemma":0.006531687,"threshold_uncertainty_score":0.99999684},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3158567748","doi":"10.1016/j.clae.2021.101447","title":"Bibliometric analysis of the literature relating to scleral contact lenses","year":2021,"lang":"en","type":"article","venue":"Contact Lens and Anterior Eye","topic":"Intraocular Surgery and Lenses","field":"Medicine","cited_by":17,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Allergan","keywords":"Keratoconus; Ophthalmology; German; Optometry; Scleral lens; Contact lens; Medicine; Optics; Art; Cornea; Physics; History; Archaeology","score_opus":0.017028415620419023,"score_gpt":0.2759525943848604,"score_spread":0.2589241787644414,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3158567748","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9925191,0.0050461753,0.000022750188,0.00093074434,0.00019062991,0.000120353245,0.000044001874,0.000016945718,0.0011093472],"genre_scores_gemma":[0.9962498,0.0005871715,0.000083291096,0.0016162354,0.00007708235,0.0000031137386,0.000015234852,0.00001202511,0.0013560684],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.998928,0.00008345782,0.00032109476,0.00025439987,0.00019995315,0.00021307346],"domain_scores_gemma":[0.9990879,0.00013286695,0.00010304009,0.0003249176,0.0002620344,0.00008921147],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00018764772,0.00014527002,0.0005620225,0.008236982,0.000116160736,0.0000742689,0.000057493748,0.00008994375,0.00018336895],"category_scores_gemma":[0.0004129548,0.00009554997,0.0003732529,0.04202614,0.000025863004,0.00010966491,0.00006566563,0.00019880186,0.0000029181053],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001421475,0.00007868325,0.61036575,0.00012734538,0.0009329206,0.0002581042,0.0011558359,0.00000131489,0.3714586,0.00023349687,0.00014551297,0.015100322],"study_design_scores_gemma":[0.00042712496,0.00018113268,0.97453415,0.0009691524,0.0007241622,0.000039962044,0.00021434054,0.000051228177,0.017815616,0.0000023694852,0.0049241646,0.00011656996],"about_ca_topic_score_codex":0.000038769667,"about_ca_topic_score_gemma":0.000019500478,"teacher_disagreement_score":0.36416847,"about_ca_system_score_codex":0.000017102246,"about_ca_system_score_gemma":0.000064522006,"threshold_uncertainty_score":0.97833604},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3159170457","doi":"10.3389/fphar.2021.646626","title":"Bibliometric Analysis of Global Research Trends on Ultrasound Microbubble: A Quickly Developing Field","year":2021,"lang":"en","type":"article","venue":"Frontiers in Pharmacology","topic":"Ultrasound and Hyperthermia Applications","field":"Engineering","cited_by":90,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Tianjin Municipal Bureau of Public Health; Natural Science Foundation of Tianjin City","keywords":"Microbubbles; Citation; Web of science; Scopus; Science Citation Index; Bibliometrics; Ultrasound; Medicine; Contrast-enhanced ultrasound; Index (typography); Medical physics; Library science; Computer science; MEDLINE; Political science; Radiology; Pathology; World Wide Web; Meta-analysis","score_opus":0.030184642845922233,"score_gpt":0.35204995057461985,"score_spread":0.3218653077286976,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3159170457","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.971692,0.002922677,0.013347241,0.00046823273,0.00097591005,0.00008730296,0.0000743218,0.00006753938,0.010364747],"genre_scores_gemma":[0.9920067,0.0012042954,0.006192102,0.00025481544,0.000061133214,0.000044864268,0.000042419928,0.000013020985,0.00018065415],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9988654,0.000103857616,0.00027512148,0.0002397687,0.00015219254,0.00036363755],"domain_scores_gemma":[0.9992057,0.00041268344,0.000026403612,0.00017755521,0.00011621582,0.00006140934],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0002604248,0.00011091237,0.00032609876,0.0278868,0.00005438256,0.0000187593,0.00023032444,0.0001302725,0.00036609994],"category_scores_gemma":[0.000071733564,0.000123523,0.00009059884,0.2014352,0.00006122238,0.000047232123,0.00002837167,0.00027479458,0.000007962923],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008907481,0.00040140355,0.31639192,0.00011103673,0.0038688465,0.00007347583,0.00046205908,0.022787578,0.14519747,0.001739842,0.36520752,0.1436698],"study_design_scores_gemma":[0.0026340531,0.00018260891,0.6421359,0.000046412093,0.0013035721,0.00003717294,0.00088535325,0.013546245,0.15941933,0.0019867846,0.1769356,0.00088695396],"about_ca_topic_score_codex":0.00003586786,"about_ca_topic_score_gemma":0.000103455255,"teacher_disagreement_score":0.325744,"about_ca_system_score_codex":0.00027552314,"about_ca_system_score_gemma":0.00008914519,"threshold_uncertainty_score":0.9831313},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3159284827","doi":"10.13140/rg.2.2.21014.37440","title":"Análise bibliométrica da produção científica brasileira acerca da temática Blockchain","year":2020,"lang":"pt","type":"article","venue":"","topic":"Healthcare during COVID-19 Pandemic","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Humanities; Philosophy; Physics","score_opus":0.0992894657647935,"score_gpt":0.333574694786852,"score_spread":0.23428522902205848,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3159284827","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.45281813,0.0033888458,0.20503274,0.31481528,0.0051771468,0.0056377477,0.00022072795,0.0050397934,0.007869603],"genre_scores_gemma":[0.9609358,0.00010573662,0.0062237363,0.028609514,0.0011490022,0.00007758064,0.000011402516,0.00012528586,0.0027619828],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98920554,0.000957827,0.0017512827,0.003481281,0.0021541568,0.0024498918],"domain_scores_gemma":[0.99224985,0.00083639525,0.0005322348,0.0029314929,0.0006030037,0.0028470226],"candidate_categories":["metaepi_narrow","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0017471586,0.0010376868,0.0011830289,0.003169526,0.0007693372,0.0016199998,0.004916551,0.00062633626,0.0010265865],"category_scores_gemma":[0.0025017452,0.001016049,0.00040458314,0.027698327,0.0003312437,0.0006735183,0.0028664076,0.0016197823,0.0035459972],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00063209224,0.0027105971,0.08720865,0.006432132,0.0006737854,0.0026898717,0.0598268,0.0021559964,0.0132880695,0.026804432,0.63694566,0.16063191],"study_design_scores_gemma":[0.004810186,0.002910337,0.022640796,0.000788154,0.0002100193,0.00025947473,0.0013503351,0.6954666,0.0052372725,0.00027331372,0.26151645,0.0045370944],"about_ca_topic_score_codex":0.0003570439,"about_ca_topic_score_gemma":0.000046099914,"teacher_disagreement_score":0.69331056,"about_ca_system_score_codex":0.00045300703,"about_ca_system_score_gemma":0.002570021,"threshold_uncertainty_score":0.99988663},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3162335385","doi":"10.22521/edupij.2021.102.1","title":"Exploring the emerging COVID-19 research trends and current status in the field of education: a bibliometric analysis and knowledge mapping","year":2021,"lang":"en","type":"article","venue":"Educational Process International Journal","topic":"COVID-19 and Mental Health","field":"Psychology","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Coronavirus disease 2019 (COVID-19); Pandemic; Thematic analysis; Bibliometrics; Field (mathematics); Web of science; Data collection; Thematic map; Higher education; China; Library science; Medical education; Data science; Political science; MEDLINE; Sociology; Geography; Social science; Qualitative research; Medicine; Computer science; Cartography","score_opus":0.37379274155039927,"score_gpt":0.6112337143634851,"score_spread":0.2374409728130858,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3162335385","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9078737,0.020239215,0.00042233,0.065693915,0.0023757082,0.00009762982,0.000013030657,0.0000038631333,0.0032806427],"genre_scores_gemma":[0.9942127,0.0032217337,0.000068220805,0.0011066947,0.00079759955,0.000117027914,0.000026578162,0.000005690597,0.00044374383],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9980691,0.0003979248,0.00040237885,0.00024238149,0.0006507952,0.00023743193],"domain_scores_gemma":[0.99622375,0.0025893343,0.0001594899,0.00014090886,0.0007212557,0.00016524455],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0020208447,0.00008863167,0.00012576034,0.02171031,0.00031450982,0.00015736338,0.00035513213,0.000025300158,0.00080865616],"category_scores_gemma":[0.0014127343,0.00006352466,0.000058354373,0.034568727,0.00009259583,0.00023524917,0.00008920007,0.0005356682,0.0000032601267],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000058977705,0.0007944898,0.34440067,0.000107493805,0.00021827489,0.0000050605736,0.0478017,0.000017983042,0.0000069815064,0.0091493875,0.004783824,0.5926551],"study_design_scores_gemma":[0.0003293905,0.000028974351,0.9017471,0.00006534532,0.00003357864,0.00014012966,0.029843045,0.00004605001,0.000012163933,0.0042743934,0.063408755,0.000071082344],"about_ca_topic_score_codex":0.0005267418,"about_ca_topic_score_gemma":0.00035477278,"teacher_disagreement_score":0.5925841,"about_ca_system_score_codex":0.0003149098,"about_ca_system_score_gemma":0.0036637841,"threshold_uncertainty_score":0.9893778},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3163219630","doi":"10.31234/osf.io/rqwea","title":"Fake collaborations: Interdisciplinary science can undermine research integrity","year":2019,"lang":"en","type":"preprint","venue":"","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Research integrity; Scientific integrity; Engineering ethics; Political science; Public relations; Academic integrity; Business; Engineering","score_opus":0.8797457237471681,"score_gpt":0.7091312008736873,"score_spread":0.17061452287348078,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3163219630","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5924136,0.0014430579,0.010114394,0.03560391,0.00903718,0.0035167944,0.0007239871,0.00018689146,0.34696016],"genre_scores_gemma":[0.95819134,0.00022948947,0.0029144003,0.00015081224,0.00024997705,0.00008780672,0.000032036703,0.000028038989,0.03811613],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9518704,0.0011424805,0.001663529,0.0035048074,0.039926723,0.0018920884],"domain_scores_gemma":[0.93085194,0.012909888,0.00053932745,0.0055933315,0.04879724,0.0013082938],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","sts","open_science","insufficient_payload"],"category_scores_codex":[0.15185748,0.00043225317,0.0008713167,0.2106572,0.0016951314,0.019752564,0.017393308,0.00067992794,0.001990702],"category_scores_gemma":[0.12237361,0.00029234862,0.00028218515,0.44048792,0.0032841153,0.000967041,0.060737632,0.0050889254,0.0025258858],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013743801,0.0014431761,0.06797353,0.00016827257,0.00009232952,0.00012833095,0.0030509392,0.0028066947,0.0027244703,0.06821386,0.69554484,0.15771611],"study_design_scores_gemma":[0.0008620145,0.0007941832,0.065116145,0.00023986687,0.000016068761,0.000027021495,0.021621171,0.12780222,0.0038731678,0.7424931,0.0356887,0.0014663369],"about_ca_topic_score_codex":0.00077178737,"about_ca_topic_score_gemma":0.0009594633,"teacher_disagreement_score":0.6742792,"about_ca_system_score_codex":0.0019858584,"about_ca_system_score_gemma":0.014833283,"threshold_uncertainty_score":0.99995285},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"methods","study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["research_integrity","bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W3164193789","doi":"10.1016/j.egyr.2021.05.020","title":"Corrigendum to “Carbon capture technologies for climate change mitigation: A bibliometric analysis of the scientific discourse during 1998–2018” [Energy Reports 6 (2020) 1200-1212]","year":2021,"lang":"en","type":"erratum","venue":"Energy Reports","topic":"Carbon Dioxide Capture Technologies","field":"Engineering","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Climate change; Climate science; Carbon capture and storage (timeline); Energy (signal processing); Carbon fibers; Environmental science; Environmental economics; Data science; Political science; Computer science; Statistics; Economics; Geology; Mathematics","score_opus":0.012923554899626636,"score_gpt":0.22967288767242833,"score_spread":0.2167493327728017,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3164193789","genre_codex":"editorial","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2820702,0.27789196,0.0011566576,0.0026378327,0.3953624,0.0037708003,0.0015216278,0.015557741,0.02003079],"genre_scores_gemma":[0.9551708,0.0036673,0.00044508147,0.000049652084,0.00092729216,0.0018691955,0.0016010961,0.00033841695,0.035931177],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9935414,0.000047594025,0.0019608664,0.0018466893,0.0014105496,0.0011928657],"domain_scores_gemma":[0.9930854,0.0000602785,0.0016428943,0.004569649,0.0004846479,0.00015715811],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0006337059,0.0010215645,0.0019014267,0.04462631,0.00018006252,0.0003137178,0.00087036047,0.0017990633,0.00001892447],"category_scores_gemma":[0.0010239349,0.0008724898,0.0012859951,0.13404728,0.00044963023,0.00019922176,0.0010141367,0.00084141333,3.6212688e-7],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004699338,0.00028797955,0.0072280522,0.0032906148,0.008414375,0.014627251,0.00083740166,0.023135366,0.013821285,0.0009415469,0.9149452,0.012423955],"study_design_scores_gemma":[0.00032565408,0.00010887194,0.010206056,0.0020836731,0.006528385,0.0026649544,0.0016422375,0.0057721534,0.062480908,0.0012348385,0.9033919,0.003560342],"about_ca_topic_score_codex":0.00049479416,"about_ca_topic_score_gemma":0.0033116254,"teacher_disagreement_score":0.6731006,"about_ca_system_score_codex":0.00054828014,"about_ca_system_score_gemma":0.00024266631,"threshold_uncertainty_score":0.9994968},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3167921806","doi":"10.3233/efi-211513","title":"A bibliometric analysis of race-related research in LIS","year":2021,"lang":"en","type":"article","venue":"Education for Information","topic":"Library Science and Administration","field":"Social Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University; McGill University; Dalhousie University","funders":"","keywords":"Scholarship; Race (biology); Sociology; Field (mathematics); Inequality; Racism; Library science; Social science; Political science; Gender studies; Computer science; Law","score_opus":0.09678664269493042,"score_gpt":0.44837305588687926,"score_spread":0.35158641319194883,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3167921806","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.89806503,0.00009622524,0.00020341348,0.0043871845,0.0004892276,0.00031835394,0.000013137063,0.000016619933,0.09641081],"genre_scores_gemma":[0.99769336,0.00007482695,0.00046010912,0.00007760785,0.000029223454,0.000032145355,0.000316366,0.0000010208969,0.0013153323],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9989736,0.000107786196,0.00031160587,0.00006800934,0.00040460774,0.00013437835],"domain_scores_gemma":[0.99886954,0.00021380454,0.00011816245,0.00010131021,0.0006497501,0.00004745562],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0018606896,0.000027930733,0.00007876867,0.0468445,0.00017869147,0.00015032223,0.000114636205,0.000058963597,0.00022144004],"category_scores_gemma":[0.0010089473,0.000029724264,0.000051252046,0.25575846,0.00006694169,0.003485079,0.000010540795,0.00005420703,0.000020240152],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000018967106,0.00035418847,0.0830828,0.000066557644,0.00007548076,1.5038167e-7,0.090701185,0.0004912045,0.00014934491,0.45270008,0.01125634,0.3611037],"study_design_scores_gemma":[0.00030586869,0.000062630104,0.5628118,0.000036938927,0.00005989823,3.6160287e-7,0.08327727,0.002665049,0.0016906045,0.010748987,0.33819517,0.00014542785],"about_ca_topic_score_codex":0.0005256144,"about_ca_topic_score_gemma":0.00035596036,"teacher_disagreement_score":0.479729,"about_ca_system_score_codex":0.00007259672,"about_ca_system_score_gemma":0.0018386707,"threshold_uncertainty_score":0.9639587},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3171172264","doi":"10.3390/polym13121957","title":"Scientometric Analysis and Systematic Review of Multi-Material Additive Manufacturing of Polymers","year":2021,"lang":"en","type":"review","venue":"Polymers","topic":"Additive Manufacturing and 3D Printing Technologies","field":"Engineering","cited_by":64,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Scopus; Systematic review; Computer science; Microscale chemistry; Nanotechnology; Materials science; Management science; Data science; Engineering; Mathematics; MEDLINE; Political science","score_opus":0.02572536010457595,"score_gpt":0.28821310498069214,"score_spread":0.2624877448761162,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3171172264","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000056139324,0.99768585,0.00037130856,0.0000016782565,0.00021986438,0.0005797304,0.000690732,0.0001924825,0.00020224175],"genre_scores_gemma":[0.0010317017,0.9981333,0.0004017982,0.0000043193086,0.000017395734,0.000081491264,0.00019464384,0.00005536892,0.00007999926],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.99756,0.00016124638,0.0011961675,0.00040819574,0.00033681636,0.0003375343],"domain_scores_gemma":[0.9981073,0.00040612483,0.00072266464,0.00063629,0.00005082888,0.00007676092],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.00038852444,0.00050751923,0.004415213,0.0030415705,0.000040940442,0.000026977425,0.00044335457,0.00024114412,0.00029979218],"category_scores_gemma":[0.00043114115,0.0004206847,0.00087534246,0.0028996943,0.00022630868,0.00005989675,0.00025074533,0.0002513865,0.0000039113866],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[2.5915094e-7,0.0000096275635,3.2136813e-7,0.74839264,0.0024289791,0.0000047976505,0.000018073588,0.0000018347934,0.0000018510981,0.0000067903893,0.000013760007,0.24912104],"study_design_scores_gemma":[0.00011671412,0.000025827776,0.000010821996,0.9396553,0.023631366,0.000022333994,0.00012900609,0.00003372183,0.02843518,0.0000012683324,0.0071957437,0.00074270234],"about_ca_topic_score_codex":0.00004906073,"about_ca_topic_score_gemma":0.0000017683873,"teacher_disagreement_score":0.24837834,"about_ca_system_score_codex":0.0000905303,"about_ca_system_score_gemma":0.000049103473,"threshold_uncertainty_score":0.9998245},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3172127826","doi":"10.26642/pbo-2021-1(48)-70-76","title":"Bibliometric analysis of research on responsible investment","year":2021,"lang":"en","type":"article","venue":"Problems of Theory and Methodology of Accounting Control and Analysis","topic":"Corporate Social Responsibility Reporting","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Popularity; Investment (military); Bibliometrics; Context (archaeology); Subject (documents); Political science; Analytics; Web of science; Regional science; Business; Data science; Library science; Politics; Sociology; Geography; Computer science; Law","score_opus":0.15796329305301177,"score_gpt":0.4012948525614604,"score_spread":0.24333155950844865,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3172127826","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99099123,0.0024296711,0.00430044,0.0003346821,0.000040686176,0.0001694841,0.00001109868,0.000017218263,0.001705506],"genre_scores_gemma":[0.9966204,0.00015273957,0.0027080853,0.00026101657,0.000054955653,0.000010010227,0.000015519092,0.000011132229,0.00016612708],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99598986,0.0015251896,0.0010954487,0.00053464615,0.00053229614,0.0003225378],"domain_scores_gemma":[0.9859829,0.010270331,0.0014718305,0.00044864157,0.0017994905,0.000026816355],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.040607803,0.00017471859,0.0014576089,0.065312326,0.00018584848,0.00006972998,0.00019318965,0.00015344929,0.000080371385],"category_scores_gemma":[0.021878954,0.0001550181,0.00046413712,0.13543983,0.0005964609,0.0002901963,0.0002788082,0.0002308519,6.894863e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001621327,0.0003089852,0.66738456,0.0009694434,0.0140771,0.00001275709,0.0003074172,0.0022264593,0.086977474,0.19952182,0.000011778327,0.026580887],"study_design_scores_gemma":[0.0012229535,0.000103337596,0.70393354,0.00014168982,0.020713853,0.0000017044437,0.0020757748,0.0066107535,0.0047545447,0.25983837,0.00030656898,0.00029694365],"about_ca_topic_score_codex":0.000605167,"about_ca_topic_score_gemma":0.00021327147,"teacher_disagreement_score":0.08222293,"about_ca_system_score_codex":0.000013943215,"about_ca_system_score_gemma":0.00010581274,"threshold_uncertainty_score":0.98789614},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3173150748","doi":"10.1111/ijfs.15215","title":"Unravelling scientific research towards the green extraction of phenolic compounds from leaves: a bibliometric analysis","year":2021,"lang":"en","type":"article","venue":"International Journal of Food Science & Technology","topic":"Phytochemicals and Antioxidant Activities","field":"Medicine","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Nutraceutical; Web of science; Bibliometrics; Scientific literature; Computer science; Political science; Library science; Chemistry; MEDLINE; Biology; Food science; Botany","score_opus":0.1000462371678488,"score_gpt":0.4001401096929394,"score_spread":0.30009387252509057,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3173150748","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9883157,0.0012980015,0.0027605006,0.00663982,0.00039875018,0.000043452543,0.000014587012,0.000010601731,0.00051859574],"genre_scores_gemma":[0.9987418,0.00018450584,0.0007866939,0.000035707315,0.00015424643,0.0000014990737,0.0000025031018,0.0000042459455,0.0000887938],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99712175,0.000042544798,0.00045339775,0.00023560261,0.0019220337,0.00022464129],"domain_scores_gemma":[0.9949777,0.000218276,0.00036133174,0.00026989102,0.004109419,0.00006337919],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0017151791,0.00007444159,0.0002807184,0.06117354,0.00018503913,0.00014998464,0.0009128242,0.00009972496,0.000050218918],"category_scores_gemma":[0.0011039731,0.00005206096,0.00017907527,0.11043991,0.001898006,0.00032964328,0.00020205024,0.0006746791,0.000002408229],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000043255262,0.00019640477,0.008313863,0.000004095931,0.00042435902,0.00004536109,0.00024575152,0.000031241983,0.9756774,0.0009011647,0.000038725368,0.014078428],"study_design_scores_gemma":[0.00034515135,0.00022499703,0.029329183,0.00010248323,0.00013077928,0.00037743212,0.0016280676,0.0005885324,0.95586604,0.010447235,0.00090744643,0.000052664567],"about_ca_topic_score_codex":0.000053181822,"about_ca_topic_score_gemma":0.000022125028,"teacher_disagreement_score":0.049266372,"about_ca_system_score_codex":0.00017980895,"about_ca_system_score_gemma":0.00084981177,"threshold_uncertainty_score":0.94946724},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3174204091","doi":"10.32552/2021.actamedica.542","title":"Global Scientific Outputs of Palliative Care Publications: A Holistic Approach with Bibliometric Analysis","year":2021,"lang":"en","type":"article","venue":"Acta Medica","topic":"Palliative Care and End-of-Life Issues","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Palliative care; Psychology; Medicine; Nursing","score_opus":0.16831768489709104,"score_gpt":0.4278691643489317,"score_spread":0.25955147945184065,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3174204091","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9279268,0.025219565,0.016258348,0.0014876677,0.00019189493,0.0003941219,0.00040448716,0.00007282523,0.028044285],"genre_scores_gemma":[0.9884334,0.00053252117,0.009571438,0.00008481026,0.000052264968,0.000025450145,0.0008378262,0.000008792724,0.0004534867],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9978946,0.00006405663,0.00036253314,0.00041792367,0.0010281751,0.00023272204],"domain_scores_gemma":[0.99684787,0.00020229645,0.00024461737,0.0006193456,0.0018543055,0.00023154885],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00023942189,0.00014225328,0.00058508024,0.012352424,0.000063067564,0.000033977445,0.00019957293,0.000092011884,0.00015973944],"category_scores_gemma":[0.0024343492,0.000106386266,0.00017038794,0.16330051,0.00039610264,0.0000916724,0.00006922618,0.00011157902,0.0000040253885],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005726829,0.00042297042,0.98228556,0.00051424064,0.0019787445,0.000024795541,0.0033563154,0.000008944098,0.0001996476,0.00073947373,0.0035752468,0.0068367887],"study_design_scores_gemma":[0.0015688938,0.00021083199,0.9533657,0.0004625018,0.0032926726,0.0000065672752,0.009670234,0.00029768213,0.0017254507,0.000052558175,0.029159192,0.00018773347],"about_ca_topic_score_codex":0.0001109515,"about_ca_topic_score_gemma":0.00007411337,"teacher_disagreement_score":0.15094809,"about_ca_system_score_codex":0.0001150489,"about_ca_system_score_gemma":0.0007939605,"threshold_uncertainty_score":0.9988417},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3176261361","doi":"10.1111/cid.13024","title":"Bibliometric analysis of the scientific production of literature on peri‐implant diseases in the Web of Science","year":2021,"lang":"en","type":"article","venue":"Clinical Implant Dentistry and Related Research","topic":"Dental Implant Techniques and Outcomes","field":"Dentistry","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Web of science; Citation database; Citation; Bibliometrics; Library science; Scopus; Typology; Science Citation Index; Work (physics); Medicine; Political science; MEDLINE; Computer science; Sociology; Engineering","score_opus":0.1127940631614157,"score_gpt":0.47405413258266493,"score_spread":0.3612600694212492,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3176261361","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99222195,0.0016368493,8.495038e-7,0.00018128661,0.00496281,0.00027125565,0.00050207553,0.000010363115,0.00021258746],"genre_scores_gemma":[0.99730265,0.00088085723,0.000012240459,0.000014112892,0.000027288374,0.0000063393836,0.000085745385,0.000006280426,0.0016644573],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99554735,0.00088819826,0.0010053014,0.0005860014,0.0016090891,0.0003640533],"domain_scores_gemma":[0.9964648,0.0015223844,0.00031038478,0.00092190155,0.00067759835,0.000102923375],"candidate_categories":["bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0055427994,0.00012620998,0.00048732778,0.014947742,0.00036830705,0.00022187574,0.00088164525,0.00021302295,0.00005052946],"category_scores_gemma":[0.0034148546,0.00006855044,0.00034468155,0.15143059,0.003285225,0.00016766741,0.00045492855,0.00091227144,0.000008110108],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00062809617,0.003958091,0.8354013,0.001080533,0.0012807943,0.02038119,0.00078922877,0.00005134313,0.09417233,0.0055029136,0.021344975,0.015409193],"study_design_scores_gemma":[0.00035205035,0.00012812558,0.96941614,0.0004330917,0.0002817487,0.012318013,0.000457943,0.00015919674,0.015628245,0.000331576,0.00040409737,0.00008979918],"about_ca_topic_score_codex":0.000025722366,"about_ca_topic_score_gemma":0.000057119905,"teacher_disagreement_score":0.13648285,"about_ca_system_score_codex":0.00002148597,"about_ca_system_score_gemma":0.00031457236,"threshold_uncertainty_score":0.99942726},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3180943902","doi":"10.1080/03057925.2021.1932422","title":"A decade review and bibliometric analysis of the journal <i>Compare</i>","year":2021,"lang":"en","type":"review","venue":"Compare A Journal of Comparative and International Education","topic":"Global Education and Multiculturalism","field":"Social Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"China Scholarship Council","keywords":"Bibliometrics; Data science; Library science; Regional science; Computer science; Geography","score_opus":0.2029282678084211,"score_gpt":0.5252226406632183,"score_spread":0.32229437285479723,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3180943902","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015405245,0.99309707,0.000016137217,0.0014673594,0.0020375326,0.00026133485,0.000022161694,0.000002299829,0.0015555987],"genre_scores_gemma":[0.009232457,0.9890356,0.00017450876,0.0005169398,0.00029521013,0.0000060568,0.000030260937,0.0000054843817,0.0007034627],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99694955,0.0009774696,0.0011767884,0.00014780687,0.00060965226,0.00013872409],"domain_scores_gemma":[0.992205,0.00044763778,0.0025184965,0.00014060696,0.0044743586,0.000213888],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0010682639,0.00021310625,0.0016166852,0.008393831,0.00025733584,0.00018893815,0.00063675176,0.00006351547,0.00027268167],"category_scores_gemma":[0.00088891893,0.00013573156,0.00055085495,0.024309717,0.00024705462,0.0004614202,0.0000817725,0.00044803115,0.0000017624542],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017098033,0.0010267239,0.0015427438,0.003848244,0.009489171,0.0000018408697,0.031703267,0.000018406894,0.000001574916,0.008492226,0.35025746,0.5936012],"study_design_scores_gemma":[0.000091088164,0.000014101482,0.0025746666,0.008958226,0.006981696,0.00014490768,0.00416058,0.0000033797094,4.7209846e-7,0.00004289956,0.97689766,0.00013033996],"about_ca_topic_score_codex":0.00013324709,"about_ca_topic_score_gemma":0.00011375487,"teacher_disagreement_score":0.6266402,"about_ca_system_score_codex":0.00023100522,"about_ca_system_score_gemma":0.002078652,"threshold_uncertainty_score":0.996429},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3181564238","doi":"10.29173/cais1219","title":"A Bibliometric Analysis of Race-Related Research in LIS","year":2021,"lang":"en","type":"article","venue":"Proceedings of the Annual Conference of CAIS / Actes du congrès annuel de l ACSI","topic":"Advanced Causal Inference Techniques","field":"Mathematics","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"McGill University; Dalhousie University; Western University","funders":"","keywords":"Diversity (politics); Equity (law); Race (biology); Work (physics); Sociology; Political science; Library science; Regional science; Social science; Computer science; Engineering; Gender studies; Anthropology","score_opus":0.1645688386437193,"score_gpt":0.4133993316957991,"score_spread":0.24883049305207983,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3181564238","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9885792,0.00032550574,0.00010852174,0.00071778224,0.000039965078,0.00036091232,0.00012459463,0.00006052267,0.009682989],"genre_scores_gemma":[0.9943481,0.00045923048,0.0045588254,0.000023812821,0.00001559235,0.00004426826,0.000005878184,0.000031190422,0.0005131124],"study_design_codex":"observational","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99655455,0.00011871579,0.001174144,0.00045737196,0.001123063,0.0005721352],"domain_scores_gemma":[0.9099807,0.0015881535,0.001135283,0.00048425177,0.086681604,0.00013000777],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0028719741,0.00027403154,0.0011508034,0.027995214,0.00007428173,0.0006718405,0.0016912253,0.00030250015,0.00011883456],"category_scores_gemma":[0.0872266,0.00022722737,0.00035296066,0.12703554,0.00065051107,0.0044684852,0.0010191043,0.00069587305,0.0000013285023],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027936304,0.001456023,0.5107773,0.0018695887,0.0017971419,0.000022388605,0.039871585,0.000028938928,0.16481142,0.25905415,0.0029795093,0.017052619],"study_design_scores_gemma":[0.00086827244,0.00045737086,0.21553113,0.0013042373,0.00082903844,0.000029057275,0.006341272,0.0011902994,0.42785397,0.3424077,0.002640921,0.0005467268],"about_ca_topic_score_codex":0.0002675599,"about_ca_topic_score_gemma":0.000068962516,"teacher_disagreement_score":0.29524612,"about_ca_system_score_codex":0.00012898713,"about_ca_system_score_gemma":0.00042394086,"threshold_uncertainty_score":0.9830216},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3184362473","doi":"","title":"Citation analysis for monitoring evidence-based research – a systematic review of meta-research studies","year":2021,"lang":"en","type":"review","venue":"University of Southern Denmark Research Portal (University of Southern Denmark)","topic":"Impact of AI and Big Data on Business and Society","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Community Based Research Centre","funders":"","keywords":"Citation; Systematic review; Meta-analysis; Citation analysis; Data science; Computer science; Library science; MEDLINE; Political science; Medicine","score_opus":0.7608620923625228,"score_gpt":0.5230539471964568,"score_spread":0.23780814516606597,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3184362473","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012293798,0.9897479,0.00070966076,0.00027813253,0.00008028741,0.0033351362,0.0041777473,0.000016354355,0.0004253939],"genre_scores_gemma":[0.0034274494,0.9842834,0.0010721218,0.0000054536285,0.000059830683,0.000007000012,0.00017617374,0.000050963845,0.010917623],"study_design_codex":"systematic_review","study_design_gemma":"qualitative","domain_scores_codex":[0.9730298,0.010730908,0.0018536052,0.0016585786,0.011444498,0.0012826261],"domain_scores_gemma":[0.9366617,0.029308058,0.002870518,0.0029688168,0.027662383,0.0005284943],"candidate_categories":["metaresearch","metaepi_narrow","sts","open_science","insufficient_payload"],"consensus_categories":["metaresearch","sts"],"category_scores_codex":[0.07726111,0.00064002106,0.007955948,0.0059875585,0.0017308821,0.00017810363,0.005453576,0.00072986254,0.004363088],"category_scores_gemma":[0.02308293,0.00054408144,0.005773672,0.017907297,0.0033604403,0.0006165475,0.0019324886,0.0016408389,0.00037833792],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031908866,0.000642543,0.00052073825,0.92656803,0.031600554,0.0003843372,0.013061105,0.000028042665,0.00001341739,0.00015865453,0.009183695,0.017519817],"study_design_scores_gemma":[0.0006428158,0.00035992332,0.00002123179,0.4169157,0.02425573,0.000006490597,0.53918105,0.00009156275,0.0000057053776,0.00026651795,0.017620232,0.0006330345],"about_ca_topic_score_codex":0.0022261005,"about_ca_topic_score_gemma":0.0005236565,"teacher_disagreement_score":0.52611995,"about_ca_system_score_codex":0.00039932688,"about_ca_system_score_gemma":0.0030710336,"threshold_uncertainty_score":0.9999274},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3185954846","doi":"10.2196/29282","title":"Global Burden of Skin Disease Representation in the Literature: Bibliometric Analysis","year":2021,"lang":"en","type":"article","venue":"JMIR Dermatology","topic":"Cutaneous Melanoma Detection and Management","field":"Medicine","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Medicine; Burden of disease; Disease; Dermatology; Bibliometrics; Psoriasis; Disease burden; Gerontology; Pathology; Library science","score_opus":0.013811066632382468,"score_gpt":0.3214766952604405,"score_spread":0.307665628628058,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3185954846","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99020475,0.0013012338,0.0002556857,0.005324605,0.00008803376,0.00024530254,0.000009039972,0.000017523811,0.0025538115],"genre_scores_gemma":[0.99890536,0.00013623816,0.00006214705,0.00049364794,0.000032275533,0.00003441742,0.00008441165,0.0000028507548,0.00024865483],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9991353,0.00015065054,0.00022505388,0.00017717486,0.00019847631,0.000113317554],"domain_scores_gemma":[0.9993912,0.00006232205,0.000064606684,0.0003370246,0.00009356274,0.000051264404],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.000059127346,0.000064136475,0.0002232284,0.009736491,0.000018701894,0.00002149307,0.000067683846,0.000050037885,0.00012059051],"category_scores_gemma":[0.00011958972,0.00004818809,0.00013814052,0.10039919,0.000028470442,0.000030154511,0.00003649569,0.00006786523,0.000010509235],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005376635,0.00075196795,0.9364874,0.0005165963,0.0010231587,0.018711196,0.002160803,0.00046308208,0.00044225538,0.0020848808,0.011680439,0.025140595],"study_design_scores_gemma":[0.0007637786,0.000023887003,0.98663616,0.000021712705,0.00045826385,0.0016730565,0.0006813033,0.0007610973,0.00040027456,0.00027501964,0.008244706,0.00006075016],"about_ca_topic_score_codex":0.000048584796,"about_ca_topic_score_gemma":0.000063256906,"teacher_disagreement_score":0.090662695,"about_ca_system_score_codex":0.000026630483,"about_ca_system_score_gemma":0.000033226166,"threshold_uncertainty_score":0.9187223},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3188723953","doi":"10.3389/fpsyt.2021.683474","title":"Bibliometrics Analysis of the Research Status and Trends of the Association Between Depression and Insulin From 2010 to 2020","year":2021,"lang":"en","type":"review","venue":"Frontiers in Psychiatry","topic":"Mental Health Research Topics","field":"Psychology","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Jilin University","keywords":"Insulin resistance; Depression (economics); Obesity; Bibliometrics; Association (psychology); Insulin; Medicine; Psychology; Library science; Computer science; Internal medicine","score_opus":0.11551003431842798,"score_gpt":0.48464005996210324,"score_spread":0.36913002564367525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3188723953","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010597767,0.9808012,0.000026124126,0.00046555395,0.005735263,0.0006452242,0.0011676796,0.0000038848784,0.00055726577],"genre_scores_gemma":[0.0006147894,0.9964037,0.0013329214,0.000037079997,0.00020152912,0.00006516106,0.00013257035,0.000029764618,0.0011824921],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99367845,0.0030403575,0.00097336253,0.0006140126,0.0011527985,0.0005409898],"domain_scores_gemma":[0.9969899,0.0009659385,0.00066891836,0.0010214169,0.00015511515,0.00019873171],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0026773338,0.00022145052,0.0015570817,0.009576084,0.00013240287,0.000036961643,0.000691474,0.00063981017,0.00012897475],"category_scores_gemma":[0.0007232493,0.0001427996,0.00031377695,0.05137525,0.00014135214,0.000041281906,0.0006633226,0.0014712375,0.0000016818724],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009151329,0.000052592775,0.4870007,0.000656102,0.0006950535,2.5962834e-7,0.00023054326,1.951246e-7,2.941982e-8,0.000013489093,0.052903444,0.45843843],"study_design_scores_gemma":[0.00030718854,0.000043790467,0.7243695,0.0017162247,0.00087133393,1.9176734e-7,0.000378802,0.000004520387,3.137101e-7,0.00015722816,0.27204522,0.00010565591],"about_ca_topic_score_codex":0.0016715336,"about_ca_topic_score_gemma":0.0006194745,"teacher_disagreement_score":0.45833278,"about_ca_system_score_codex":0.00034780512,"about_ca_system_score_gemma":0.000361566,"threshold_uncertainty_score":0.9687882},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W3188822421","doi":"","title":"Plotting the Literature on Precautionary Measures of COVID-19: A Scientometric Analysis of Web of Science","year":2021,"lang":"en","type":"article","venue":"Lincoln (University of Nebraska)","topic":"COVID-19 and Mental Health","field":"Psychology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Coronavirus disease 2019 (COVID-19); Scholarship; Web of science; Social distance; 2019-20 coronavirus outbreak; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Scientometrics; Isolation (microbiology); Library science; Periodical literature; Scopus; Pandemic; Social science; Political science; Sociology; World Wide Web; MEDLINE; Computer science; Medicine; Law; Virology; Outbreak; Biology","score_opus":0.08815230148730743,"score_gpt":0.379996700175142,"score_spread":0.2918443986878345,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3188822421","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9926263,0.0011074637,0.0003246456,0.0012790432,0.00018263931,0.00018752977,0.00022587784,0.000013412002,0.0040530646],"genre_scores_gemma":[0.9981776,0.00007665561,0.00036076384,0.00024869572,0.000012739571,3.4571738e-7,0.000016059172,0.000003733023,0.0011034404],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9982654,0.00021069555,0.0002304206,0.00035494743,0.0007358344,0.00020271032],"domain_scores_gemma":[0.9978651,0.0006058823,0.00041301173,0.00049144204,0.00046601976,0.00015855527],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012266613,0.0000888851,0.00035350144,0.0027701682,0.0002763688,0.000006257483,0.00055139227,0.00008517271,0.00038185695],"category_scores_gemma":[0.0004654694,0.00008516076,0.00023348375,0.019337095,0.0010418302,0.00011231849,0.00020251043,0.00014231092,0.0000026686726],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026802064,0.0065349983,0.5353399,0.0030638862,0.0034074886,0.00022183047,0.19352181,0.006931852,0.09925833,0.0899096,0.004210996,0.054919116],"study_design_scores_gemma":[0.0020803045,0.0006824697,0.9399065,0.0002733281,0.000871133,0.000009665836,0.040136218,0.0021342465,0.003021238,0.0003882339,0.010274388,0.00022231587],"about_ca_topic_score_codex":0.00096430857,"about_ca_topic_score_gemma":0.0005129719,"teacher_disagreement_score":0.4045666,"about_ca_system_score_codex":0.00022767289,"about_ca_system_score_gemma":0.0023132602,"threshold_uncertainty_score":0.92908335},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3195052279","doi":"","title":"Covid-19 and Emergency Medicine: A Scientometric Assessment of Global Publications","year":2021,"lang":"en","type":"article","venue":"Library philosophy and practice","topic":"COVID-19 and healthcare impacts","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Citation impact; Coronavirus disease 2019 (COVID-19); Web of science; Scientometrics; Medicine; Scopus; Bibliometrics; Library science; MEDLINE; Political science; Internal medicine; Computer science","score_opus":0.17324767683834677,"score_gpt":0.49306591190014953,"score_spread":0.31981823506180274,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3195052279","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013779034,0.010239646,0.00034942874,0.72945476,0.00025906516,0.00030375374,0.00007558302,0.00007973422,0.24545902],"genre_scores_gemma":[0.9614982,0.0076430193,0.0032307743,0.026427181,0.00038720327,0.000012074479,0.00008153303,0.000011705271,0.0007083037],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985035,0.0002253789,0.00037601293,0.00033630632,0.0003776388,0.00018116392],"domain_scores_gemma":[0.9971994,0.0011350375,0.00022404325,0.0003260119,0.00016238175,0.000953083],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00049717544,0.000121540994,0.00027075488,0.0002815733,0.00016525839,0.000026061127,0.00006975766,0.00008395072,0.00134981],"category_scores_gemma":[0.008640412,0.00010459774,0.00003797075,0.0033358233,0.00014891327,0.0013139949,0.00012593214,0.00018985399,0.000002316219],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002685367,0.00053213147,0.39475042,0.0024335468,0.00014846543,0.0000846994,0.00049923296,8.48941e-7,0.0002479686,0.5717526,0.0261873,0.0030942636],"study_design_scores_gemma":[0.0016235245,0.0006311727,0.4639557,0.00009982469,0.00028263018,0.00069573434,0.00039174032,0.000060541897,0.000016411721,0.03934098,0.49274954,0.00015219451],"about_ca_topic_score_codex":0.00009242681,"about_ca_topic_score_gemma":0.0000027051701,"teacher_disagreement_score":0.94771916,"about_ca_system_score_codex":0.000056547728,"about_ca_system_score_gemma":0.0022416613,"threshold_uncertainty_score":0.9997102},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3195712101","doi":"10.1108/jec-06-2021-0091","title":"A retrospective overview of <i>Journal of Enterprising Communities</i>: <i>People and Places in the Global Economy</i> from2007 to 2021 using abibliometric analysis","year":2021,"lang":"en","type":"article","venue":"Journal of Enterprising Communities People and Places in the Global Economy","topic":"Entrepreneurship Studies and Influences","field":"Business, Management and Accounting","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Dalhousie University","funders":"","keywords":"Originality; Indigenous; Entrepreneurship; Value (mathematics); Citation; Library science; Social science; Political science; Sociology; Law; Computer science","score_opus":0.029261791463801215,"score_gpt":0.2871278414481956,"score_spread":0.2578660499843944,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3195712101","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9783645,0.014089931,0.00021803656,0.0012419504,0.0002793909,0.00014320463,0.000036698486,0.000002897205,0.0056233755],"genre_scores_gemma":[0.9932658,0.0038717568,0.00025610122,0.0023756262,0.0002109148,0.0000021771386,0.0000071757117,0.0000083480845,0.0000020745692],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9973407,0.00049284543,0.0013517581,0.00014395974,0.00031734922,0.00035342356],"domain_scores_gemma":[0.99653393,0.00092109124,0.0015970155,0.00041925992,0.00049265247,0.000036075537],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0029624312,0.0003006736,0.0010985667,0.0014577572,0.00038239852,0.0010499108,0.0010771984,0.0000817902,0.000070629474],"category_scores_gemma":[0.00018497887,0.00022192915,0.00030173513,0.005242439,0.00022756754,0.0016591528,0.00065554195,0.00051406125,7.098073e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001880707,0.00018248167,0.9834717,0.00016366696,0.0004915774,0.00002514304,0.011656445,0.0019984888,0.0000014669938,0.000799019,0.0002695029,0.00075245555],"study_design_scores_gemma":[0.0023745012,0.00029480128,0.50854594,0.002008492,0.0013842782,0.00073706644,0.46637866,0.0020470794,0.000008903024,0.008716642,0.0069733616,0.0005302435],"about_ca_topic_score_codex":0.009697472,"about_ca_topic_score_gemma":0.03001704,"teacher_disagreement_score":0.47492573,"about_ca_system_score_codex":0.0002063515,"about_ca_system_score_gemma":0.00010056907,"threshold_uncertainty_score":0.99998707},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3196254136","doi":"","title":"PUTAJ- Humanities and Social Sciences: A Ten Years Bibliometric Analysis","year":2020,"lang":"en","type":"article","venue":"","topic":"Education and Islamic Studies","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Library science; Subject (documents); Bibliometrics; Social science; Sociology; Computer science","score_opus":0.16812602063890048,"score_gpt":0.3862911110645878,"score_spread":0.21816509042568732,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3196254136","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6184199,0.00037244768,0.0000718748,0.031901345,0.00008751781,0.00007932196,0.000003519064,0.00009955214,0.3489645],"genre_scores_gemma":[0.9919299,0.00019973499,0.00026366283,0.0045687794,0.00018913556,0.000003101099,5.5861307e-7,0.0000018049973,0.002843305],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.9993054,0.00005127029,0.00008361244,0.0001439235,0.00026997153,0.0001457912],"domain_scores_gemma":[0.9997464,0.00012489493,0.000031805874,0.000022487724,0.000051442516,0.000023002556],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00029915536,0.000040706844,0.0001107064,0.009365246,0.0007275042,0.00019216599,0.00012374166,0.00002388797,0.0006839205],"category_scores_gemma":[0.00023510327,0.000036365345,0.000046939353,0.08951824,0.0005879126,0.00013200815,0.00004209361,0.000033565146,0.000031554762],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000026137698,0.00004210901,0.33070534,0.000011591314,0.00034848822,0.0000013008039,0.3671447,0.0000021140584,0.000011093634,0.20325546,0.077445544,0.02102967],"study_design_scores_gemma":[0.00013716408,0.000044184424,0.49067777,0.0000011517893,0.00018107916,6.81159e-8,0.25172696,0.00009201897,0.000003914692,0.0014272797,0.25552872,0.0001796877],"about_ca_topic_score_codex":0.0014932599,"about_ca_topic_score_gemma":0.0006121957,"teacher_disagreement_score":0.37351,"about_ca_system_score_codex":0.000015045156,"about_ca_system_score_gemma":0.00007094194,"threshold_uncertainty_score":0.9298345},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3199112149","doi":"10.1080/08164622.2021.1973866","title":"Bibliometric analysis of the keratoconus literature","year":2021,"lang":"en","type":"article","venue":"Clinical and Experimental Optometry","topic":"Corneal surgery and disorders","field":"Medicine","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Keratoconus; Scopus; Bibliometrics; Index (typography); Subject (documents); Library science; Impact factor; Medicine; MEDLINE; Optometry; Ophthalmology; Political science; Computer science; Law","score_opus":0.03211646747737853,"score_gpt":0.41010618127422366,"score_spread":0.37798971379684515,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3199112149","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9796663,0.017492821,0.000005844753,0.00032485824,0.00023439169,0.000084587155,0.000012530982,0.000010835569,0.0021678181],"genre_scores_gemma":[0.9976525,0.0003816937,0.00009649595,0.0012741709,0.000050673374,0.000004934441,0.000022510558,0.00000520757,0.00051186845],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9989905,0.000070765294,0.00038614988,0.00023905796,0.00019731878,0.00011625293],"domain_scores_gemma":[0.99915624,0.00027327792,0.00006923818,0.0002743052,0.00008586698,0.00014110473],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00021340357,0.00008906257,0.0004338748,0.0073871515,0.000049460286,0.000026000616,0.000053497064,0.000111151014,0.00040604573],"category_scores_gemma":[0.00037707054,0.00005564749,0.00048909045,0.13271552,0.00018389159,0.00004615453,0.00012290255,0.000177944,0.0000037353916],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000119190605,0.00063316687,0.98743933,0.000023090408,0.000558616,0.000024929219,0.00006374905,8.809947e-7,0.0029776248,0.000053484462,0.00046345757,0.007642474],"study_design_scores_gemma":[0.0007942848,0.00009455773,0.97349143,0.000043960405,0.00038653731,0.0000170109,0.00025181338,0.0000824497,0.022798145,0.000010432024,0.001959586,0.00006981744],"about_ca_topic_score_codex":0.0000031526822,"about_ca_topic_score_gemma":2.917102e-7,"teacher_disagreement_score":0.12532836,"about_ca_system_score_codex":0.000006416513,"about_ca_system_score_gemma":0.000039495968,"threshold_uncertainty_score":0.885719},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3199933884","doi":"10.1155/2021/9274918","title":"Changes in the Structures and Directions of Rock Excavation Research from 1999 to 2020: A Bibliometric Study","year":2021,"lang":"en","type":"article","venue":"Advances in Civil Engineering","topic":"Tunneling and Rock Mechanics","field":"Engineering","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Natural Science Foundation of Hunan Province; Education Department of Hunan Province","keywords":"Excavation; Mining engineering; Geology; Archaeology; Forensic engineering; Engineering; Geotechnical engineering; History","score_opus":0.022443028341507534,"score_gpt":0.3017072764829584,"score_spread":0.2792642481414509,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3199933884","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9697521,0.023955371,0.0055174595,0.00004315942,0.00038501067,0.00019441434,0.000009925383,0.00005993518,0.000082668186],"genre_scores_gemma":[0.9940881,0.0051160385,0.00063689484,0.000004940892,0.000076498225,0.000052739175,0.00000312362,0.000016657481,0.000005016906],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99921626,0.00003962214,0.00016107706,0.00016462663,0.0002278423,0.00019055473],"domain_scores_gemma":[0.99934524,0.00039884218,0.000010453085,0.00017174987,0.000045733468,0.000028004672],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00033679113,0.0000898583,0.00014054656,0.004258938,0.000034021,0.000022534548,0.00011545576,0.00003836811,0.000007855382],"category_scores_gemma":[0.00033318755,0.00007997956,0.000010930592,0.01954207,0.0000027603658,0.0001041005,0.000055014956,0.0002519159,7.340651e-7],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000002965119,0.00003572266,0.0031625847,0.00007469836,0.000010817251,0.000015541365,0.0029960792,0.9720615,0.0073186653,0.000082404185,0.000025852121,0.014213125],"study_design_scores_gemma":[0.0018668908,0.0004192699,0.32150698,0.0010270518,0.000039709743,0.000029012743,0.019564444,0.5888412,0.043333523,0.003496665,0.018882023,0.0009932271],"about_ca_topic_score_codex":0.000095369694,"about_ca_topic_score_gemma":0.0070747836,"teacher_disagreement_score":0.38322034,"about_ca_system_score_codex":0.0000367346,"about_ca_system_score_gemma":0.000006567635,"threshold_uncertainty_score":0.93893164},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3200317969","doi":"10.3389/fped.2021.633525","title":"Global Trends and Future Prospects of Child Nutrition: A Bibliometric Analysis of Highly Cited Papers","year":2021,"lang":"en","type":"review","venue":"Frontiers in Pediatrics","topic":"Child Nutrition and Water Access","field":"Nursing","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Citation; Science Citation Index; Overweight; Bibliometrics; Childhood obesity; Obesity; Environmental health; Medical education; Pediatrics; Library science; Pathology; Computer science","score_opus":0.011524502379103593,"score_gpt":0.28678308968243477,"score_spread":0.2752585873033312,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3200317969","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00022646891,0.9946585,0.000027609081,0.0003260853,0.0020313151,0.00045222527,0.0014033054,0.00003534175,0.00083910004],"genre_scores_gemma":[0.00036392745,0.9955578,0.0013464922,0.00004809482,0.0010618819,0.000036730697,0.0015288304,0.0000350154,0.00002125172],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99692607,0.00022243362,0.0012304996,0.0006687002,0.00060259865,0.0003496722],"domain_scores_gemma":[0.9982725,0.00008901899,0.0008329123,0.00046888084,0.00019031555,0.0001464217],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00024168416,0.00046476288,0.0032030304,0.12144569,0.00006062568,0.00007785909,0.0003815904,0.0006308076,0.000040172952],"category_scores_gemma":[0.00008769365,0.0004404349,0.0010175147,0.4027699,0.00009796376,0.00013901398,0.00010466101,0.0003855968,2.4381737e-7],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006133474,0.0007597566,0.04773844,0.024236474,0.00096576463,0.00003191935,0.00006382513,0.0000025696536,1.4999932e-8,0.00005060345,0.03636336,0.8897259],"study_design_scores_gemma":[0.0014066872,0.00011230421,0.020902783,0.0019747233,0.013770165,0.00001901973,0.000053152904,0.000015413916,0.0000011499687,0.00003569709,0.9611988,0.0005100669],"about_ca_topic_score_codex":0.000023532451,"about_ca_topic_score_gemma":0.000013971163,"teacher_disagreement_score":0.9248355,"about_ca_system_score_codex":0.00025912965,"about_ca_system_score_gemma":0.00005539942,"threshold_uncertainty_score":0.99980474},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3201111863","doi":"10.3390/su131810378","title":"Chinese Universities’ Cross-Border Research Collaboration in the Social Sciences and Its Impact","year":2021,"lang":"en","type":"article","venue":"Sustainability","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Natural Science Foundation of China","keywords":"Scopus; Institution; Political science; Quality (philosophy); Bibliometrics; Library science; Social science; Sociology; MEDLINE; Law","score_opus":0.32639008186427493,"score_gpt":0.723295831982531,"score_spread":0.3969057501182561,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3201111863","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98076195,0.0007632902,0.000021140444,0.010638372,0.000077474324,0.0003203944,0.000021937936,0.000006501538,0.007388952],"genre_scores_gemma":[0.9977526,0.000028576082,0.000019716734,0.00005889678,0.00005241808,0.000008045959,0.0000019785978,0.0000027859621,0.002074978],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9891788,0.0021793512,0.00034721513,0.0007178237,0.006880741,0.00069611956],"domain_scores_gemma":[0.96114177,0.010742877,0.00010153635,0.00058973185,0.027261218,0.00016284383],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.07064912,0.000099803794,0.00019424934,0.013297918,0.0016475981,0.0051208213,0.0011839883,0.000101995174,0.00042804726],"category_scores_gemma":[0.10245134,0.000055878416,0.00007150397,0.27402657,0.0011241168,0.0011669269,0.0008488218,0.00040691855,0.000012328011],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006377305,0.0002040801,0.9652968,0.000019142284,0.0000041419767,0.000070545175,0.006709171,0.00010128896,0.000050137132,0.013989816,0.0016241072,0.011866988],"study_design_scores_gemma":[0.00025378985,0.00009550056,0.88070446,8.879977e-7,6.433966e-7,0.0000045804372,0.06206643,0.0013833769,0.00003740405,0.05239802,0.0029848681,0.00007001979],"about_ca_topic_score_codex":0.00039187307,"about_ca_topic_score_gemma":0.00036542205,"teacher_disagreement_score":0.26072866,"about_ca_system_score_codex":0.0011016894,"about_ca_system_score_gemma":0.005767235,"threshold_uncertainty_score":0.99986917},"labels":[{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"grok","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"opus","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W3201236133","doi":"10.1080/14767058.2021.1974387","title":"Trends, gaps, and future directions of research in cervical remodeling during pregnancy: a bibliometric analysis","year":2021,"lang":"en","type":"review","venue":"The Journal of Maternal-Fetal & Neonatal Medicine","topic":"Preterm Birth and Chorioamnionitis","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Columbia College","funders":"","keywords":"Scopus; Obstetrics and gynaecology; Medicine; Pregnancy; Obstetrics; Citation; Web of science; Cervix; Citation analysis; MEDLINE; Gynecology; Family medicine; Library science; Meta-analysis; Political science; Internal medicine; Biology","score_opus":0.059050537974263885,"score_gpt":0.38911398896406996,"score_spread":0.33006345098980605,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3201236133","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01796365,0.9800885,0.00001464435,0.00097486016,0.000449272,0.00024435774,0.000045443703,0.000010639654,0.00020863149],"genre_scores_gemma":[0.01325313,0.98463655,0.000113357746,0.0000123474865,0.0015711554,0.00001098879,0.00003670029,0.000046230965,0.0003195525],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99479204,0.0008320354,0.0018939241,0.00031836904,0.0016879688,0.00047568523],"domain_scores_gemma":[0.9968176,0.0007123614,0.00097457384,0.0005306087,0.0005932534,0.00037159966],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0021545538,0.00039605223,0.0031986681,0.06392052,0.00011787477,0.000027673072,0.0004586443,0.00030495392,0.00036432553],"category_scores_gemma":[0.00030904677,0.00022168884,0.00065327814,0.07535062,0.0003348948,0.00018245234,0.00023703606,0.0020990914,0.0000019657484],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019465118,0.0000579653,0.00062892976,0.011806413,0.0013105596,0.0012432643,0.00032954558,0.0000014816538,0.000010186922,0.00000820694,0.000033108437,0.9843757],"study_design_scores_gemma":[0.0055680857,0.0016295457,0.03589,0.35581037,0.023226181,0.05021534,0.0008934961,0.00011242564,0.000053828782,0.000115577044,0.5259027,0.00058246637],"about_ca_topic_score_codex":0.00032152393,"about_ca_topic_score_gemma":0.000050228366,"teacher_disagreement_score":0.9837932,"about_ca_system_score_codex":0.00018292402,"about_ca_system_score_gemma":0.00026800504,"threshold_uncertainty_score":0.9466891},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3202687862","doi":"10.1093/heapro/daab137","title":"A bibliometric analysis of <i>Health Promotion International</i> content regarding unions, unionization and collective agreements","year":2021,"lang":"en","type":"article","venue":"Health Promotion International","topic":"School Health and Nursing Education","field":"Health Professions","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Ontario Tech University; York University","funders":"","keywords":"Collective bargaining; Promotion (chess); Health promotion; Government (linguistics); Collective action; Public relations; Political science; Content analysis; Business; Sociology; Law; Health care; Social science; Politics","score_opus":0.1848110419441196,"score_gpt":0.4767290542662429,"score_spread":0.29191801232212333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3202687862","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.79316556,0.0019263589,0.04588203,0.13741262,0.013102742,0.0029568938,0.0005081784,0.00023037971,0.004815206],"genre_scores_gemma":[0.98232764,0.0029663723,0.0026821175,0.002412845,0.00039350177,0.00027966202,0.0033832588,0.00003619401,0.005518411],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9940195,0.0018564308,0.0018955689,0.0006504382,0.0010147806,0.00056326523],"domain_scores_gemma":[0.9939954,0.00036446739,0.0019390648,0.00033281907,0.0030161983,0.00035203347],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0034369212,0.00022704242,0.0006183243,0.026618103,0.0010404424,0.000049498187,0.00023085804,0.00019224302,0.0009235827],"category_scores_gemma":[0.0015997821,0.00024007093,0.00012689517,0.04279321,0.00006608674,0.0004957793,0.00012552112,0.0004982525,0.00002566141],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015305621,0.0021291578,0.89720196,0.00074075436,0.0011300967,0.0000022194595,0.007446021,0.000105528175,0.00028195314,0.006411317,0.023337765,0.061060175],"study_design_scores_gemma":[0.0021811319,0.00045455684,0.9792341,0.0011624885,0.00012298183,0.00001029409,0.004181037,0.004129246,0.000082620885,0.00095514226,0.0072739753,0.00021241045],"about_ca_topic_score_codex":0.00069625315,"about_ca_topic_score_gemma":0.0003382298,"teacher_disagreement_score":0.18916205,"about_ca_system_score_codex":0.002646644,"about_ca_system_score_gemma":0.003188679,"threshold_uncertainty_score":0.9999897},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3202986186","doi":"10.3390/su131910798","title":"An Analysis of Peer-Reviewed Publications on Open Educational Practices (OEP) from 2007 to 2020: A Bibliometric Mapping Analysis","year":2021,"lang":"en","type":"article","venue":"Sustainability","topic":"Open Education and E-Learning","field":"Computer Science","cited_by":28,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Scopus; Open science; Library science; Web of science; Data science; Computer science; Political science; MEDLINE; Mathematics","score_opus":0.05117221953669503,"score_gpt":0.4196123526328995,"score_spread":0.36844013309620444,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3202986186","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7154377,0.0003276126,0.05896468,0.21250619,0.00035693415,0.0010469112,0.000063829146,0.00009464167,0.0112015],"genre_scores_gemma":[0.96835285,0.000021772506,0.027162472,0.00076499284,0.00006140418,0.00013117344,0.00043281732,0.000006168778,0.0030663696],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99652874,0.0008669055,0.0005997116,0.0009976771,0.00070811535,0.0002988378],"domain_scores_gemma":[0.9889906,0.0011595171,0.00069366227,0.002022868,0.006796589,0.00033674002],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0032141332,0.00015331531,0.0005240884,0.016263554,0.00028423223,0.00144387,0.0020587316,0.00006074856,0.003223746],"category_scores_gemma":[0.02290597,0.0001564563,0.00026004505,0.2352617,0.000038373168,0.0018021526,0.00061530375,0.00020420567,0.000042417178],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014486126,0.0025442715,0.90200675,0.000055110766,0.0021034267,0.000002184416,0.0044253385,0.010587415,0.000049682203,0.048635747,0.0078956,0.021680016],"study_design_scores_gemma":[0.00008236073,0.000038741437,0.9366523,0.000004751178,0.0006308484,3.1282795e-7,0.0071846494,0.011933186,0.000029967616,0.0017576327,0.041511986,0.00017326293],"about_ca_topic_score_codex":0.003786846,"about_ca_topic_score_gemma":0.0006257602,"teacher_disagreement_score":0.2529151,"about_ca_system_score_codex":0.00056709594,"about_ca_system_score_gemma":0.002505209,"threshold_uncertainty_score":0.9995927},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W3203402935","doi":"10.1136/bmjopen-2021-051839","title":"Publication practices of sub-Saharan African Cochrane authors: a bibliometric study","year":2021,"lang":"en","type":"article","venue":"BMJ Open","topic":"Child and Adolescent Health","field":"Health Professions","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; St. Joseph’s Healthcare Hamilton; Impact; Programs for Assessment of Technology in Health Research Institute","funders":"Government of the United Kingdom","keywords":"Systematic review; Medicine; Cochrane Library; MEDLINE; Cochrane collaboration; Psychological intervention; Meta-analysis; Family medicine; Nursing; Internal medicine","score_opus":0.2856161133475696,"score_gpt":0.5762948029386101,"score_spread":0.29067868959104054,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3203402935","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.88123506,0.000949074,0.000050554372,0.043544907,0.0009219358,0.024236483,0.000080759404,0.00009912014,0.04888209],"genre_scores_gemma":[0.99280506,0.00024680846,0.0003563101,0.0013624689,0.00033105258,0.0003176779,0.00007833201,0.000024028857,0.004478269],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9959907,0.0017207705,0.00095668616,0.00038938312,0.0005651812,0.00037727828],"domain_scores_gemma":[0.9956428,0.0007789679,0.0016789113,0.00080399634,0.0008821321,0.000213231],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.005314742,0.0001226111,0.00044493162,0.0041075055,0.0004405165,0.000094851646,0.0005572109,0.000103315186,0.0010945589],"category_scores_gemma":[0.007012823,0.0001064222,0.000047476828,0.030856026,0.000023689898,0.0005546299,0.00063488417,0.00047093665,0.000268818],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006698647,0.0053379443,0.661356,0.0024504883,0.00014743743,0.000046114397,0.014011373,0.0000010812638,0.0016240855,0.0014344242,0.28937075,0.023550468],"study_design_scores_gemma":[0.002443077,0.00023004928,0.9385978,0.00084213255,0.000049298516,0.0000048778616,0.012614572,0.00003014888,0.00019142519,0.0001500577,0.04466728,0.00017927463],"about_ca_topic_score_codex":0.0013349557,"about_ca_topic_score_gemma":0.0010450197,"teacher_disagreement_score":0.27724183,"about_ca_system_score_codex":0.00007846208,"about_ca_system_score_gemma":0.0017291241,"threshold_uncertainty_score":0.99981856},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W3203891641","doi":"10.2196/32721","title":"Virtual Reality in Health Care: Bibliometric Analysis","year":2021,"lang":"en","type":"article","venue":"JMIR Serious Games","topic":"Virtual Reality Applications and Impacts","field":"Computer Science","cited_by":76,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"The Internet; Health care; Medical education; Public relations; Psychology; Political science; Medicine; Computer science; World Wide Web","score_opus":0.02382220927478649,"score_gpt":0.3358496085904542,"score_spread":0.3120273993156677,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3203891641","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6873318,0.0050789346,0.28102028,0.019755643,0.00027959858,0.0007343678,0.000069414564,0.00044238305,0.0052875816],"genre_scores_gemma":[0.99567515,0.00043127677,0.002469245,0.0010442607,0.000029329212,0.00005607043,0.000035970224,0.000006865563,0.00025185375],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99812096,0.00018397898,0.0004087994,0.000515863,0.00036942534,0.00040098233],"domain_scores_gemma":[0.99835956,0.00009285994,0.0001349917,0.0010406877,0.00014799768,0.00022391707],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00039979586,0.00013447109,0.00035054877,0.017209116,0.000098142984,0.00028209196,0.0005544354,0.00006726233,0.000023301163],"category_scores_gemma":[0.00011389698,0.00013359604,0.00013342741,0.1999624,0.000031928117,0.00034981378,0.00024736585,0.00014910381,0.00003461628],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000055646424,0.00027375354,0.004391067,0.0000431374,0.00010115984,0.00003046668,0.003429202,0.0010368219,0.00015385615,0.022881778,0.0037551683,0.963898],"study_design_scores_gemma":[0.00040200958,0.00019803015,0.9141215,0.000017321932,0.000015684327,0.000013309886,0.00056513184,0.0066470895,0.00048436725,0.00062579854,0.07662349,0.0002862894],"about_ca_topic_score_codex":0.0011530181,"about_ca_topic_score_gemma":0.00161077,"teacher_disagreement_score":0.9636117,"about_ca_system_score_codex":0.00020452484,"about_ca_system_score_gemma":0.00044660675,"threshold_uncertainty_score":0.99393},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3205724352","doi":"10.2147/idr.s335745","title":"Update on the COVID-19 Vaccine Research Trends: A Bibliometric Analysis","year":2021,"lang":"en","type":"article","venue":"Infection and Drug Resistance","topic":"SARS-CoV-2 and COVID-19 Research","field":"Medicine","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Coronavirus disease 2019 (COVID-19); Citation; Web of science; Pandemic; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Library science; 2019-20 coronavirus outbreak; China; Medicine; Geography; Infectious disease (medical specialty); Virology; Disease; Internal medicine; Meta-analysis; Computer science; Outbreak","score_opus":0.0835997424656843,"score_gpt":0.4226466915997224,"score_spread":0.3390469491340381,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3205724352","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9305232,0.003826872,0.00044600107,0.03466603,0.000092946495,0.00021770524,0.000012479421,0.00011387604,0.030100873],"genre_scores_gemma":[0.94246686,0.0005029917,0.000030616753,0.052085,0.00013735857,0.00007183601,0.000009499254,0.00001732334,0.0046785],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99742407,0.00049046404,0.0002590456,0.0005165048,0.00091678736,0.00039315922],"domain_scores_gemma":[0.99749756,0.0011613067,0.00005795834,0.0006386586,0.0005012728,0.00014327333],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0021830297,0.00014459343,0.00030373895,0.03470956,0.0004995668,0.00018700626,0.00010514384,0.00008312639,0.00032929674],"category_scores_gemma":[0.0027474586,0.0001000664,0.00017742721,0.18256977,0.00011254837,0.00008313695,0.00008883057,0.00054329017,0.00010713067],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023330082,0.0009770902,0.681891,0.00049639127,0.0014836729,0.0008019203,0.0006308788,0.000009809048,0.008875734,0.010501157,0.2653923,0.026606997],"study_design_scores_gemma":[0.0010661168,0.00005565827,0.11983816,0.000026328613,0.0001358657,0.000022089052,0.00008873577,0.00010249685,0.008109417,0.0005356543,0.86990744,0.0001120476],"about_ca_topic_score_codex":0.00043553978,"about_ca_topic_score_gemma":0.0033861045,"teacher_disagreement_score":0.60451514,"about_ca_system_score_codex":0.00023291605,"about_ca_system_score_gemma":0.00032886342,"threshold_uncertainty_score":0.97623116},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3207540549","doi":"10.1080/10888705.2021.1980727","title":"The Who, Where, and What of Publications in the <i>Journal of Applied Animal Welfare Science</i> from 2009 to 2019: A Bibliometric Analysis","year":2021,"lang":"en","type":"article","venue":"Journal of Applied Animal Welfare Science","topic":"Human-Animal Interaction Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of British Columbia, Okanagan Campus; University of British Columbia","funders":"","keywords":"Library science; Context (archaeology); Bibliometrics; Welfare; Salient; Content analysis; Animal welfare; Social science; Sociology; Political science; History; Computer science; Law; Biology; Ecology","score_opus":0.013344916181686126,"score_gpt":0.3170096083078018,"score_spread":0.30366469212611563,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3207540549","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9885026,0.0035676223,0.00008899636,0.005813069,0.00010662924,0.00013491957,0.000019381283,0.0000021868925,0.0017645869],"genre_scores_gemma":[0.99654275,0.0022457384,0.0007904823,0.00026137638,0.00012893039,0.000004656304,0.0000016030425,0.000008096359,0.000016344791],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9971103,0.00005488174,0.0008716997,0.00035600833,0.0012401835,0.00036689],"domain_scores_gemma":[0.99625534,0.00011513448,0.0011014959,0.00039633308,0.0019387107,0.0001929953],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0031745704,0.00018178334,0.0004139202,0.006926203,0.0007254649,0.0007590211,0.0014127077,0.00006528474,0.000042903514],"category_scores_gemma":[0.0005442046,0.000111660476,0.0001550646,0.045245755,0.0009954448,0.00016896208,0.00042690866,0.0003086355,0.0000015947481],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006349453,0.00014597211,0.0015478967,0.000007359852,0.00019758072,0.0000067087094,0.00090778334,0.00021148675,0.9870171,0.003005198,0.0016071354,0.004710857],"study_design_scores_gemma":[0.0009314031,0.0012849926,0.73378557,0.00005382293,0.00039411843,0.00018339917,0.04187601,0.000106051506,0.16475926,0.00014996843,0.05612364,0.00035178906],"about_ca_topic_score_codex":0.000029761386,"about_ca_topic_score_gemma":0.00018405124,"teacher_disagreement_score":0.8222578,"about_ca_system_score_codex":0.00006963679,"about_ca_system_score_gemma":0.00047541287,"threshold_uncertainty_score":0.975048},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W3207720485","doi":"10.5958/2249-3182.2020.00010.6","title":"A space-time investigation into the progress of doctoral research in library and information science (LIS) in India","year":2020,"lang":"en","type":"article","venue":"Gyankosh- The Journal of Library and Information Management","topic":"ICT in Developing Communities","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Shastri Indo-Canadian Institute","funders":"","keywords":"Library science; Space (punctuation); Information science; Data science; Computer science","score_opus":0.027477681828811554,"score_gpt":0.2538518395856318,"score_spread":0.22637415775682024,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3207720485","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.90666616,0.00045143944,0.0010198449,0.077397816,0.00014546503,0.00076370547,0.0000017551358,0.000041311523,0.013512487],"genre_scores_gemma":[0.98454654,0.0017812782,0.010961619,0.0026555082,0.000025060277,0.000007485824,0.0000047476087,0.0000032889652,0.000014485016],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9982741,0.00030129257,0.0006484586,0.000049068512,0.00055586593,0.0001712226],"domain_scores_gemma":[0.9991006,0.00018169383,0.00038669398,0.00021610905,0.00004873845,0.00006613708],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0017971575,0.00008656683,0.00012377273,0.0011249472,0.00020639099,0.0006779574,0.001269087,0.000030525476,0.00000516086],"category_scores_gemma":[0.00004184702,0.00005508284,0.000015141624,0.0026308994,0.000470281,0.09689255,0.0014249717,0.00035326113,0.0000041218022],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015815895,0.000021105074,0.018800884,0.00052894285,0.000022949795,0.0000038793305,0.17224975,0.000611362,0.00002309162,0.76164,0.0010849767,0.044854883],"study_design_scores_gemma":[0.0019114059,0.00070659904,0.80014884,0.0011158518,0.000013892332,0.000042765634,0.020925917,0.08409189,0.0035245477,0.030491225,0.05666984,0.0003572271],"about_ca_topic_score_codex":0.0000024904373,"about_ca_topic_score_gemma":6.0714484e-8,"teacher_disagreement_score":0.78134793,"about_ca_system_score_codex":0.000014179621,"about_ca_system_score_gemma":0.00015371504,"threshold_uncertainty_score":0.91573876},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W3208611076","doi":"10.21037/apm-21-2757","title":"Pressure ulcers in patients with diabetes: a bibliometrics analysis","year":2021,"lang":"en","type":"article","venue":"Annals of Palliative Medicine","topic":"Pressure Ulcer Prevention and Management","field":"Health Professions","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Bibliometrics; Citation; Centrality; China; Science Citation Index; Diabetes mellitus; Library science; Index (typography); Family medicine; Gerontology; Political science; Law","score_opus":0.1126698826476841,"score_gpt":0.4506884416778527,"score_spread":0.33801855903016864,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3208611076","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9551692,0.003422744,0.004606273,0.013050489,0.00042486214,0.001712307,0.00012512218,0.00006698009,0.021422014],"genre_scores_gemma":[0.9929876,0.00030224695,0.00017598287,0.0032011452,0.000048156395,0.000057516914,0.0001442347,0.000012431922,0.0030707337],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99755865,0.00056176,0.0005924702,0.0002979877,0.0006235467,0.0003655904],"domain_scores_gemma":[0.9975787,0.00047879585,0.00041206073,0.0003594944,0.0010491135,0.00012187755],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0010021992,0.00014669618,0.00060444814,0.0037972191,0.00008093438,0.000002751867,0.00016253829,0.00008086296,0.00210381],"category_scores_gemma":[0.0009438214,0.00010457847,0.00008255868,0.016096836,0.00008782838,0.000097590724,0.00011845708,0.00027561386,0.000013561758],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006230369,0.00023120533,0.9812916,0.00035380063,0.0015411925,0.000003713943,0.0017722383,0.00009787292,0.000007576765,0.00044916454,0.013498191,0.00069115387],"study_design_scores_gemma":[0.002345727,0.0002012307,0.97150856,0.0003822872,0.00045965266,1.2630386e-9,0.0011109392,0.00013866395,0.00003479943,0.00022673748,0.023489716,0.00010166977],"about_ca_topic_score_codex":0.000120389144,"about_ca_topic_score_gemma":0.0003973709,"teacher_disagreement_score":0.037818342,"about_ca_system_score_codex":0.000018160426,"about_ca_system_score_gemma":0.000077246026,"threshold_uncertainty_score":0.9988084},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3208729672","doi":"10.3389/fped.2021.762531","title":"Global Research Trends in Pediatric Trauma From 1968 to 2021: A Bibliometric Analysis","year":2021,"lang":"en","type":"article","venue":"Frontiers in Pediatrics","topic":"Trauma and Emergency Care Studies","field":"Medicine","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Southeast University; Health Commission of Hubei Province","keywords":"Medicine; Bibliometrics; Citation; Data science; Library science; Computer science","score_opus":0.04981567414039937,"score_gpt":0.37386996858272875,"score_spread":0.32405429444232936,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3208729672","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94343793,0.03723512,0.0014588835,0.0010472677,0.0019479571,0.0001908062,0.00021539434,0.000036279776,0.014430347],"genre_scores_gemma":[0.9711492,0.013305654,0.012425789,0.00011434135,0.0015945911,0.000056108944,0.00022988525,0.000027910119,0.0010965338],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9960071,0.00024382997,0.00073130085,0.0009162448,0.0011554294,0.0009461327],"domain_scores_gemma":[0.99828756,0.000167484,0.000065871536,0.000616684,0.00047404467,0.00038833628],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0008684568,0.0002746139,0.000909325,0.14802362,0.00009471059,0.000050728137,0.00028126774,0.00026024715,0.00047438598],"category_scores_gemma":[0.00087820564,0.00028222,0.0003630092,0.61385584,0.000047531732,0.00011156332,0.00021963993,0.0005550622,0.000049276066],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010022687,0.00027232192,0.7507536,0.000031944124,0.00013793877,0.00049969973,0.0004120349,0.00012457228,9.4921074e-7,0.00000905182,0.1382319,0.10942577],"study_design_scores_gemma":[0.001322217,0.00012603002,0.98707694,0.0000067953824,0.00091185735,0.0000023194157,0.0013482099,0.00011202291,0.0000059951935,0.00029941273,0.008525868,0.00026232642],"about_ca_topic_score_codex":0.0017743297,"about_ca_topic_score_gemma":0.004375926,"teacher_disagreement_score":0.46583223,"about_ca_system_score_codex":0.0006887891,"about_ca_system_score_gemma":0.0002800918,"threshold_uncertainty_score":0.999963},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3209634622","doi":"10.20944/preprints202110.0380.v1","title":"Mapping Maternal Health in the New Media Environment: A Scientometric Analysis","year":2021,"lang":"en","type":"preprint","venue":"Preprints.org","topic":"Maternal Mental Health During Pregnancy and Postpartum","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Maternal health; Web of science; Latin Americans; Millennium Development Goals; Health care; Global health; Library science; Political science; Medicine; Geography; Public relations; Environmental health; MEDLINE; Economic growth; Developing country; Population; Computer science; Health services","score_opus":0.14344979431729252,"score_gpt":0.3767326587241007,"score_spread":0.2332828644068082,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3209634622","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9865237,0.0040117498,0.0002647599,0.0060175736,0.00059858744,0.0010903304,0.000029529878,0.000058182493,0.0014055929],"genre_scores_gemma":[0.9920266,0.003520864,0.0004728762,0.0021178839,0.00026553165,0.00011416563,0.00019988393,0.000030831427,0.0012513302],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9956824,0.00044025996,0.0010278478,0.0011238692,0.0009976882,0.0007279024],"domain_scores_gemma":[0.9972435,0.00014147324,0.00050423073,0.0015962995,0.00003351595,0.00048093952],"candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0022765016,0.00039195333,0.0009905904,0.0020180005,0.00013020456,0.00006127641,0.000707898,0.00024118382,0.0023461527],"category_scores_gemma":[0.00015872235,0.00031856276,0.0004064475,0.0022841725,0.00007928571,0.00007067031,0.0012985775,0.0013978637,0.00048244104],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007820414,0.00025668176,0.98773694,0.0024856033,0.00032665787,0.00017024786,0.0075045526,0.00025054452,0.00005808871,0.000029123748,0.000017172757,0.0010861753],"study_design_scores_gemma":[0.000618568,0.000026414124,0.9923522,0.0034755063,0.00019543283,0.0000641291,0.00032642158,0.00011923961,0.0010859143,0.00019658123,0.0013149776,0.00022462486],"about_ca_topic_score_codex":0.0026987565,"about_ca_topic_score_gemma":0.00010069983,"teacher_disagreement_score":0.007178131,"about_ca_system_score_codex":0.0005683959,"about_ca_system_score_gemma":0.00055722176,"threshold_uncertainty_score":0.9999266},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3214883256","doi":"10.1515/jcim-2021-0417","title":"Trends in the St. John’s wort ( <i>Hypericum perforatum</i> ) research literature: a bibliometric analysis","year":2021,"lang":"en","type":"article","venue":"Journal of Complementary and Integrative Medicine","topic":"Natural Compound Pharmacology Studies","field":"Agricultural and Biological Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Impact","funders":"McMaster University","keywords":"Hypericum perforatum; Hypericum; Bibliometrics; Alternative medicine; Medicine; Traditional medicine; Library science; Computer science","score_opus":0.10624130967043492,"score_gpt":0.40507099549662945,"score_spread":0.2988296858261945,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3214883256","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93215775,0.02321447,0.000001629682,0.043309413,0.00019002907,0.000064713786,0.000033255,0.0000024679734,0.0010262732],"genre_scores_gemma":[0.99392986,0.0033779764,0.0000717564,0.001756094,0.00054187735,0.000004193494,0.00010872184,5.250379e-7,0.000208967],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9974426,0.0010176371,0.00047788923,0.00018258055,0.00063291524,0.00024638046],"domain_scores_gemma":[0.9973053,0.001598291,0.00016571512,0.00004843001,0.00080448424,0.00007780571],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0026138343,0.00014884614,0.00047612187,0.005604195,0.00028824832,0.000063321735,0.00033313505,0.000036627644,0.0016708969],"category_scores_gemma":[0.00020394959,0.000037334237,0.00014012327,0.10795149,0.0002606906,0.00020065645,0.00009821721,0.0009411062,5.1367516e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00054196373,0.000596601,0.36361906,0.00001787763,0.0018658437,0.0010240447,0.007882872,0.0000032631783,0.07627206,0.00086045626,0.39145172,0.15586422],"study_design_scores_gemma":[0.0006183838,0.0017759822,0.8616151,0.000096250435,0.0002717919,0.00017519042,0.01892089,0.000028499437,0.00045119488,0.0006553078,0.11529141,0.00009995617],"about_ca_topic_score_codex":0.00012609812,"about_ca_topic_score_gemma":0.001815213,"teacher_disagreement_score":0.4979961,"about_ca_system_score_codex":0.000044308803,"about_ca_system_score_gemma":0.000012878298,"threshold_uncertainty_score":0.9992417},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W3217037246","doi":"10.18280/ijsdp.160612","title":"Research in Petroleum and Environment: A Bibliometric Analysis in South America","year":2021,"lang":"en","type":"article","venue":"International Journal of Sustainable Development and Planning","topic":"Chemistry and Chemical Engineering","field":"Environmental Science","cited_by":31,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Escuela Superior Politécnica del Litoral","keywords":"Environmental planning; Environmental pollution; Natural resource; Pollution; Petroleum; Environmental protection; Resource (disambiguation); Exploitation of natural resources; Environmental resource management; Environmental remediation; Geography; Environmental science; Political science; Ecology; Computer science; Contamination","score_opus":0.015646011801789945,"score_gpt":0.26574322148469826,"score_spread":0.25009720968290833,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3217037246","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99764234,0.00061355165,0.00044036176,0.00012080062,0.000013465242,0.000013441111,3.889309e-7,0.000001277627,0.0011543619],"genre_scores_gemma":[0.9982451,0.00013003108,0.0012915776,0.00001670077,0.00001802924,0.0000016284843,0.0000037229531,0.0000028188836,0.0002903513],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99903464,0.000019311856,0.00023794865,0.00012376985,0.00039552295,0.00018880694],"domain_scores_gemma":[0.99972224,0.00009427698,0.00005794695,0.00003153066,0.0000288989,0.00006512293],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00049137144,0.000058214715,0.00011905767,0.006071626,0.000028920584,0.00005708092,0.000108220454,0.000031112813,0.00014175726],"category_scores_gemma":[0.00007991854,0.000057850742,0.000018266164,0.008567718,0.000038427184,0.00015868366,0.0001958959,0.00022750946,0.0000014837102],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006190408,0.00007505169,0.91914964,0.000026169342,0.00013445872,0.006352157,0.0058780797,0.05599778,0.002164388,0.000019951733,0.000037248818,0.010103183],"study_design_scores_gemma":[0.00082808867,0.000016038854,0.9459614,0.0000905494,0.000016086573,0.00008339964,0.03138527,0.0033545066,0.006057159,0.0002503909,0.011776897,0.00018021322],"about_ca_topic_score_codex":0.000018993382,"about_ca_topic_score_gemma":0.0000011303645,"teacher_disagreement_score":0.052643273,"about_ca_system_score_codex":0.000247894,"about_ca_system_score_gemma":0.00002859301,"threshold_uncertainty_score":0.5417649},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W350340533","doi":"","title":"Publishing Trends and Open Access Journals at York University","year":2011,"lang":"en","type":"article","venue":"YorkSpace (York University)","topic":"Publishing and Scholarly Communication","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Publishing; Library science; Media studies; Political science; World Wide Web; Computer science; Sociology; Law","score_opus":0.24217321604234088,"score_gpt":0.282328559418204,"score_spread":0.04015534337586313,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W350340533","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07882577,0.00035236002,0.00005067971,0.0040389257,0.0002998733,0.00011339669,0.000045620174,0.000178385,0.91609496],"genre_scores_gemma":[0.76905805,0.00020011366,0.00039248893,0.00017317216,0.000118880016,1.7784971e-7,0.000032264958,0.000017494787,0.23000735],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99865127,0.00032041062,0.00013019104,0.0003532426,0.00021740873,0.00032747615],"domain_scores_gemma":[0.99859405,0.00011219424,0.0002905831,0.0005708428,0.00016967021,0.00026264624],"candidate_categories":["sts","scholarly_communication","insufficient_payload"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.0006042184,0.00018673774,0.0002466217,0.0009107008,0.0019335636,0.009719148,0.0041176137,0.0001392685,0.002975842],"category_scores_gemma":[0.00008269023,0.00020387057,0.00008458962,0.0005947723,0.00027968021,0.017399767,0.0031016956,0.0005545229,0.000026682948],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00039116593,0.00020660592,0.025892055,0.000018120554,0.00022052982,0.00010553836,0.07291286,0.0000056797858,0.000024205376,0.6212194,0.26431096,0.014692897],"study_design_scores_gemma":[0.0007512498,0.000042369305,0.011881637,0.00006325627,0.00007422785,0.0000062067356,0.01894024,0.000023111623,0.000011362493,0.00056890777,0.96734643,0.00029099872],"about_ca_topic_score_codex":0.008373734,"about_ca_topic_score_gemma":0.018551232,"teacher_disagreement_score":0.7030355,"about_ca_system_score_codex":0.00026407398,"about_ca_system_score_gemma":0.00009793323,"threshold_uncertainty_score":0.9993658},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","scholarly_communication"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W394390834","doi":"10.29173/cais241","title":"Visibilité de la production scientifique : une étude scientométrique au Pôle Universitaire Lyonnais","year":2013,"lang":"fr","type":"article","venue":"Proceedings of the Annual Conference of CAIS / Actes du congrès annuel de l ACSI","topic":"Cultural Insights and Digital Impacts","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Political science; Humanities; Art","score_opus":0.06435642797461559,"score_gpt":0.2686356409985202,"score_spread":0.20427921302390462,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W394390834","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94231194,0.00035165893,0.00026424232,0.01943191,0.00074323674,0.00073116965,0.0000581937,0.00007562273,0.036032047],"genre_scores_gemma":[0.98244977,0.00010222601,0.0009931956,0.0001773078,0.0001792711,0.000017270733,0.0000025044649,0.000022019172,0.016056426],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9968382,0.00009542882,0.00060333585,0.00078965834,0.00076626416,0.0009071291],"domain_scores_gemma":[0.9358445,0.00019827932,0.0009645153,0.0004961373,0.06197594,0.0005206608],"candidate_categories":["metaresearch","metaepi_narrow","scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.001400018,0.0004374221,0.00052541576,0.00029245095,0.0004442244,0.013828266,0.0028101972,0.0003119678,0.000038136754],"category_scores_gemma":[0.015349381,0.00032855978,0.00029308512,0.002197694,0.0024090402,0.048891444,0.0015046665,0.0004471584,0.000018771014],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001163575,0.0012934835,0.061383754,0.0017095521,0.00023853235,0.000014496896,0.3043925,0.00015080796,0.16774583,0.37325907,0.04133736,0.048358247],"study_design_scores_gemma":[0.0013205516,0.0011698649,0.42261276,0.0036543193,0.00023466969,0.0006956609,0.013591654,0.008319844,0.34600708,0.085179485,0.11569247,0.0015216462],"about_ca_topic_score_codex":0.007321567,"about_ca_topic_score_gemma":0.0002378575,"teacher_disagreement_score":0.361229,"about_ca_system_score_codex":0.00038453934,"about_ca_system_score_gemma":0.0014385316,"threshold_uncertainty_score":0.9999167},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4200345448","doi":"10.1071/py21069","title":"Internationalisation of general practice journals: a bibliometric analysis of the Science Citation Index database","year":2021,"lang":"en","type":"article","venue":"Australian Journal of Primary Health","topic":"Primary Care and Health Outcomes","field":"Health Professions","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Taipei Veterans General Hospital","keywords":"Internationalization; General practice; Index (typography); Subject (documents); Citation; Bibliometrics; Science Citation Index; Citation index; Population health","score_opus":0.17556736909796392,"score_gpt":0.5089626947258716,"score_spread":0.33339532562790775,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200345448","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8891302,0.0031177443,0.0038523457,0.09727689,0.0026734401,0.00047553508,0.00024031747,0.000010060974,0.0032234695],"genre_scores_gemma":[0.9683951,0.0029473584,0.009921487,0.017574545,0.00036003476,0.0000041110343,0.000047565387,0.000016491818,0.0007333142],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.993193,0.0016385955,0.002564618,0.00020920984,0.0019494651,0.00044515202],"domain_scores_gemma":[0.98692405,0.0014143928,0.005659791,0.0004649518,0.005214048,0.00032277632],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.010812459,0.00013601541,0.00076440093,0.026517477,0.00048226127,0.000025084762,0.0006364288,0.000095231444,0.00046839734],"category_scores_gemma":[0.00492606,0.00010115459,0.0002976345,0.08878628,0.00020140543,0.0014120312,0.00017958625,0.00093809166,0.0000028540642],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005720607,0.00084340596,0.8484991,0.0021445234,0.0018491681,0.000054651984,0.006927102,0.0015098727,0.017544601,0.005164538,0.070487775,0.044403166],"study_design_scores_gemma":[0.000879254,0.0001369656,0.9881947,0.00027985137,0.00041218786,0.000048717066,0.001516172,0.00005486759,0.0004834454,0.00021387008,0.007695501,0.00008444299],"about_ca_topic_score_codex":0.0008277614,"about_ca_topic_score_gemma":0.00015685172,"teacher_disagreement_score":0.13969558,"about_ca_system_score_codex":0.0017872906,"about_ca_system_score_gemma":0.02164028,"threshold_uncertainty_score":0.98451614},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4200350299","doi":"10.21203/rs.3.rs-1062152/v1","title":"The Role of Decision Impact Studies in Genomic Medicine in Cancer Care: A Scoping Review and Bibliometric Analysis Protocol","year":2021,"lang":"en","type":"review","venue":"Research Square","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"Protocol (science); Cancer; Medicine; Alternative medicine; Internal medicine; Pathology","score_opus":0.7632741537937451,"score_gpt":0.782732483875663,"score_spread":0.019458330081917885,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200350299","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000032902164,0.89820904,3.5594616e-7,0.0003251577,0.000016713278,0.10114551,0.000027067155,0.0000048725637,0.0002383763],"genre_scores_gemma":[0.00007032137,0.90575606,0.000042900712,0.000017495926,0.000090010486,0.09389687,0.00002384323,0.000044350552,0.00005814789],"study_design_codex":"design_other","study_design_gemma":"systematic_review","domain_scores_codex":[0.9891151,0.0029512267,0.0023801199,0.0009668586,0.0035844878,0.0010021768],"domain_scores_gemma":[0.9290043,0.064372055,0.00039067358,0.001498634,0.004280655,0.00045367386],"candidate_categories":["metaresearch","bibliometrics","research_integrity"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.03718505,0.000370581,0.005167136,0.08323281,0.0001766629,0.000040301744,0.0007037455,0.0005944881,0.00026468694],"category_scores_gemma":[0.13488151,0.00019239982,0.00064630306,0.2550959,0.0010783853,0.000057073983,0.0010925119,0.0052347896,0.000007837288],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003266625,0.000046977035,0.00533801,0.35880938,0.00061216607,0.00009418593,0.00011921488,8.661562e-7,2.9815305e-7,0.000019342177,0.00008824693,0.63483864],"study_design_scores_gemma":[0.00068369124,0.0006057283,0.0024637943,0.94311076,0.0010954717,0.000009993033,0.00059037696,0.0000057715347,6.6584846e-7,0.00036107382,0.050919663,0.00015303063],"about_ca_topic_score_codex":0.011248775,"about_ca_topic_score_gemma":0.055067502,"teacher_disagreement_score":0.63468564,"about_ca_system_score_codex":0.002364868,"about_ca_system_score_gemma":0.010732344,"threshold_uncertainty_score":0.9970602},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4200395585","doi":"10.1007/s40037-021-00695-4","title":"Bibliometrics: Methods for studying academic publishing","year":2021,"lang":"en","type":"article","venue":"Perspectives on Medical Education","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":938,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Royal College of Physicians and Surgeons of Canada; University of Ottawa","funders":"","keywords":"Bibliometrics; Terminology; Publishing; Data science; Strengths and weaknesses; Population health; Computer science; Population; Library science; Medicine; Psychology; Political science","score_opus":0.652587060367001,"score_gpt":0.6967628458869507,"score_spread":0.04417578551994972,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200395585","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12756842,0.03384215,0.69538605,0.10588891,0.0070561157,0.00096526346,0.00002620779,0.00012455856,0.029142316],"genre_scores_gemma":[0.8358486,0.006136074,0.14205182,0.006544131,0.0024207246,0.00032180033,0.000040270832,0.000053969547,0.0065826005],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.98567086,0.00097252044,0.00080706837,0.0013116566,0.010596315,0.00064159604],"domain_scores_gemma":[0.90955573,0.060009766,0.0007112044,0.0019139363,0.02511396,0.0026953814],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.05948411,0.00017035092,0.0003682181,0.13695721,0.00041316703,0.004798167,0.002418063,0.0003491818,0.0022256284],"category_scores_gemma":[0.7677514,0.0001292189,0.00022387369,0.41679198,0.00017749392,0.0016280803,0.00049591,0.0010730888,0.00010692501],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017092674,0.00063028897,0.0051643876,0.0000071091395,0.000019598398,0.0000016973703,0.0020814126,0.0000032151559,0.00020612084,0.0243053,0.05017921,0.91738456],"study_design_scores_gemma":[0.0019848754,0.0005301767,0.099297084,0.00012583559,0.000037809892,0.00005040143,0.20438051,0.017545173,0.0029247259,0.18374409,0.48861232,0.0007670138],"about_ca_topic_score_codex":0.000032507895,"about_ca_topic_score_gemma":0.000003341181,"teacher_disagreement_score":0.9166176,"about_ca_system_score_codex":0.0004534263,"about_ca_system_score_gemma":0.004114283,"threshold_uncertainty_score":0.9986865},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4200425828","doi":"10.1016/j.ijcard.2021.12.021","title":"A survey of retractions in the cardiovascular literature","year":2021,"lang":"en","type":"review","venue":"International Journal of Cardiology","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":37,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Health Sciences Centre; Sunnybrook Health Science Centre; University of Toronto","funders":"","keywords":"Medicine; Interquartile range; Scientific misconduct; Misconduct; Internal medicine; Alternative medicine; Pathology; Law","score_opus":0.09250250443613138,"score_gpt":0.4167842958079029,"score_spread":0.3242817913717715,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200425828","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000011992387,0.9891327,0.00029180603,0.00037809508,0.0030366955,0.000103021855,0.00011610472,0.0000016384432,0.0069279415],"genre_scores_gemma":[0.0017239403,0.9958372,0.000039412804,0.000056425,0.0021964167,0.000003777276,0.00006608736,0.0000066848183,0.00007001602],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9915834,0.0064416556,0.0008091939,0.00012565864,0.00089480355,0.00014529684],"domain_scores_gemma":[0.99520725,0.0023127764,0.0006472354,0.00016391595,0.0016345095,0.000034299548],"candidate_categories":["research_integrity"],"consensus_categories":[],"category_scores_codex":[0.011537089,0.00012942051,0.0014423481,0.00032527215,0.000049907856,0.000047059453,0.0012451739,0.001168715,0.000027228347],"category_scores_gemma":[0.0051385467,0.000084990126,0.0015605264,0.00048151021,0.00018483847,0.00012118431,0.000044440134,0.0032132366,0.0000033760107],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019338657,0.000035782705,0.00057270087,0.0003390106,0.0064968825,0.0016258089,0.005223606,0.000098506665,9.375529e-8,0.010180745,0.018211352,0.9571962],"study_design_scores_gemma":[0.000073318384,0.000013058065,0.00034854183,0.0026740818,0.0002767363,0.0010012776,0.00038010644,2.0102449e-7,3.8381543e-8,0.00019181731,0.9949723,0.00006851569],"about_ca_topic_score_codex":0.00066724786,"about_ca_topic_score_gemma":0.00018901643,"teacher_disagreement_score":0.976761,"about_ca_system_score_codex":0.0001706649,"about_ca_system_score_gemma":0.0011732561,"threshold_uncertainty_score":0.9990864},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4200434117","doi":"10.5152/turkjorthod.2021.21030","title":"Reporting Sample Size Calculation in Randomized Clinical Trials Published in 4 Orthodontic Journals","year":2021,"lang":"en","type":"article","venue":"","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Sample size determination; Randomized controlled trial; Sample (material); Clinical trial; Large sample","score_opus":0.9305638248511154,"score_gpt":0.6830218970389912,"score_spread":0.24754192781212425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200434117","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":"methods","domain_gemma":"methods","model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":"methods","prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.69625276,0.00197628,0.27044445,0.008328093,0.0012975414,0.0027233965,0.00000410216,0.000010524079,0.018962871],"genre_scores_gemma":[0.9244865,0.000097371485,0.065250896,0.0006605572,0.00018001017,0.000060257327,0.0000065141694,0.0000101123405,0.009247771],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.18262753,0.540191,0.26305196,0.0027722917,0.0106569575,0.00070025434],"domain_scores_gemma":[0.010496321,0.8701295,0.108629435,0.006072607,0.0042360732,0.0004360302],"candidate_categories":["metaresearch","metaepi_broad","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.9396527,0.00022985318,0.018691622,0.00058768614,0.000055338784,0.0024044567,0.0006789373,0.0001853366,0.06934137],"category_scores_gemma":[0.99388576,0.0000994583,0.0065446594,0.0034903614,0.000042425363,0.0005786543,0.00013261836,0.00036560852,0.00023734964],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007767307,0.00015049805,0.9365101,0.000025585809,0.0002675996,0.00011804862,0.00025547302,0.0004158061,0.00006574341,0.0011371793,0.013488779,0.046788458],"study_design_scores_gemma":[0.12315146,0.000023178149,0.53698856,0.000422645,0.0009573708,0.00009998527,0.0020103082,0.1589502,0.00008035415,0.16116956,0.015456751,0.00068966794],"about_ca_topic_score_codex":0.00017803875,"about_ca_topic_score_gemma":0.0014204256,"teacher_disagreement_score":0.39952156,"about_ca_system_score_codex":0.0000394986,"about_ca_system_score_gemma":0.00033612477,"threshold_uncertainty_score":0.9986311},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4200496290","doi":"10.1161/circheartfailure.121.008685","title":"Factors Associated With Racial and Ethnic Diversity Among Heart Failure Trial Participants: A Systematic Bibliometric Review","year":2021,"lang":"en","type":"article","venue":"Circulation Heart Failure","topic":"Heart Failure Treatment and Management","field":"Medicine","cited_by":59,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hamilton Health Sciences; Population Health Research Institute; McMaster University","funders":"","keywords":"Medicine; Ethnic group; Heart failure; Diversity (politics); MEDLINE; Gerontology; Internal medicine","score_opus":0.11610561275819081,"score_gpt":0.32771820513228284,"score_spread":0.21161259237409202,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200496290","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9780289,0.006117116,0.00009990249,0.011617366,0.00013595427,0.0037460427,0.00001326763,0.00020272557,0.000038704195],"genre_scores_gemma":[0.9979217,0.00015045861,0.00030160957,0.0008556422,0.000095949115,0.00012685246,0.00022199047,0.000033555334,0.00029223898],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.996965,0.00047047803,0.00059647247,0.00057435205,0.00088123005,0.00051249465],"domain_scores_gemma":[0.9981523,0.0003394041,0.00021442786,0.0004902221,0.0004020413,0.00040158426],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0008324893,0.00035128265,0.001206175,0.0023106292,0.00038467217,0.000085758285,0.00006423097,0.00021495207,0.00034821933],"category_scores_gemma":[0.0014734911,0.0002668301,0.00025575727,0.013330496,0.00009461158,0.00036814625,0.0001483253,0.00028538596,0.00003821366],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000396007,0.00077510066,0.93970484,0.020404575,0.0013525533,0.00010141221,0.00088116433,0.000015827438,0.00026540592,0.00008617166,0.03590127,0.00011564755],"study_design_scores_gemma":[0.023467956,0.00055848697,0.94319534,0.020005297,0.003887853,0.000030209712,0.000674681,0.0001446758,0.00011472308,0.00002125857,0.0073417397,0.0005577886],"about_ca_topic_score_codex":0.000048217385,"about_ca_topic_score_gemma":0.00044442684,"teacher_disagreement_score":0.02855953,"about_ca_system_score_codex":0.00019593105,"about_ca_system_score_gemma":0.000162389,"threshold_uncertainty_score":0.99997836},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"systematic_review","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4205376244","doi":"10.31219/osf.io/jm3hf","title":"An infographic of data sharing in North America","year":2022,"lang":"en","type":"preprint","venue":"","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université Laval","funders":"Canadian Institutes of Health Research","keywords":"Infographic; Data sharing; Political science; Geography; Regional science; Computer science; Data mining","score_opus":0.2607468458245488,"score_gpt":0.430734309988772,"score_spread":0.16998746416422317,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4205376244","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08038446,0.00025505573,0.88781005,0.0017246252,0.00040171115,0.0011864078,0.00036772026,0.0003225991,0.02754734],"genre_scores_gemma":[0.5710574,0.0044277124,0.4180894,0.0005148709,0.00006367437,0.00019124319,0.0049860193,0.000035966288,0.0006337243],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99685216,0.00021899964,0.00040027066,0.0014318271,0.00081124116,0.00028547383],"domain_scores_gemma":[0.9880226,0.00013318093,0.0003093552,0.011418623,0.00003800708,0.000078193734],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.0017964349,0.00014574736,0.00023706524,0.0008866884,0.000045365556,0.0016079664,0.03574179,0.000033570337,0.00015454332],"category_scores_gemma":[0.00022078496,0.00014801776,0.000031077772,0.0016423505,0.000055817716,0.028728232,0.100694045,0.0008112019,0.0000051229395],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000042710068,0.001398733,0.6292022,0.0011703839,0.00034939934,0.00036225902,0.0018647027,0.032993898,0.000035245914,0.09927502,0.0027311032,0.23057437],"study_design_scores_gemma":[0.00024762485,0.00016003438,0.18708578,0.000038327766,0.00001474249,0.0000011413302,0.00030160704,0.6457174,0.0000045374504,0.0015386496,0.16435957,0.0005305788],"about_ca_topic_score_codex":0.0050916346,"about_ca_topic_score_gemma":0.0027133224,"teacher_disagreement_score":0.6127235,"about_ca_system_score_codex":0.000032044358,"about_ca_system_score_gemma":0.00014524012,"threshold_uncertainty_score":0.99942845},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4205547057","doi":"10.14429/djlit.42.1.17480","title":"A Bibliometric Study of Papers Published in Library and Information Science Research during 1994 2020","year":2021,"lang":"en","type":"article","venue":"DESIDOC Journal of Library & Information Technology","topic":"Middle East and Rwanda Conflicts","field":"Social Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Indiana University Bloomington; University of Illinois at Urbana-Champaign; Wuhan University; Tampereen Yliopisto; University of Alberta; University of Wisconsin-Milwaukee; Florida State University","keywords":"Library science; Distribution (mathematics); Citation; Bibliometrics; Period (music); Science Citation Index; Index (typography); Demography; Geography; Regional science; Mathematics; Computer science; Sociology; World Wide Web","score_opus":0.021831510318585372,"score_gpt":0.3014450265080845,"score_spread":0.27961351618949914,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4205547057","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95479405,0.00039579644,0.000007809394,0.004671861,0.00018099752,0.00022613737,0.0000026522837,0.00004846915,0.039672248],"genre_scores_gemma":[0.9968449,0.0023429482,0.0005494537,0.00010415417,0.000052566793,0.0000036347,0.0000036193458,0.0000040110604,0.00009471929],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9974358,0.00019127413,0.00096662587,0.00008907934,0.0009780921,0.0003391226],"domain_scores_gemma":[0.99867976,0.0001469187,0.00055757764,0.00018119182,0.00029961797,0.0001349501],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0016868475,0.000084792446,0.00024351604,0.055626664,0.00036061744,0.0006183372,0.00066216564,0.00016146712,0.00033726785],"category_scores_gemma":[0.001382703,0.00007619901,0.000033219494,0.10306687,0.00057602907,0.07239666,0.0003712139,0.0005595373,0.0000045681],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040745846,0.00051560794,0.5200183,0.00032732217,0.00008149057,0.00012340515,0.09162985,0.000081557526,0.0016594137,0.25094083,0.001377974,0.13283679],"study_design_scores_gemma":[0.004594406,0.0010183135,0.4968173,0.00036975558,0.000018127215,0.00023183729,0.26653787,0.00009206041,0.021018194,0.0046248026,0.20429981,0.00037753413],"about_ca_topic_score_codex":0.000013938945,"about_ca_topic_score_gemma":0.0000045133656,"teacher_disagreement_score":0.24631605,"about_ca_system_score_codex":0.000030470135,"about_ca_system_score_gemma":0.0012331998,"threshold_uncertainty_score":0.955077},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4206476323","doi":"10.21272/mmi.2021.4-14","title":"Analysis of territories marketing activities among small and medium business: a bibliometric analysis","year":2021,"lang":"en","type":"article","venue":"Marketing and Management of Innovations","topic":"Diverse Scientific Research in Ukraine","field":"Environmental Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Tourism; Marketing; Marketing research; Politics; Place branding; Marketing science; Marketing management; Sociology; Public relations; Political science; Business; Relationship marketing","score_opus":0.01951918305988723,"score_gpt":0.2569764943964904,"score_spread":0.23745731133660317,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4206476323","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9799471,0.00007522313,0.00090409647,0.00022445584,0.00006012659,0.00009664763,0.000034784665,0.00001494276,0.018642643],"genre_scores_gemma":[0.9907421,0.0006179814,0.0058574704,0.00001238328,0.000007761015,0.000012912396,0.000060366267,0.0000052519936,0.0026838067],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9983662,0.0001832899,0.00036846253,0.00035184118,0.00053241686,0.00019780918],"domain_scores_gemma":[0.9988769,0.00044055574,0.00022385824,0.00027972806,0.00013257776,0.000046348418],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.004134944,0.00010192004,0.00027660365,0.017196152,0.00015750705,0.00011082978,0.0001510468,0.000029802624,0.00061963656],"category_scores_gemma":[0.0008577015,0.00009994314,0.000064582186,0.19872606,0.00043440284,0.00021014069,0.00071523356,0.000059300026,4.305917e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001916405,0.000073871095,0.97903943,0.00023361009,0.0021400296,0.000007744235,0.00013490651,0.00047511733,0.000995746,0.00007791764,0.00037884284,0.016423617],"study_design_scores_gemma":[0.00012029926,0.000005855036,0.9939508,0.000048717196,0.001163594,2.433737e-7,0.0015203862,0.0018770407,0.00016510685,0.000012016334,0.0010469868,0.00008898824],"about_ca_topic_score_codex":0.00044875723,"about_ca_topic_score_gemma":0.0004524858,"teacher_disagreement_score":0.1815299,"about_ca_system_score_codex":0.000029185325,"about_ca_system_score_gemma":0.000008737697,"threshold_uncertainty_score":0.9939431},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4206834335","doi":"10.1016/j.psj.2022.101713","title":"Scientometric analysis and perspective of IgY technology study","year":2022,"lang":"en","type":"article","venue":"Poultry Science","topic":"Animal health and immunology","field":"Veterinary","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Guelph","funders":"National Natural Science Foundation of China","keywords":"China; Library science; Geography; Veterinary medicine; Biotechnology; Medicine; Biology; Computer science","score_opus":0.05154118043715782,"score_gpt":0.41193442035227307,"score_spread":0.36039323991511524,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4206834335","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.996578,0.00052272325,0.00006508353,0.00027387828,0.00009809132,0.00017938482,0.000009342487,0.000032423985,0.0022410983],"genre_scores_gemma":[0.99958575,0.000005575295,0.00023366886,0.000046073845,0.000004648618,0.00002228576,3.4439552e-7,0.0000027970348,0.000098851575],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9986174,0.00007185544,0.00017498549,0.00043572325,0.00038658595,0.0003134356],"domain_scores_gemma":[0.99929863,0.00006229214,0.00010528703,0.000285742,0.00018675037,0.0000612773],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0010060704,0.000067092114,0.00019280573,0.005161071,0.0006323747,0.000012178211,0.0005271183,0.00002016475,0.0001969002],"category_scores_gemma":[0.0003869693,0.000062819774,0.000029573484,0.030272463,0.0008239751,0.00012832225,0.00064757984,0.00018745547,0.000008239797],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015204622,0.0002883294,0.98339075,0.0000032050107,0.000036528527,0.0000151773975,0.0016907606,0.0000031440566,0.008899451,0.0036120797,0.0000059051827,0.0019026397],"study_design_scores_gemma":[0.0002084872,0.0052075367,0.92483026,5.2625245e-7,0.000046675912,0.000058836504,0.06877684,0.00021543493,0.00008196709,0.0004147051,0.00008757189,0.00007117724],"about_ca_topic_score_codex":0.0006990297,"about_ca_topic_score_gemma":0.0000081678845,"teacher_disagreement_score":0.06708608,"about_ca_system_score_codex":0.00016498544,"about_ca_system_score_gemma":0.00017823438,"threshold_uncertainty_score":0.9903396},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4210271902","doi":"10.21203/rs.3.rs-1281534/v1","title":"A Bibliometric Analysis of Medical Research Literature on Commonly Sold Herbal Medicines","year":2022,"lang":"en","type":"preprint","venue":"Research Square","topic":"Complementary and Alternative Medicine Studies","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Impact; Ottawa Hospital","funders":"McMaster University","keywords":"Scopus; Alternative medicine; Traditional medicine; Medicine; China; Bibliometrics; MEDLINE; Family medicine; Geography; Library science; Political science; Computer science","score_opus":0.24453593937811194,"score_gpt":0.5569769258757121,"score_spread":0.3124409864976002,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210271902","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92599046,0.029825363,0.000025678777,0.025801292,0.0006574972,0.0023893816,0.0010109484,0.00008266373,0.014216723],"genre_scores_gemma":[0.9787363,0.014757542,0.000060398797,0.00026170604,0.0016035608,0.00050468097,0.002235447,0.00006158537,0.0017788202],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9722162,0.0032909664,0.0011723181,0.0013361983,0.0206653,0.001319031],"domain_scores_gemma":[0.9858633,0.0076253605,0.00023773861,0.0018714421,0.0035678947,0.00083424826],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.020807402,0.0004924186,0.002157487,0.26153278,0.00067057536,0.00009618065,0.0017089052,0.00024180376,0.014166964],"category_scores_gemma":[0.011312945,0.0003638554,0.00080502394,0.30595383,0.0013725857,0.000055028733,0.0064740363,0.010661284,0.000022481421],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0030615688,0.00473776,0.6845843,0.011932598,0.018874144,0.0066970163,0.009044994,0.0001452915,0.00013155765,0.004106504,0.1645689,0.092115365],"study_design_scores_gemma":[0.0041672885,0.011262143,0.8225627,0.01268647,0.001664104,0.000038096467,0.014231846,0.0027196642,0.00016479059,0.0011785412,0.12877811,0.0005462568],"about_ca_topic_score_codex":0.0022677444,"about_ca_topic_score_gemma":0.00043678758,"teacher_disagreement_score":0.13797837,"about_ca_system_score_codex":0.0006700329,"about_ca_system_score_gemma":0.0012673673,"threshold_uncertainty_score":0.9998813},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4210401899","doi":"10.1016/j.jemep.2022.100758","title":"Impact of institutional affiliation bias on editorial publication decisions: A bibliometric analysis of three ophthalmology journals","year":2022,"lang":"en","type":"article","venue":"Ethics Medicine and Public Health","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":7,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; University of Toronto","funders":"","keywords":"Editorial board; Impact factor; Ophthalmology; Blinding; Medicine; Library science; Family medicine; MEDLINE; Political science; Law","score_opus":0.8774203146863786,"score_gpt":0.6741541490313949,"score_spread":0.20326616565498368,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210401899","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.937891,0.0034165285,0.00727487,0.04391784,0.005735944,0.00036288658,0.00043467982,0.000011021324,0.00095519715],"genre_scores_gemma":[0.99628794,0.0016997244,0.00020609314,0.00042750261,0.0012016852,0.000017873394,0.00011436304,0.0000063776274,0.000038446873],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9681299,0.0027218896,0.0025685122,0.0007722947,0.02511961,0.0006878121],"domain_scores_gemma":[0.9206057,0.05580285,0.0028196876,0.0011240112,0.018387783,0.0012599513],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.17317446,0.00015487855,0.00089281425,0.6918303,0.0008141091,0.0003113652,0.001401347,0.00024383207,0.0029371772],"category_scores_gemma":[0.4408161,0.000100440026,0.00028199368,0.8475218,0.00046645466,0.00052151957,0.00051811273,0.0015184936,0.0000050315543],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000110918416,0.0007726742,0.681527,0.000015417112,0.00041319238,0.0000037915233,0.0011420692,0.0012002592,0.000039862876,0.011958489,0.10166185,0.20115447],"study_design_scores_gemma":[0.0007902309,0.0029092752,0.96602243,0.000014086784,0.000031201813,0.000020499761,0.00044622083,0.0052137235,0.0000010836433,0.005976637,0.01848538,0.00008921777],"about_ca_topic_score_codex":0.0035055103,"about_ca_topic_score_gemma":0.00005930147,"teacher_disagreement_score":0.28449544,"about_ca_system_score_codex":0.00043482005,"about_ca_system_score_gemma":0.0059512635,"threshold_uncertainty_score":0.9996841},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4210511706","doi":"10.2196/preprints.31097","title":"Exploring the Shift in International Trends in Mobile Health Research From 2000 to 2020: Bibliometric Analysis (Preprint)","year":2021,"lang":"en","type":"preprint","venue":"","topic":"Mobile Health and mHealth Applications","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"mHealth; Preprint; The Internet; Popularity; Web of science; Health care; Telemedicine; Descriptive statistics; World Wide Web; Computer science; Data science; Library science; Geography; MEDLINE; Political science; Statistics; Mathematics","score_opus":0.3750816759979351,"score_gpt":0.5511372054892327,"score_spread":0.1760555294912976,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210511706","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9309678,0.003484794,0.001488576,0.043740075,0.0022339448,0.010315835,0.00042425274,0.00015600296,0.007188685],"genre_scores_gemma":[0.8359336,0.013228366,0.0014518171,0.0030515164,0.0006712924,0.14292976,0.0013612132,0.00005533488,0.001317087],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.98709065,0.0041121203,0.0031056467,0.0020297004,0.0016060168,0.0020558878],"domain_scores_gemma":[0.9911391,0.004056904,0.0005479635,0.0025356845,0.0006144669,0.00110589],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.013716361,0.00041648682,0.0012540615,0.1326516,0.0007916203,0.00012725608,0.0018614057,0.00047961037,0.01139225],"category_scores_gemma":[0.001048035,0.00035663173,0.00029423783,0.24130623,0.0000674914,0.00020321744,0.0034432632,0.0068903863,0.00059016835],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013429527,0.0006370915,0.38531008,0.00040729894,0.00020666009,0.000018605164,0.023204036,0.011932022,0.000003033925,0.0005024888,0.012970286,0.5646741],"study_design_scores_gemma":[0.0005267894,0.000054786917,0.91398215,0.0004740866,0.000020876687,1.7735455e-7,0.009237451,0.0012362166,0.0000019504757,0.00054501725,0.07368118,0.00023932302],"about_ca_topic_score_codex":0.20196405,"about_ca_topic_score_gemma":0.24312851,"teacher_disagreement_score":0.56443477,"about_ca_system_score_codex":0.004316616,"about_ca_system_score_gemma":0.005374301,"threshold_uncertainty_score":0.99988854},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4210843327","doi":"10.12688/f1000research.108772.1","title":"Scopus based bibliometric and scientometric analysis of occupational therapy publications from 2001 to 2020","year":2022,"lang":"en","type":"preprint","venue":"F1000Research","topic":"Delphi Technique in Research","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Bibliometrics; Occupational therapy; Publishing; Scientometrics; Medicine; Impact factor; Multidisciplinary approach; Library science; MEDLINE; Family medicine; Social science; Political science; Sociology; Psychiatry","score_opus":0.5239209140536851,"score_gpt":0.6060467038501911,"score_spread":0.08212578979650598,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210843327","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95083904,0.0032566076,0.0053642476,0.014579503,0.00035438855,0.004124773,0.0048320694,0.0003202673,0.016329084],"genre_scores_gemma":[0.9769619,0.003779661,0.012299228,0.00022234494,0.00021644711,0.00183054,0.0012493428,0.000053818618,0.0033867033],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98849237,0.0016796608,0.0006048512,0.0011044425,0.007218498,0.0009002075],"domain_scores_gemma":[0.991034,0.0043744883,0.00024042363,0.0012703993,0.0023089417,0.0007717927],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.013451254,0.00021752548,0.0005695721,0.27712426,0.0008318398,0.0008707464,0.0032322311,0.00029521715,0.027519247],"category_scores_gemma":[0.0099800015,0.00023493079,0.00027492628,0.5390044,0.00078918156,0.00019905297,0.002754655,0.001107772,0.00003734525],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019590648,0.00066531246,0.8530144,0.000042693388,0.0008272265,0.000006937154,0.0012164128,0.0017514931,0.0004116803,0.0013879838,0.110221796,0.030258158],"study_design_scores_gemma":[0.00021457774,0.000106924825,0.9272961,0.000018505061,0.00007997314,5.80983e-8,0.00053548906,0.004403143,0.000119663666,0.0009470246,0.06600985,0.00026870947],"about_ca_topic_score_codex":0.058607787,"about_ca_topic_score_gemma":0.0021571433,"teacher_disagreement_score":0.26188016,"about_ca_system_score_codex":0.00077769277,"about_ca_system_score_gemma":0.0043654433,"threshold_uncertainty_score":0.9983594},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4211001317","doi":"10.6007/ijarbss/v12-i1/12056","title":"A 59 Years (1962-2021) Bibliometric Analysis of Organizational Support Research Articles","year":2022,"lang":"en","type":"article","venue":"International Journal of Academic Research in Business and Social Sciences","topic":"Organizational and Employee Performance","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Thriving; Scopus; Organizational analysis; Publish or perish; Publishing; Bibliometrics; Library science; Yearbook; Citation; Publication; Management; Political science; Knowledge management; Sociology; Social science; Computer science; MEDLINE","score_opus":0.16415434745862767,"score_gpt":0.4483718198579301,"score_spread":0.28421747239930245,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4211001317","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9849482,0.0002867064,0.0007916039,0.013381955,0.00029460213,0.000052240735,0.000011757856,0.0000034401098,0.00022949053],"genre_scores_gemma":[0.99841225,0.00071727275,0.0005046706,0.00008312106,0.00020713841,0.0000034921693,0.0000032197645,0.0000035903495,0.00006526961],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99424696,0.00036261734,0.0005238208,0.00021819443,0.0043444573,0.00030392918],"domain_scores_gemma":[0.9963626,0.0005841779,0.00021831118,0.000064248205,0.0027129252,0.000057720128],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.010323239,0.000057930058,0.00019339667,0.04783639,0.000470808,0.00019771738,0.0021728072,0.000048214617,0.00022157766],"category_scores_gemma":[0.000663532,0.00005190923,0.000044599743,0.17796248,0.00057771336,0.0009455352,0.0010476931,0.00065113517,0.0000029468083],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000076746124,0.00032195015,0.8439268,0.00001742501,0.00025803834,0.00010412337,0.00934713,0.0057171,0.0027267807,0.09268432,0.0021006977,0.042718932],"study_design_scores_gemma":[0.00029869346,0.00012910899,0.9742006,0.00001579231,0.000008642799,0.00003650509,0.0009582123,0.0052491403,0.00016333333,0.015977371,0.0028781246,0.00008447505],"about_ca_topic_score_codex":0.00018357566,"about_ca_topic_score_gemma":0.0000084618505,"teacher_disagreement_score":0.13027385,"about_ca_system_score_codex":0.00024211725,"about_ca_system_score_gemma":0.0016026959,"threshold_uncertainty_score":0.9629556},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4213005437","doi":"10.1186/s41073-022-00121-1","title":"Characteristics of ‘mega’ peer-reviewers","year":2022,"lang":"en","type":"article","venue":"Research Integrity and Peer Review","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University; St. Michael's Hospital; University of Ottawa; Ottawa Public Health; University of Toronto; McGill University; Institute for Work & Health; Ottawa Hospital","funders":"","keywords":"Peer review; Receipt; Mega-; Control (management); Peer group; Psychology; Medical education; Medicine; Computer science; Social psychology; World Wide Web; Political science","score_opus":0.850221086539673,"score_gpt":0.6727948306867655,"score_spread":0.1774262558529075,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4213005437","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.074303545,0.59978276,0.0003899149,0.27034512,0.0015833759,0.003260995,0.0018694651,0.000046385616,0.04841843],"genre_scores_gemma":[0.5085676,0.40391073,0.0005851237,0.0014528848,0.00017508576,0.00024919797,0.00010429606,0.000028791574,0.0849263],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9455048,0.0039571417,0.0014733401,0.0008800773,0.04729683,0.0008878273],"domain_scores_gemma":[0.9345699,0.0145398965,0.00061864237,0.002243379,0.046927497,0.0011007093],"candidate_categories":["metaresearch","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.27370593,0.00015107995,0.0008125117,0.014870534,0.0006951483,0.000750122,0.0031018483,0.00007152838,0.008623855],"category_scores_gemma":[0.29525575,0.000102913946,0.00025328275,0.09895887,0.00046855534,0.0002879187,0.0028723339,0.002721567,0.00019163216],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000015664444,0.00015340834,0.009204247,0.00066258654,0.000012674318,0.000017793036,0.00007860716,1.3137237e-7,0.00003621878,0.005264833,0.5241783,0.4603755],"study_design_scores_gemma":[0.00011579502,0.00026985566,0.011330526,0.00025593693,0.000007696708,0.00002553641,0.00020258597,0.0001376688,0.00002366545,0.0040096412,0.9835073,0.00011380342],"about_ca_topic_score_codex":0.00028959915,"about_ca_topic_score_gemma":0.0000070037213,"teacher_disagreement_score":0.46026167,"about_ca_system_score_codex":0.00014950777,"about_ca_system_score_gemma":0.00059349544,"threshold_uncertainty_score":0.9995792},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W4213241381","doi":"10.1016/j.joi.2022.101256","title":"Female inventors over time: Factors affecting female Inventors’ innovation performance","year":2022,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Innovation Policy and R&D","field":"Economics, Econometrics and Finance","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"HEC Montréal","funders":"HEC Montréal","keywords":"Business; Production (economics); European patent office; Panel data; Demographic economics; Industrial organization; Economics; International trade; Microeconomics","score_opus":0.0561466103949233,"score_gpt":0.24998779703667268,"score_spread":0.19384118664174937,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4213241381","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98480004,0.0001922441,0.00033342006,0.00008556531,0.0014394544,0.00012154307,0.000046937737,0.000017948962,0.012962867],"genre_scores_gemma":[0.99656403,0.000053623844,0.000313246,0.0007563661,0.00027360016,0.0000032577498,0.00002596528,0.000024282526,0.0019856407],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9975646,0.000020208183,0.0016970881,0.00014500042,0.00024290556,0.00033018505],"domain_scores_gemma":[0.99669623,0.000114311,0.002685862,0.0001913587,0.0002425259,0.00006971365],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0025130513,0.00017403229,0.00041456393,0.004160278,0.0004223343,0.000093627204,0.00036739637,0.00009485387,0.0011358225],"category_scores_gemma":[0.001075487,0.0001813779,0.00016188122,0.0072349464,0.00003610332,0.0010748603,0.00014950894,0.0006463935,0.00009102733],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000050872404,0.00017509496,0.94185436,0.00008826686,0.000119130804,0.000005170022,0.0039874776,0.0022497089,0.00006256653,0.041502047,0.0077623874,0.0021429132],"study_design_scores_gemma":[0.003370756,0.0018943776,0.5782592,0.00006794091,0.000032537424,0.00014184767,0.003262587,0.005646558,0.0032594674,0.0034871816,0.3994334,0.0011441016],"about_ca_topic_score_codex":0.000034886434,"about_ca_topic_score_gemma":5.8264874e-7,"teacher_disagreement_score":0.391671,"about_ca_system_score_codex":0.0005711061,"about_ca_system_score_gemma":0.00011377651,"threshold_uncertainty_score":0.99977726},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4220683548","doi":"10.3390/jrfm15030136","title":"Influence of Senior Executives Characteristics on Corporate Environmental Disclosures: A Bibliometric Analysis","year":2022,"lang":"en","type":"article","venue":"Journal of risk and financial management","topic":"Corporate Social Responsibility Reporting","field":"Business, Management and Accounting","cited_by":50,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Hubei University","keywords":"Accounting; Corporate governance; Business; Thematic analysis; Sustainability; Environmental reporting; Sustainability reporting; Political science; Public relations; Qualitative research; Sociology; Corporate social responsibility; Finance; Social science","score_opus":0.012093439378150347,"score_gpt":0.21994334129179918,"score_spread":0.20784990191364883,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220683548","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99876726,0.00012904721,0.00054022035,0.000035256897,0.00014130409,0.00017710675,0.000036611087,0.000008088745,0.00016512668],"genre_scores_gemma":[0.9991394,0.00036803546,0.00010241809,0.00012467435,0.00017689151,0.0000062402946,0.000007976084,0.000011712939,0.00006262285],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9979556,0.000056034933,0.0008608314,0.00021063424,0.00072985975,0.00018704507],"domain_scores_gemma":[0.99613434,0.00009738392,0.003474183,0.00017676935,0.000093804374,0.000023501758],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0015821448,0.00015765651,0.0004641043,0.01860118,0.00031003752,0.00008529308,0.0002175657,0.000029137236,0.000042869113],"category_scores_gemma":[0.0004627168,0.00014589811,0.00023154393,0.025964513,0.00009087783,0.00036222846,0.00040552355,0.0002636477,0.0000029776593],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000612205,0.00034932894,0.889186,0.00008092202,0.00018999161,0.00023353819,0.00020777837,0.0027810743,0.0001666239,0.0013957986,0.000067648594,0.10472904],"study_design_scores_gemma":[0.00042472073,0.00010484215,0.9881549,0.00001942777,0.0007764075,0.0000035020812,0.0006832872,0.00015869515,0.0000105936515,0.0020753583,0.0074437875,0.00014449787],"about_ca_topic_score_codex":0.00006523274,"about_ca_topic_score_gemma":0.000008765671,"teacher_disagreement_score":0.104584545,"about_ca_system_score_codex":0.00007808432,"about_ca_system_score_gemma":0.000026743239,"threshold_uncertainty_score":0.99473906},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4220686081","doi":"10.1061/9780784483961.070","title":"Digital Twin in the Architecture, Engineering, and Construction Industry: A Bibliometric Review","year":2022,"lang":"en","type":"review","venue":"Construction Research Congress 2022","topic":"Digital Transformation in Industry","field":"Engineering","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of New Brunswick","funders":"","keywords":"Digitization; Context (archaeology); Interoperability; Computer science; Architecture; Building information modeling; Scopus; Construction industry; Data science; Citation analysis; Engineering management; Citation; Engineering; Knowledge management; World Wide Web; Construction engineering; Telecommunications","score_opus":0.0809509736026532,"score_gpt":0.3509509599944974,"score_spread":0.2699999863918442,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220686081","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000029038558,0.974704,0.00003940255,0.00007291187,0.0012715976,0.001939829,0.00046152566,0.0002170839,0.021264613],"genre_scores_gemma":[0.000021197711,0.9978004,0.00009745314,0.000020290096,0.00015909979,0.0013945346,0.0002647126,0.000094018025,0.00014825675],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9956829,0.00047437867,0.0010624939,0.00052586885,0.0015289034,0.00072543335],"domain_scores_gemma":[0.9974871,0.0013946235,0.00014976585,0.0006584961,0.00011476381,0.00019524722],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0019937644,0.0005347335,0.0010990433,0.022138862,0.00021782168,0.0007426023,0.00087235036,0.0005764376,0.001185008],"category_scores_gemma":[0.0009227234,0.0004535229,0.00023469808,0.050692134,0.0005599065,0.000658798,0.0002454362,0.0071240626,0.00004008308],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000015616164,0.000013484721,0.000020621108,0.020464431,0.00008693304,0.000038257167,0.00002078175,0.00006957476,3.7021668e-8,0.00097077026,0.0048610074,0.97345257],"study_design_scores_gemma":[0.00018394225,0.000036833513,0.000006448538,0.008322905,0.00006744427,0.0020704037,0.00028008007,0.00007346124,3.1894967e-7,0.00009510055,0.98846066,0.00040239695],"about_ca_topic_score_codex":0.0000051772126,"about_ca_topic_score_gemma":0.0000014394031,"teacher_disagreement_score":0.98359966,"about_ca_system_score_codex":0.00041335818,"about_ca_system_score_gemma":0.00030612457,"threshold_uncertainty_score":0.9997916},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4220858960","doi":"10.1186/s42269-022-00740-0","title":"Assessment of wetland potential and bibliometric review: a critical analysis of the Ramsar sites of India","year":2022,"lang":"en","type":"article","venue":"Bulletin of the National Research Centre/Bulletin of the National Research Center","topic":"Flood Risk Assessment and Management","field":"Environmental Science","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Jadavpur University","keywords":"Wetland; Ramsar site; Geography; Context (archaeology); Period (music); Ecology; Archaeology","score_opus":0.05584845189360171,"score_gpt":0.4049330084756946,"score_spread":0.34908455658209286,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220858960","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5580239,0.005301616,0.000047506215,0.30469236,0.0006250352,0.007385052,0.0037058021,0.000019230738,0.12019946],"genre_scores_gemma":[0.9949613,0.0014548241,0.00045391405,0.00031213675,0.000031689655,0.00006259112,0.00002102588,0.000014578132,0.002687939],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9740923,0.0033648193,0.0009425296,0.00045407348,0.020618461,0.0005278219],"domain_scores_gemma":[0.99223113,0.0035371163,0.00053672143,0.0004667593,0.00311819,0.00011007554],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.023601463,0.00017060473,0.00045654745,0.0039626425,0.0005503695,0.000029315534,0.002247033,0.000071676935,0.0152352415],"category_scores_gemma":[0.014827699,0.00010583513,0.0004991086,0.017910982,0.0020289572,0.000036376987,0.0057835234,0.00088392047,0.000011308923],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023750053,0.0024653536,0.13633794,0.00075702305,0.0008621028,0.0000010762828,0.000101426944,0.0042371326,0.007633972,0.010514222,0.8363467,0.0005055706],"study_design_scores_gemma":[0.001025541,0.00018321985,0.7523589,0.0003024075,0.0001733008,0.000002618375,0.0001739807,0.0014909903,0.0019405433,0.0014775321,0.24074124,0.00012970963],"about_ca_topic_score_codex":0.0006607157,"about_ca_topic_score_gemma":0.000032761476,"teacher_disagreement_score":0.616021,"about_ca_system_score_codex":0.00078040466,"about_ca_system_score_gemma":0.00037741367,"threshold_uncertainty_score":0.99347085},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4220877524","doi":"10.1007/s11192-022-04338-1","title":"Interdisciplinarity and impact: the effects of the citation time window","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":42,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Université du Québec à Montréal","funders":"National Planning Office of Philosophy and Social Science","keywords":"Citation; Impact factor; Window (computing); Citation analysis; Scientometrics; Computer science; Bibliometrics; Data science; Citation impact; Statistics; Data mining; Mathematics; Library science; Political science; World Wide Web; Law","score_opus":0.2744588650714582,"score_gpt":0.5399536004006232,"score_spread":0.265494735329165,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220877524","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9933278,0.0011457123,0.0004284798,0.0014297593,0.0013937205,0.0005191043,0.000058800422,0.000011354456,0.0016853003],"genre_scores_gemma":[0.9969249,0.000026688434,0.00007250484,0.0001155192,0.000029020994,0.00002020112,0.0000015103891,0.0000072704447,0.0028024302],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98200136,0.00076254556,0.0005336002,0.000574779,0.01561731,0.0005104088],"domain_scores_gemma":[0.98061275,0.015337603,0.00050726935,0.0012520904,0.0020106167,0.00027963938],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.039560106,0.00013243733,0.00024681468,0.06512869,0.0014785513,0.0013092889,0.0041198824,0.000044008237,0.0003657265],"category_scores_gemma":[0.09398838,0.000060919396,0.0002330613,0.54746395,0.00057902344,0.00035674177,0.0058919247,0.00045579096,0.000040432344],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000048550923,0.0004722862,0.69177157,0.000030287503,0.00004792501,0.000006802905,0.0034699466,0.00058611605,0.010552847,0.001568631,0.05059177,0.2408533],"study_design_scores_gemma":[0.0003983947,0.0003706958,0.9784735,0.0000040565405,0.000011274314,0.000013406973,0.0005296089,0.0066401265,0.0010800678,0.008498293,0.0038697487,0.00011083769],"about_ca_topic_score_codex":0.000051014402,"about_ca_topic_score_gemma":0.0000029873113,"teacher_disagreement_score":0.48233527,"about_ca_system_score_codex":0.0002523708,"about_ca_system_score_gemma":0.00023502152,"threshold_uncertainty_score":0.99982136},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4220963441","doi":"10.1016/j.ncrna.2022.03.001","title":"Hotspots and development frontiers of circRNA based on bibliometric analysis","year":2022,"lang":"en","type":"review","venue":"Non-coding RNA Research","topic":"Circular RNAs in diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Web of science; Citation analysis; Circular RNA; Data science; China; Citation; Library science; Biology; Political science; Computer science; microRNA; MEDLINE; Genetics; Gene","score_opus":0.14324621472133983,"score_gpt":0.4198828508526507,"score_spread":0.27663663613131084,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220963441","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018519752,0.99544275,0.0007508953,0.000006015106,0.00013058155,0.00059288647,0.000090860114,0.000008574315,0.0011254557],"genre_scores_gemma":[0.012435813,0.98569167,0.00069635996,0.000015126667,0.000085245185,0.00021721935,0.0005566197,0.00006808712,0.0002338375],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9961558,0.0005313063,0.00055039837,0.0008936481,0.0013195326,0.00054933014],"domain_scores_gemma":[0.99813056,0.0002794332,0.00023483443,0.0009374391,0.00018616943,0.00023158579],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0023138914,0.0003359867,0.0010841207,0.060778912,0.00026816176,0.00007810963,0.00081978703,0.00028627214,0.00017178511],"category_scores_gemma":[0.0010492847,0.00032357534,0.0005198382,0.061980907,0.00018259497,0.0000033191964,0.00060477504,0.0005608132,0.000007976345],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000046618457,0.00019639921,0.0012385839,0.00425205,0.001680881,0.000036424146,0.0000150724045,0.000039825463,0.00023150632,0.0000025934435,0.004263923,0.9879961],"study_design_scores_gemma":[0.00031986472,0.00019568385,0.0020968644,0.00060326693,0.00065033603,0.0000035042829,0.000037236863,0.00008852929,0.00066319463,0.000001873148,0.9949163,0.00042333046],"about_ca_topic_score_codex":0.0000252511,"about_ca_topic_score_gemma":0.000004194643,"teacher_disagreement_score":0.9906524,"about_ca_system_score_codex":0.00021602484,"about_ca_system_score_gemma":0.0012229712,"threshold_uncertainty_score":0.9999216},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4223486970","doi":"10.1080/21645698.2022.2038525","title":"Genetically modified foods: bibliometric analysis on consumer perception and preference","year":2022,"lang":"en","type":"article","venue":"GM crops & food","topic":"Genetically Modified Organisms Research","field":"Agricultural and Biological Sciences","cited_by":75,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Agriculture and Agri-Food Canada","funders":"","keywords":"Per capita; Preference; Consumption (sociology); Government (linguistics); Perception; Marketing; Business; Genetically modified food; Public policy; Public economics; Genetically modified organism; Agricultural economics; Economics; Economic growth; Psychology; Sociology; Social science","score_opus":0.06917925277951521,"score_gpt":0.26017339313580756,"score_spread":0.19099414035629236,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4223486970","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99735445,0.00024546648,0.00006965765,0.00067007856,0.000038315364,0.00027545076,0.00012257612,0.00005762563,0.0011663882],"genre_scores_gemma":[0.998988,0.0001833858,0.00026693416,0.00021208098,0.000044749762,0.00007111789,0.00006240212,0.000002184008,0.00016919505],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9977115,0.00025885724,0.00023215488,0.0005828984,0.000781408,0.00043322166],"domain_scores_gemma":[0.99925125,0.00023036945,0.000052042462,0.00014461346,0.000104406834,0.00021733438],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00036815513,0.00016606327,0.000236797,0.0032112675,0.00050118566,0.00016423003,0.00040077357,0.000078151796,0.003433934],"category_scores_gemma":[0.00009996556,0.00007729125,0.00010768418,0.046488203,0.00012057936,0.000041362,0.0004093731,0.00028918486,0.000060412218],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020016545,0.00054826925,0.029185379,0.000012193145,0.00036649703,0.000008624265,0.00019444502,0.00054867606,0.41652286,0.0009934427,0.00039374342,0.5510257],"study_design_scores_gemma":[0.00019544523,0.0024050719,0.99255025,0.0000013791828,0.000097435804,0.000005755137,0.00019506385,0.0006328278,0.0011161376,0.0004268719,0.002156661,0.00021708453],"about_ca_topic_score_codex":0.0002268656,"about_ca_topic_score_gemma":0.00022294324,"teacher_disagreement_score":0.9633649,"about_ca_system_score_codex":0.0000481114,"about_ca_system_score_gemma":0.000016121801,"threshold_uncertainty_score":0.99747705},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4223542744","doi":"10.33263/lianbs121.027","title":"A Bibliometric Analysis of Moderna mRNA-1273 Vaccine for COVID-19","year":2022,"lang":"en","type":"article","venue":"Letters in Applied NanoBioScience","topic":"SARS-CoV-2 and COVID-19 Research","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Allergy and Infectious Diseases; National Institutes of Health; Moderna; School of Medicine, Emory University; Institut National de la Santé et de la Recherche Médicale; Imperial College London; Vanderbilt University Medical Center; Vanderbilt University; McMaster University; Irving Medical Center, Columbia University; Emory University; Brigham and Women's Hospital","keywords":"Scopus; Library science; Index (typography); Coronavirus disease 2019 (COVID-19); Bibliometrics; Medicine; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Family medicine; MEDLINE; Geography; Political science; Computer science; Law; World Wide Web; Internal medicine; Infectious disease (medical specialty)","score_opus":0.04808331648198753,"score_gpt":0.3591258333526371,"score_spread":0.31104251687064954,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4223542744","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9795109,0.00015880198,0.014912008,0.0036584034,0.000101989535,0.0010277643,0.000060268783,0.000049873775,0.00051996094],"genre_scores_gemma":[0.8389566,0.0000041008707,0.0014068342,0.1591547,0.00002652733,0.00041483366,0.0000081732205,0.000015003627,0.00001323006],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9970909,0.00004864894,0.00047881517,0.0006962843,0.001110313,0.00057503785],"domain_scores_gemma":[0.9985977,0.0004911819,0.00017676497,0.00059051544,0.00004916327,0.000094717994],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0015839918,0.00017330427,0.0005639137,0.081736915,0.00022101116,0.000029066212,0.000724068,0.0000515439,0.00005196351],"category_scores_gemma":[0.00041056224,0.00016371017,0.00021662748,0.23791312,0.00015846004,0.000055790973,0.00031408152,0.00026891922,0.0000038013104],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005406129,0.00028360443,0.089683264,0.000116822484,0.00011205193,0.000035198547,0.00029735107,0.00067320146,0.9036931,0.0003602597,0.0014354886,0.002769068],"study_design_scores_gemma":[0.02052781,0.0014927975,0.13999026,0.000041934985,0.0015206935,0.000078320336,0.00092899456,0.16156368,0.39857015,0.0011046261,0.27267718,0.0015035336],"about_ca_topic_score_codex":0.00030244593,"about_ca_topic_score_gemma":0.000028126395,"teacher_disagreement_score":0.5051229,"about_ca_system_score_codex":0.00051335496,"about_ca_system_score_gemma":0.000370735,"threshold_uncertainty_score":0.9286708},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4223594566","doi":"10.1177/26314541221078909","title":"Research Landscape of Multigenerational Workforce Literature: A Bibliographic Coupling and Co-Citation Analysis","year":2022,"lang":"en","type":"article","venue":"NHRD Network Journal","topic":"Human Resource and Talent Management","field":"Business, Management and Accounting","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of New Brunswick","funders":"","keywords":"Bibliographic coupling; Workforce; Scopus; Extant taxon; Citation; Relevance (law); Diversity (politics); Citation analysis; Sample (material); Sociology; Facet (psychology); Psychology; Political science; Library science; Computer science; Social psychology; MEDLINE","score_opus":0.045354678874393935,"score_gpt":0.3009868133700668,"score_spread":0.25563213449567285,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4223594566","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.988926,0.004481668,0.0009981839,0.00035949302,0.00019156527,0.0001744479,0.000003667599,0.000024651921,0.0048403456],"genre_scores_gemma":[0.99680084,0.00040727068,0.00016955935,0.000296631,0.0018732044,0.000019973864,0.00011965556,0.000015421838,0.00029744027],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981097,0.00006058149,0.00036145293,0.00021211337,0.0009237156,0.00033243294],"domain_scores_gemma":[0.9991386,0.00010718332,0.00029254516,0.0001386321,0.00029760468,0.000025415911],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.002829852,0.00011531156,0.00020214738,0.0058791013,0.0013884897,0.0007434366,0.00022109837,0.00003572772,0.0008060966],"category_scores_gemma":[0.000019718756,0.000109095294,0.0001561414,0.012628549,0.00005702592,0.0004329386,0.00019673494,0.00054452237,0.0000066120515],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011810482,0.000087659966,0.44168055,0.000050160397,0.00046264505,0.000037835624,0.00018608979,0.49660823,0.000027620037,0.008967003,0.049872607,0.0019014993],"study_design_scores_gemma":[0.0018569044,0.000106260035,0.24802282,0.00014828422,0.00073429826,0.000021189127,0.0019802498,0.55660313,0.000004431282,0.0051107868,0.18495154,0.00046008927],"about_ca_topic_score_codex":0.000013461455,"about_ca_topic_score_gemma":0.000022036167,"teacher_disagreement_score":0.19365774,"about_ca_system_score_codex":0.00001764648,"about_ca_system_score_gemma":0.000013936857,"threshold_uncertainty_score":0.99991155},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4224260724","doi":"10.1177/00031348221091965","title":"Vascular Surgery Research in the Coronavirus Disease 2019 Pandemic: A Sex-Based Bibliometric Analysis","year":2022,"lang":"en","type":"review","venue":"The American Surgeon","topic":"COVID-19 and healthcare impacts","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre Hospitalier de l’Université de Montréal; University of British Columbia; University of Toronto; Université de Montréal","funders":"","keywords":"Medicine; Pandemic; MEDLINE; Scopus; Vascular surgery; Coronavirus disease 2019 (COVID-19); Bibliometrics; Family medicine; English language; Disease; Internal medicine; Library science; Cardiac surgery; Infectious disease (medical specialty)","score_opus":0.47291877864158804,"score_gpt":0.5460853306450997,"score_spread":0.07316655200351163,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4224260724","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006858569,0.9890207,0.000010650004,0.0016940418,0.0001104175,0.001937495,0.00021826856,0.00007237204,0.00007747285],"genre_scores_gemma":[0.011599142,0.9836107,0.000003014665,0.0029273177,0.00020708187,0.00059966947,0.00077818724,0.000088189096,0.00018670749],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9862515,0.007830152,0.001064494,0.0008226716,0.0028156224,0.0012155881],"domain_scores_gemma":[0.9648296,0.03105821,0.0007206967,0.0026791992,0.00014747173,0.0005647944],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.019035086,0.0005179022,0.0038180798,0.09863785,0.00040140463,0.00009132634,0.0008902223,0.00010601,0.0005961633],"category_scores_gemma":[0.0029270377,0.0002816211,0.002492877,0.372976,0.0005859277,0.000046156492,0.00022536606,0.0023767499,0.000092875205],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015878274,0.00023275358,0.05919061,0.0037430369,0.0006970673,0.00044220206,0.00006749406,0.000010103924,3.760561e-8,0.000001910366,0.0026258198,0.93283015],"study_design_scores_gemma":[0.00014390702,0.00012237139,0.12201678,0.0009473309,0.003635012,0.000016512315,0.00013909586,0.000028398083,1.5124519e-8,0.0000072892053,0.8726824,0.00026091808],"about_ca_topic_score_codex":0.036439788,"about_ca_topic_score_gemma":0.00033293414,"teacher_disagreement_score":0.93256927,"about_ca_system_score_codex":0.0010952471,"about_ca_system_score_gemma":0.006600599,"threshold_uncertainty_score":0.9999636},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4224274019","doi":"10.1111/1911-3846.12779","title":"<i>Contemporary Accounting Research</i>: A Retrospective between 1984 and 2021 using Bibliometric Analysis*","year":2022,"lang":"en","type":"article","venue":"Contemporary Accounting Research","topic":"Auditing, Earnings Management, Governance","field":"Business, Management and Accounting","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Accounting research; Accounting; Audit; Subject (documents); Bibliometrics; Scope (computer science); Bibliographic coupling; Quality (philosophy); Diversity (politics); Citation; Library science; Regional science; Political science; Sociology; Computer science; Business","score_opus":0.09424337518649772,"score_gpt":0.34611079883870577,"score_spread":0.25186742365220804,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4224274019","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92007095,0.004481215,0.0032382247,0.0021946859,0.00043798494,0.0019469932,0.00009749063,0.0003479612,0.067184485],"genre_scores_gemma":[0.9930468,0.000056484783,0.00028799917,0.00053097226,0.0033705703,0.00022319745,0.0002678014,0.0002272476,0.0019889243],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9855826,0.00094509317,0.001544201,0.002555642,0.006903442,0.0024690046],"domain_scores_gemma":[0.9840032,0.0018215257,0.009238345,0.0017064019,0.0030795997,0.00015096465],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.035219643,0.0007247609,0.0013412634,0.10527919,0.007402085,0.0041120593,0.0025687155,0.0002394661,0.0010326325],"category_scores_gemma":[0.024392337,0.0008274376,0.000413221,0.2651235,0.0009010912,0.0063857078,0.010339887,0.0047374424,0.0002854538],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013986435,0.00015742405,0.9427964,0.00031122315,0.00068538974,0.00019032182,0.00023948778,0.00016067135,0.00066943414,0.0016249395,0.040615752,0.012409104],"study_design_scores_gemma":[0.001756103,0.00010555373,0.5917467,0.00024093158,0.00017816495,0.000009285027,0.0042278385,0.0068836077,0.00004469133,0.0024194329,0.39114583,0.0012418595],"about_ca_topic_score_codex":0.012522339,"about_ca_topic_score_gemma":0.000074989526,"teacher_disagreement_score":0.3510497,"about_ca_system_score_codex":0.0008561086,"about_ca_system_score_gemma":0.00068331446,"threshold_uncertainty_score":0.99988055},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4224277908","doi":"10.2196/33842","title":"Global Scientific Research Landscape on Medical Informatics From 2011 to 2020: Bibliometric Analysis","year":2022,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Health informatics; Informatics; Public health informatics; Bibliometrics; Health care; Data science; Health Administration Informatics; The Internet; Library science; Medicine; Computer science; Public health; Political science; World Wide Web; Health policy; International health; Nursing","score_opus":0.1692497379824417,"score_gpt":0.4944459984231761,"score_spread":0.3251962604407344,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4224277908","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97964793,0.00014098632,0.0015595438,0.009444728,0.0013660582,0.00082647667,0.00013412937,0.000113315335,0.0067668143],"genre_scores_gemma":[0.98820597,0.000117334785,0.0012855545,0.0080610225,0.0006200728,0.00029521267,0.00088416267,0.000015891472,0.0005147854],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98800755,0.00022217972,0.0018209243,0.00023857811,0.008857816,0.0008529741],"domain_scores_gemma":[0.99513924,0.000923634,0.00023250446,0.00087826926,0.00071358465,0.0021127644],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0062973914,0.00021199846,0.0005425047,0.026094973,0.0008473804,0.00020243016,0.0008723072,0.00032368425,0.028677898],"category_scores_gemma":[0.0028201973,0.00017816472,0.00020385293,0.13017939,0.00030662224,0.00025324812,0.0006302006,0.0014705849,0.0030969507],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031754372,0.00095210073,0.06474955,0.00021071757,0.0003227057,0.000066654866,0.020692976,0.00027656165,0.0000010714784,0.00047851133,0.6133844,0.2985472],"study_design_scores_gemma":[0.0009712787,0.0037151938,0.07684287,0.00042586963,0.0005194295,0.00018287993,0.12650855,0.31661904,0.00016190908,0.0017943089,0.4712994,0.0009592677],"about_ca_topic_score_codex":0.00066316844,"about_ca_topic_score_gemma":0.00022964244,"teacher_disagreement_score":0.31634247,"about_ca_system_score_codex":0.0005600946,"about_ca_system_score_gemma":0.0023714988,"threshold_uncertainty_score":0.99767923},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4224316587","doi":"10.1136/bmjopen-2021-059445","title":"Assessing the impact of predatory journals on policy and guidance documents: a cross-sectional study protocol","year":2022,"lang":"en","type":"article","venue":"BMJ Open","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Memorial University of Newfoundland; Ottawa Hospital; Dalhousie University; University of Ottawa","funders":"Ottawa Hospital Anesthesia Alternate Funds Association; Dalhousie University; University of Ottawa","keywords":"Medicine; Cross-sectional study; Protocol (science); Environmental health; Alternative medicine; Pathology","score_opus":0.7721978534082363,"score_gpt":0.7632492341181518,"score_spread":0.008948619290084503,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4224316587","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8438905,0.000018300334,0.00003662629,0.0002560812,0.000094362986,0.14098231,0.000053794138,0.000004239762,0.014663785],"genre_scores_gemma":[0.9287419,5.1234707e-7,0.000062298255,0.00008125259,0.000082332685,0.068542145,3.7010975e-7,0.000006842877,0.002482351],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98748684,0.0012989913,0.00089715305,0.0004902836,0.009500004,0.00032670426],"domain_scores_gemma":[0.99460614,0.002447668,0.00068932684,0.0008009519,0.0012866196,0.00016931782],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.07873661,0.00010160687,0.00023495252,0.014013605,0.0010468466,0.0136302365,0.0038981538,0.000025566456,0.0023080227],"category_scores_gemma":[0.025896512,0.000052136114,0.00010406357,0.06517633,0.00012291798,0.0014818667,0.0048836237,0.0002773885,0.000014760508],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019682094,0.00044881352,0.9841341,0.000002364014,0.000027671844,0.0000062123586,0.00016033593,0.0006872142,0.000097232434,0.00007267324,0.00555147,0.008615096],"study_design_scores_gemma":[0.0011010548,0.0010154115,0.99098027,0.0000060166067,7.941286e-7,0.000019711413,0.00064750883,0.00061726297,0.0000246227,0.0025387972,0.0029780066,0.00007054338],"about_ca_topic_score_codex":0.001169105,"about_ca_topic_score_gemma":0.0000070373944,"teacher_disagreement_score":0.08485139,"about_ca_system_score_codex":0.00030446443,"about_ca_system_score_gemma":0.0012613776,"threshold_uncertainty_score":0.998604},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"not_applicable","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","scholarly_communication","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4225140398","doi":"10.1007/s11024-022-09468-7","title":"China’s Research Evaluation Reform: What are the Consequences for Global Science?","year":2022,"lang":"en","type":"article","venue":"Minerva","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":57,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bureau de Coopération Interuniversitaire; Université du Québec à Montréal; Université de Montréal","funders":"Canada Research Chairs","keywords":"China; Science Citation Index; Political science; Science policy; Coronavirus disease 2019 (COVID-19); Higher education; Citation; Citation impact; Position (finance); Public administration; Economic growth; Economics; Law","score_opus":0.8411194510935742,"score_gpt":0.6756913006516972,"score_spread":0.16542815044187698,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4225140398","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92416805,0.0048450287,0.00023355719,0.050248224,0.00214486,0.0014694526,0.00010732408,0.000020675838,0.016762841],"genre_scores_gemma":[0.9964975,0.000073385396,0.00022345674,0.00026381438,0.00011513169,0.00039699834,0.0000055289674,0.000005295293,0.0024188582],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9625664,0.0010594776,0.00054006296,0.00096411636,0.03383564,0.0010343179],"domain_scores_gemma":[0.98636407,0.002234629,0.0002608425,0.0012365162,0.009595339,0.00030859647],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.20777427,0.00011022779,0.0001918551,0.01718655,0.004162999,0.0071831825,0.0061384374,0.0000393037,0.0008530358],"category_scores_gemma":[0.06150865,0.00005952724,0.00012809566,0.24621883,0.0018242926,0.0015539818,0.0022655565,0.00035734338,0.00008656083],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009742548,0.00021581977,0.013198419,0.000005593078,0.000015239875,0.000010773091,0.00067451893,0.0007099098,0.0022863643,0.02067651,0.054510564,0.90759885],"study_design_scores_gemma":[0.0012747489,0.0010140084,0.35938197,0.000018920085,0.000017577846,0.00009145323,0.07495495,0.12306834,0.0012091561,0.2894324,0.14915444,0.00038201467],"about_ca_topic_score_codex":0.000883209,"about_ca_topic_score_gemma":0.00015690073,"teacher_disagreement_score":0.90721685,"about_ca_system_score_codex":0.0017660822,"about_ca_system_score_gemma":0.001920539,"threshold_uncertainty_score":0.99923885},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4225293074","doi":"10.3389/fsurg.2022.881555","title":"Mapping Knowledge Structure and Themes Trends of Post-operative Rehabilitation of Hip Fractures in the Elderly: A Bibliometrics and Visualization Study","year":2022,"lang":"en","type":"article","venue":"Frontiers in Surgery","topic":"Hip and Femur Fractures","field":"Medicine","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Rehabilitation; Bibliometrics; Medicine; Hip fracture; Web of science; Library science; Physical therapy; Osteoporosis; Meta-analysis; Computer science","score_opus":0.018218843561246728,"score_gpt":0.3217526115075404,"score_spread":0.3035337679462937,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4225293074","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9937504,0.005277554,0.0001121641,0.00024610484,0.00018348945,0.00031812987,0.000023917995,0.000005623042,0.00008260964],"genre_scores_gemma":[0.99945915,0.00008241876,0.00027689253,0.00007428057,0.000020254858,0.000023112123,0.000035723762,0.000009612294,0.00001856398],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99867374,0.00042687985,0.00035568324,0.00017770486,0.0002587732,0.000107194945],"domain_scores_gemma":[0.9987173,0.0009046693,0.00013645603,0.00013804443,0.000081022576,0.000022491378],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00087055773,0.00010223535,0.00038550285,0.0070311395,0.000057776895,0.000010414805,0.00005144959,0.00005113782,0.000036403184],"category_scores_gemma":[0.00075195276,0.00007232812,0.000038633592,0.006815724,0.00008809733,0.00008502937,0.000038050068,0.0001936446,1.4807619e-8],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001923475,0.0003268168,0.85899997,0.00010312629,0.000035977082,0.000006064657,0.0783062,0.000023245078,0.00028346822,0.000019640567,0.0027571218,0.058946054],"study_design_scores_gemma":[0.00058250054,0.0005540235,0.9143418,0.000037600585,0.000022458638,0.0000046170962,0.08321611,0.00018478025,0.00012087711,0.00019456995,0.00067441777,0.00006624845],"about_ca_topic_score_codex":0.000084491105,"about_ca_topic_score_gemma":0.00004054963,"teacher_disagreement_score":0.058879808,"about_ca_system_score_codex":0.000033055498,"about_ca_system_score_gemma":0.00005400473,"threshold_uncertainty_score":0.62738127},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4225336645","doi":"10.1109/access.2022.3160457","title":"Educational Data Mining: A Bibliometric Analysis of an Emerging Field","year":2022,"lang":"en","type":"article","venue":"IEEE Access","topic":"Online Learning and Analytics","field":"Computer Science","cited_by":43,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Multidisciplinary approach; Data science; Field (mathematics); Extant taxon; Bibliometrics; Computer science; Educational data mining; Analytics; Library science; Social science; Sociology; Mathematics","score_opus":0.1044720757957683,"score_gpt":0.4291229528722822,"score_spread":0.3246508770765139,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4225336645","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9444125,0.00011897426,0.048991043,0.0053965645,0.0004668987,0.000031269563,0.00003537266,0.000044532975,0.0005028451],"genre_scores_gemma":[0.9915003,0.0000067408428,0.007741833,0.0003323525,0.00008199273,0.0000034399368,0.000070909344,0.0000037681743,0.00025865008],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988706,0.000072622766,0.00019323731,0.0003245011,0.00041001776,0.00012901843],"domain_scores_gemma":[0.998462,0.0002627203,0.00014869003,0.0010064186,0.00006807368,0.000052072024],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00045081027,0.00005833298,0.0001447774,0.029455524,0.00014755728,0.00015952415,0.0035923324,0.000014131723,0.00023790031],"category_scores_gemma":[0.00014145876,0.000060976334,0.00004952984,0.14793618,0.000009722477,0.0007767221,0.0009427724,0.000116432944,0.0000012597534],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001327112,0.0011931864,0.42289016,0.000040745,0.0012402231,0.00002426305,0.0014429427,0.16235498,0.00029458737,0.0029452413,0.031007329,0.37655306],"study_design_scores_gemma":[0.00006974367,0.000053059946,0.04742867,0.0000020298623,0.00012364982,0.000002247432,0.00005800501,0.94973725,0.000084431485,0.00032104854,0.0020220696,0.0000977918],"about_ca_topic_score_codex":0.00020048337,"about_ca_topic_score_gemma":0.000021786425,"teacher_disagreement_score":0.78738225,"about_ca_system_score_codex":0.000012595092,"about_ca_system_score_gemma":0.00012197473,"threshold_uncertainty_score":0.9815448},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4225588997","doi":"10.3389/fpubh.2022.847420","title":"Antimicrobial Stewardship in Surgery: A Literature Bibliometric Analysis","year":2022,"lang":"en","type":"article","venue":"Frontiers in Public Health","topic":"Antibiotic Use and Resistance","field":"Immunology and Microbiology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"China International Medical Foundation; Shanghai Shen Kang Hospital Development Center","keywords":"Antimicrobial stewardship; Bibliometrics; Citation; Library science; Medicine; Web of science; MEDLINE; Ranking (information retrieval); Stewardship (theology); Antibiotic resistance; Political science; Internal medicine; Meta-analysis; Antibiotics; Computer science; Biology","score_opus":0.02237268627922744,"score_gpt":0.2587835507543104,"score_spread":0.23641086447508297,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4225588997","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9397573,0.040775433,0.0022796432,0.01154821,0.004275181,0.00045825084,0.0003218835,0.000094294715,0.0004898406],"genre_scores_gemma":[0.99466765,0.0016577262,0.00033438602,0.0016193601,0.000022645017,0.00001628619,0.0005836429,0.00001624465,0.0010820884],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99660516,0.0012882095,0.0006113002,0.000483699,0.00010176319,0.0009098703],"domain_scores_gemma":[0.9991865,0.00013645654,0.00021692568,0.00035350755,0.000041106694,0.00006551351],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.002891863,0.00017667696,0.0007647568,0.09280477,0.0002832041,0.00005178179,0.00036728062,0.00017790587,0.0002819142],"category_scores_gemma":[0.00017617078,0.0001742727,0.00021029785,0.19440398,0.00009706681,0.00015605446,0.00014344678,0.0008554769,0.000012041956],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000060919647,0.00045274975,0.7752854,0.00007957848,0.0003088833,0.00005249418,0.0012047053,0.00003555979,0.00005214415,0.00013368923,0.20572565,0.016608229],"study_design_scores_gemma":[0.0011011008,0.000090795285,0.7578171,0.000052841708,0.000029478275,0.000027412274,0.0013802671,0.000049316553,0.000037666854,0.00008091574,0.23899913,0.0003339889],"about_ca_topic_score_codex":0.000440116,"about_ca_topic_score_gemma":0.00017409977,"teacher_disagreement_score":0.10159921,"about_ca_system_score_codex":0.00046500308,"about_ca_system_score_gemma":0.0007881369,"threshold_uncertainty_score":0.91747755},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4226014969","doi":"10.1007/s12564-022-09759-4","title":"A bibliometric mapping of shadow education research: achievements, limitations, and the future","year":2022,"lang":"en","type":"article","venue":"Asia Pacific Education Review","topic":"Global Educational Reforms and Inequalities","field":"Social Sciences","cited_by":61,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Nazarbayev University; International Development Research Centre","keywords":"Shadow (psychology); Metadata; Socioeconomic status; Educational research; Sociology; Higher education; Economics education; Social science; Psychology; Computer science; Pedagogy; Primary education; Economic growth; World Wide Web; Economics","score_opus":0.12894635699541473,"score_gpt":0.41701059608287117,"score_spread":0.28806423908745643,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4226014969","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013380354,0.58292055,0.000004763936,0.178629,0.0021458643,0.0019525168,0.000020253869,0.000022910013,0.2209238],"genre_scores_gemma":[0.6003716,0.38496822,0.0003873237,0.0021266644,0.0005081693,0.0014364771,0.00011022183,0.000010040755,0.010081259],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99700975,0.0012715257,0.00040438128,0.00019226527,0.0009160293,0.00020606602],"domain_scores_gemma":[0.9981671,0.0004931181,0.0002649912,0.0002657237,0.00071616104,0.00009293739],"candidate_categories":["bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.00631422,0.0000817246,0.00018621607,0.0034401638,0.00131757,0.00007887741,0.00032818838,0.000026884903,0.0005515191],"category_scores_gemma":[0.0015894332,0.00006024242,0.00007075201,0.028118076,0.00032184523,0.000244369,0.00007630568,0.00022857892,0.000016858989],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000003798519,0.00020419755,0.0010352426,0.0003120679,0.0000113742435,1.34203875e-8,0.008221859,1.09515234e-7,0.0000012427037,0.5829517,0.13367185,0.2735865],"study_design_scores_gemma":[0.00005534061,0.000014223879,0.0070537813,0.00019175903,0.000012169676,0.0000015013588,0.26294473,3.250582e-7,3.140317e-7,0.010872297,0.71880406,0.00004950443],"about_ca_topic_score_codex":0.0013204969,"about_ca_topic_score_gemma":0.00005059556,"teacher_disagreement_score":0.58699125,"about_ca_system_score_codex":0.00021538309,"about_ca_system_score_gemma":0.0032434813,"threshold_uncertainty_score":0.9999826},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4226297244","doi":"10.1162/qss_c_00178","title":"Towards a moralization of bibliometrics? A response to Kyle Siler","year":2022,"lang":"en","type":"article","venue":"Quantitative Science Studies","topic":"Advanced Causal Inference Techniques","field":"Mathematics","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Publication; Bibliometrics; Epistemology; Point (geometry); Sociology; Classics; Philosophy; History; Political science; Law; Computer science; Library science; Mathematics","score_opus":0.4986595425254802,"score_gpt":0.570894191989019,"score_spread":0.0722346494635388,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4226297244","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.88683987,0.00041465857,0.1097823,0.00096172554,0.00020740942,0.0005468166,0.00004559421,0.00017293959,0.0010286912],"genre_scores_gemma":[0.8311115,0.00003950048,0.16844232,0.00010003086,0.000004270667,0.0001643218,3.559905e-7,0.0000105918525,0.00012712597],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9977988,0.00022236856,0.0003369241,0.00033484516,0.0010245577,0.00028249616],"domain_scores_gemma":[0.99708974,0.0012698313,0.00022434093,0.00029175612,0.0010622097,0.00006210471],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.003944249,0.00012928931,0.00030917613,0.0053611635,0.00049590797,0.000022268248,0.00047128054,0.000012288864,0.00005657141],"category_scores_gemma":[0.021947762,0.000113562644,0.00004392042,0.02838833,0.00061808847,0.0003923363,0.0007790055,0.00010281093,0.000006284152],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035913338,0.00018512661,0.0006070775,0.000058420177,0.000042694504,0.0000067635856,0.02714677,0.0002550015,0.20741585,0.75775427,0.003957395,0.0022115163],"study_design_scores_gemma":[0.00034456985,0.0040928423,0.006922289,0.000088727575,0.00003446014,0.000005938963,0.06466841,0.00054421206,0.14352544,0.7771622,0.0021221098,0.0004888195],"about_ca_topic_score_codex":0.000016460343,"about_ca_topic_score_gemma":0.0000048614984,"teacher_disagreement_score":0.063890405,"about_ca_system_score_codex":0.0002604658,"about_ca_system_score_gemma":0.0001902928,"threshold_uncertainty_score":0.99226373},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W423015428","doi":"10.1016/j.joi.2015.05.001","title":"Modelling count response variables in informetric studies: Comparison among count, linear, and lognormal regression models","year":2015,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Advanced Text Analysis Techniques","field":"Computer Science","cited_by":54,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Capital Medical University","keywords":"Count data; Statistics; Negative binomial distribution; Poisson regression; Akaike information criterion; Mathematics; Regression analysis; Linear regression; Poisson distribution; Overdispersion; Log-normal distribution; Generalized linear model; Binomial regression; Regression diagnostic; Econometrics; Polynomial regression; Population","score_opus":0.103223804081119,"score_gpt":0.3570945740796192,"score_spread":0.2538707699985002,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W423015428","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.17380138,0.0037154844,0.8218133,0.000059706494,0.00016024518,0.000112353264,0.0000011187186,0.000042383173,0.00029405425],"genre_scores_gemma":[0.68673,0.002247306,0.31082717,0.00009233856,0.000042733995,0.0000029471632,7.0284295e-7,0.000010189707,0.000046578996],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.996476,0.00010515386,0.0016057367,0.0001951677,0.0012315819,0.0003863743],"domain_scores_gemma":[0.9949323,0.0012002634,0.0015050469,0.00037619343,0.0017142985,0.0002719241],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0060026147,0.00025620664,0.0007645148,0.0047702654,0.000095942305,0.00017513546,0.0008552774,0.00017694189,5.2453373e-7],"category_scores_gemma":[0.002962127,0.00019221741,0.000105454244,0.006009157,0.000101941405,0.005555848,0.00041682617,0.0006421939,0.0000012358203],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034110877,0.0000925449,0.014880593,0.000034874352,0.000067810586,0.00004908803,0.0030228484,0.96374387,0.000006136033,0.0015121952,0.0007795466,0.015469374],"study_design_scores_gemma":[0.0009498698,0.000508794,0.00043741913,0.00020656371,0.000025689204,0.00007024848,0.00094119983,0.98872674,0.0002075702,0.005726562,0.001970956,0.0002283904],"about_ca_topic_score_codex":0.000041235755,"about_ca_topic_score_gemma":0.000008155375,"teacher_disagreement_score":0.51292866,"about_ca_system_score_codex":0.00061564107,"about_ca_system_score_gemma":0.00042554547,"threshold_uncertainty_score":0.7838399},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"methods","study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4232503042","doi":"10.1007/978-1-4614-6170-8_110038","title":"Bibliometrics","year":2014,"lang":"en","type":"book-chapter","venue":"","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Bibliometrics; Computer science; Library science","score_opus":0.062186757456040645,"score_gpt":0.27360002029525443,"score_spread":0.2114132628392138,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4232503042","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[4.5688626e-7,0.00030023625,0.00088183925,0.0000178416,0.00021768303,0.00010825424,0.00006637739,0.00045133906,0.997956],"genre_scores_gemma":[0.00015015184,0.000087743414,0.00076998136,0.00022210903,0.00032584256,0.0000013139244,0.000063319356,0.0003393135,0.9980402],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987464,0.0000041203775,0.00020135286,0.00029510295,0.0005709895,0.00018203747],"domain_scores_gemma":[0.9988813,0.00010142602,0.00014508558,0.0005899545,0.00016423532,0.0001180229],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00026890094,0.0002654886,0.00030124534,0.042743552,0.000021689686,0.00006031745,0.0002779803,0.00038482968,0.026469667],"category_scores_gemma":[0.00007038381,0.00022926931,0.00013991588,0.0025297166,0.000052839663,0.000027127355,0.00011086821,0.0002610269,0.45631874],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000011933523,0.000001229803,0.0000027369952,0.000006868572,0.000021728987,0.0000036404865,5.491447e-7,1.4241348e-7,0.000006129354,0.52532715,0.4702773,0.004351335],"study_design_scores_gemma":[0.000092253336,0.000024722522,0.000012656846,0.000018553035,0.000037576523,0.0000057898546,1.1579486e-7,0.0000093265535,0.0000157891,0.017945804,0.98153555,0.00030187395],"about_ca_topic_score_codex":0.000008897049,"about_ca_topic_score_gemma":0.000010099293,"teacher_disagreement_score":0.51125824,"about_ca_system_score_codex":0.00006958402,"about_ca_system_score_gemma":0.000030937965,"threshold_uncertainty_score":0.97442025},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4234189662","doi":"10.1002/asi.20880","title":"Modifying the journal impact factor by fractional citation weighting: The audience factor","year":2008,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":178,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thomson Reuters (Canada)","funders":"","keywords":"Impact factor; Normalization (sociology); Citation; Weighting; Computer science; Citation analysis; Factor (programming language); Field (mathematics); Citation impact; Discipline; Information retrieval; Statistics; Sociology; Mathematics; Library science; Social science; Political science; Law","score_opus":0.32657725380462954,"score_gpt":0.5119929688617303,"score_spread":0.18541571505710075,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4234189662","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9520446,0.00022198255,0.029396757,0.01770973,0.00032746146,0.0001809829,0.000029881088,0.000007634373,0.000080963466],"genre_scores_gemma":[0.9970457,0.00091966055,0.0010102775,0.0009105585,0.00006081862,0.0000030019125,1.8973272e-7,0.0000027665578,0.000047018766],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9922488,0.00007500657,0.0008125613,0.00012826711,0.006304288,0.00043107686],"domain_scores_gemma":[0.98679394,0.0026844474,0.00253445,0.00033430487,0.007501177,0.00015169068],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["sts"],"category_scores_codex":[0.015175322,0.00011350404,0.00022254264,0.0054881265,0.003502267,0.0011890295,0.0034765124,0.00005840816,0.000017034421],"category_scores_gemma":[0.020850388,0.00004212226,0.0003188631,0.06675902,0.00397756,0.0040474595,0.0003780067,0.0006843707,0.0000064889186],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009780682,0.000082490544,0.10500408,0.000004960155,0.00012871377,7.0435715e-7,0.008819265,0.00045739138,0.019147512,0.0010482842,0.18348764,0.68172115],"study_design_scores_gemma":[0.0015516072,0.0018263817,0.52625203,0.00002487265,0.00004332952,0.0031596264,0.0603164,0.05702914,0.006229599,0.013651215,0.32944673,0.00046907336],"about_ca_topic_score_codex":0.00001700896,"about_ca_topic_score_gemma":5.5379877e-7,"teacher_disagreement_score":0.68125206,"about_ca_system_score_codex":0.00029150018,"about_ca_system_score_gemma":0.0010178919,"threshold_uncertainty_score":0.9998478},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4241175141","doi":"10.14429/djlit.37.11573","title":"Metamaterials Research: A Scientometric Assessment of Global Publications Output during 2007-16","year":2017,"lang":"en","type":"article","venue":"DESIDOC Journal of Library & Information Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Citation impact; Citation; China; Library science; Metamaterial; World class; Web of science; Geography; Mathematics; Political science; Physics; Computer science; Engineering; Optics; Law","score_opus":0.5649173997868181,"score_gpt":0.5944565569745612,"score_spread":0.029539157187743137,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4241175141","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94991535,0.0004803391,0.010419322,0.010975781,0.0009796995,0.00033630425,0.00011004519,0.00005108886,0.026732087],"genre_scores_gemma":[0.98761904,0.0004896647,0.011148626,0.000042924705,0.000084032195,0.000009426547,0.0000041022777,0.0000064255496,0.0005957739],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9873117,0.0003534617,0.0025891971,0.00026436572,0.008789293,0.00069198664],"domain_scores_gemma":[0.9852946,0.0011452909,0.0037682357,0.0015954026,0.007826623,0.0003698625],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.029095357,0.0001516526,0.0005843825,0.14585528,0.00094758434,0.0065263915,0.008144996,0.00026553348,0.00086686295],"category_scores_gemma":[0.0632099,0.00010797067,0.00021837257,0.14890714,0.00091194996,0.018520527,0.0022870041,0.00057459425,0.00014898564],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000063166226,0.00024261995,0.7403365,0.00004025058,0.000095957344,0.000012334132,0.00006868298,0.000045956338,0.00094249914,0.11232102,0.050925683,0.094905324],"study_design_scores_gemma":[0.0010754724,0.00042638468,0.8718072,0.000045251334,0.0000116584615,0.00017490734,0.00067893456,0.00029798655,0.0109382225,0.05671687,0.057664678,0.00016241585],"about_ca_topic_score_codex":0.000017481158,"about_ca_topic_score_gemma":9.2157336e-7,"teacher_disagreement_score":0.13147072,"about_ca_system_score_codex":0.0002854057,"about_ca_system_score_gemma":0.0013216106,"threshold_uncertainty_score":0.9997507},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4241465860","doi":"10.1504/ijbbm.2018.092803","title":"City logistics: a review and bibliometric analysis","year":2018,"lang":"en","type":"review","venue":"International Journal of Bibliometrics in Business and Management","topic":"Urban and Freight Transport Logistics","field":"Engineering","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Citation; Field (mathematics); Social network analysis; Data science; Citation analysis; City logistics; Bibliometrics; Computer science; Library science; Engineering; World Wide Web; Social media; Transport engineering","score_opus":0.09743947623305906,"score_gpt":0.3360887360137579,"score_spread":0.23864925978069884,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4241465860","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000003709838,0.9557279,0.041808885,0.000032716795,0.000893078,0.00021831937,0.00007259071,0.000016395128,0.0012264289],"genre_scores_gemma":[0.00024435375,0.9976653,0.0017134129,0.00006211314,0.00018730003,0.0000099342205,0.00004073159,0.000028009696,0.0000488504],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975457,0.000029281262,0.0012739596,0.00023280452,0.0007146019,0.00020362216],"domain_scores_gemma":[0.9982669,0.00018441219,0.00047819255,0.00018703338,0.00077027973,0.00011323056],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0010162774,0.00032708477,0.0015419668,0.5763725,0.000019616271,0.00021816624,0.0006059018,0.00014766754,0.00006194814],"category_scores_gemma":[0.00019748951,0.00026340925,0.00024858338,0.5269472,0.00009979752,0.00019357991,0.00015018939,0.0002803462,0.0000032058733],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000036081817,0.000089393674,0.0006005805,0.040327564,0.0048182304,0.0005312883,0.000004147001,0.0000814049,6.128105e-9,0.0006794856,0.005434844,0.9474295],"study_design_scores_gemma":[0.00023594295,0.000021608666,0.008013529,0.008440915,0.007477338,0.000085367574,0.0000017611743,0.00009945668,2.1667365e-8,0.00011449681,0.97522724,0.0002823078],"about_ca_topic_score_codex":0.000011852474,"about_ca_topic_score_gemma":0.000008774932,"teacher_disagreement_score":0.9697924,"about_ca_system_score_codex":0.00014501337,"about_ca_system_score_gemma":0.000025724674,"threshold_uncertainty_score":0.9999818},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4241889296","doi":"10.1002/asi.20888","title":"Conference proceedings as a source of scientific information: A bibliometric analysis","year":2008,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":157,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Scientific literature; Library science; Natural science; Computer science; Sociology of scientific knowledge; Data science; Engineering ethics; Sociology; Social science; Epistemology; Engineering; Philosophy; Geology","score_opus":0.19550787465052466,"score_gpt":0.4623797083122703,"score_spread":0.26687183366174566,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4241889296","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98421335,0.00006186569,0.011766738,0.0030569118,0.0001541136,0.00019970766,0.000010954738,0.0000113405395,0.00052503817],"genre_scores_gemma":[0.99516815,0.00022444755,0.004161401,0.00033271837,0.000011417454,0.0000051847915,4.3857438e-7,0.0000016231888,0.000094646544],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99026775,0.000014166449,0.0012682957,0.00017381314,0.007887334,0.00038862287],"domain_scores_gemma":[0.96304363,0.000609794,0.0035111408,0.0003966988,0.032245107,0.00019365725],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.020302929,0.0001063433,0.00045700071,0.3879982,0.0010427777,0.0016255957,0.0034192104,0.00007028617,0.000012200737],"category_scores_gemma":[0.037297126,0.00006118925,0.00043771928,0.88342136,0.006742647,0.00787826,0.00070760195,0.00025853465,0.000010573918],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007316556,0.00009902123,0.3403982,0.000029023751,0.0002601891,2.8418634e-7,0.005985707,0.00010582247,0.0034196666,0.007811286,0.02899052,0.6128271],"study_design_scores_gemma":[0.0017714786,0.0019586163,0.50007063,0.000030597752,0.00024076737,0.0005887469,0.053463895,0.048233222,0.0138195,0.009815167,0.36952487,0.00048248557],"about_ca_topic_score_codex":0.000026413703,"about_ca_topic_score_gemma":5.1607225e-7,"teacher_disagreement_score":0.6123446,"about_ca_system_score_codex":0.00011603249,"about_ca_system_score_gemma":0.0013879016,"threshold_uncertainty_score":0.9994108},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4248643252","doi":"10.23880/oajprs-16000116","title":"Global Asthma Research with Special Reference to India: A Scientometric Assessment of Publication Output during 2007-16","year":2018,"lang":"en","type":"article","venue":"Open Access Journal of Pulmonary & Respiratory Sciences","topic":"Inhalation and Respiratory Drug Delivery","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Asthma; Medicine; Library science; Computer science; Internal medicine","score_opus":0.22207067140611822,"score_gpt":0.5039208769829117,"score_spread":0.2818502055767934,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4248643252","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.83939534,0.00016188071,0.00011286443,0.0006586408,0.00036486558,0.0005611109,0.000010175083,0.0000069519333,0.15872818],"genre_scores_gemma":[0.99619395,0.000018904508,0.0010886712,0.0006899819,0.0012970896,0.000010727426,0.0000012902258,0.000010345831,0.000689063],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9945826,0.00042352354,0.00091815996,0.00050594285,0.0030596019,0.0005102045],"domain_scores_gemma":[0.9936183,0.00010661661,0.00084095745,0.00039653236,0.0045167506,0.00052081904],"candidate_categories":["scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.008037439,0.00018216303,0.00045932358,0.0032639517,0.0007698321,0.0011141652,0.0027007146,0.00009697329,0.0010254657],"category_scores_gemma":[0.00066468015,0.00012360029,0.000074030126,0.011535566,0.0012652213,0.0037264614,0.0009676591,0.00039667627,0.000045095996],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013559135,0.0008532146,0.950641,0.00009909853,0.00006951391,0.00028843535,0.0001638164,0.000111639165,0.0041520507,0.0012879926,0.012131626,0.028845696],"study_design_scores_gemma":[0.0009703879,0.0028524674,0.9489901,0.00038896996,0.000021293008,0.00025525733,0.0006202643,0.00005180499,0.0024865963,0.00005473694,0.043121845,0.00018627351],"about_ca_topic_score_codex":0.00006158956,"about_ca_topic_score_gemma":0.00003857444,"teacher_disagreement_score":0.15803912,"about_ca_system_score_codex":0.0005709704,"about_ca_system_score_gemma":0.003551874,"threshold_uncertainty_score":0.99992275},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4254696421","doi":"10.1002/asi.20285","title":"The diffusion of theories: A functional approach","year":2005,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University; Rogers Communications (Canada)","funders":"","keywords":"Typology; Citation; Context (archaeology); Diffusion; Citation analysis; Diffusion theory; Diffusion of innovations; Sociology; Innovation diffusion; Epistemology; Computer science; Data science; Knowledge management; Social science; World Wide Web","score_opus":0.1571127675202829,"score_gpt":0.4602075098049632,"score_spread":0.30309474228468036,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4254696421","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92782044,0.0002627875,0.048736587,0.021472862,0.0002980089,0.00022915081,0.00001021508,0.000007892999,0.0011620749],"genre_scores_gemma":[0.99308026,0.0003477501,0.006000985,0.0004427329,0.000039869796,0.000004189354,1.0785907e-7,0.0000013824732,0.00008271273],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99457556,0.000023884773,0.00069015793,0.00010040807,0.004364133,0.0002458655],"domain_scores_gemma":[0.9895999,0.0013564777,0.0015627644,0.00033393688,0.007074201,0.00007270949],"candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.022274127,0.000059522532,0.0001812004,0.005592714,0.0009036698,0.00053559634,0.002234419,0.00003886126,0.0000027431358],"category_scores_gemma":[0.019373389,0.000024568175,0.000209091,0.08023848,0.0044495575,0.0016255905,0.00050209084,0.00020416791,0.0000020857804],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004061122,0.00003872799,0.006503583,0.000002901531,0.000018667632,1.3103329e-8,0.0006447338,0.000082898485,0.0022208313,0.0502954,0.01255004,0.9276016],"study_design_scores_gemma":[0.0012402699,0.0009021489,0.054176405,0.000012309838,0.000025258803,0.0001586213,0.064645864,0.05008943,0.0058348062,0.069654316,0.7530654,0.0001951423],"about_ca_topic_score_codex":0.0000031385696,"about_ca_topic_score_gemma":4.6278126e-7,"teacher_disagreement_score":0.92740643,"about_ca_system_score_codex":0.00008997659,"about_ca_system_score_gemma":0.00054642546,"threshold_uncertainty_score":0.9982598},"labels":[{"model":"gemma","categories":["sts"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4280533913","doi":"10.3390/su14106003","title":"Inventory Models in a Sustainable Supply Chain: A Bibliometric Analysis","year":2022,"lang":"en","type":"article","venue":"Sustainability","topic":"Sustainable Supply Chain Management","field":"Business, Management and Accounting","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Supply chain; China; Sustainability; Bibliometrics; Sustainable development; Business; Citation; Consumption (sociology); Publication; Environmental economics; Production (economics); Per capita; Agriculture; Regional science; Agricultural economics; Marketing; Library science; Economics; Political science; Computer science; Geography; Social science; Sociology","score_opus":0.013897476550779093,"score_gpt":0.23715716488799038,"score_spread":0.2232596883372113,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4280533913","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9657549,0.0007759935,0.0020952215,0.007201248,0.0002649892,0.003865461,0.000009544909,0.00053522526,0.019497368],"genre_scores_gemma":[0.987574,0.000007449857,0.000044540495,0.0016702429,0.00020995442,0.001769458,0.00011862037,0.000068908514,0.008536854],"study_design_codex":"observational","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99429387,0.0002379174,0.00088413566,0.0012719479,0.0013848551,0.0019272632],"domain_scores_gemma":[0.9964613,0.00014765716,0.00038662527,0.0013863354,0.0015580563,0.000060030758],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.006251319,0.00048360782,0.00075388246,0.16081272,0.00089427264,0.00054458313,0.0012690427,0.00009532699,0.0020080463],"category_scores_gemma":[0.0016375566,0.0005613923,0.00048164616,0.3621646,0.00017327063,0.0026757838,0.0035766028,0.00065796287,0.0000193912],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020884846,0.00087814464,0.48275292,0.0013670382,0.0002245546,0.000785727,0.000511726,0.24218181,6.2309255e-7,0.26378322,0.0047153817,0.0025900255],"study_design_scores_gemma":[0.0018329795,0.00006369653,0.20919219,0.0000046191776,0.00047144256,0.000002024367,0.07789519,0.1922131,0.0000010283829,0.36133617,0.15599898,0.0009885781],"about_ca_topic_score_codex":0.022458807,"about_ca_topic_score_gemma":0.00048563114,"teacher_disagreement_score":0.27356073,"about_ca_system_score_codex":0.0066779456,"about_ca_system_score_gemma":0.000567056,"threshold_uncertainty_score":0.99968374},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4280604520","doi":"10.21203/rs.3.rs-1639305/v1","title":"Graph Neural Networks: a bibliometrics overview","year":2022,"lang":"en","type":"preprint","venue":"Research Square","topic":"Advanced Graph Neural Networks","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Computer science; Field (mathematics); Data science; Scopus; Artificial intelligence; Subject (documents); Graph; Convolutional neural network; Science and engineering; Library science; Engineering ethics; Political science; Engineering","score_opus":0.14965168741352466,"score_gpt":0.4316644671126066,"score_spread":0.2820127796990819,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4280604520","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009598502,0.19927727,0.764763,0.005449646,0.00850571,0.0058240024,0.00017433184,0.0025607503,0.0038468007],"genre_scores_gemma":[0.8806549,0.08128332,0.029352846,0.0014499506,0.002108765,0.003004806,0.00043029545,0.00035632018,0.0013587946],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9891484,0.002015166,0.0006744371,0.0020874867,0.0040231235,0.002051382],"domain_scores_gemma":[0.99268013,0.0018316032,0.00028489862,0.003811478,0.0008049129,0.00058696006],"candidate_categories":["metaepi_narrow","bibliometrics","scholarly_communication","open_science","research_integrity"],"consensus_categories":["bibliometrics","open_science"],"category_scores_codex":[0.0035329275,0.0005741825,0.0007035683,0.016290437,0.00092127186,0.0012740742,0.007523279,0.0004240471,0.00026517367],"category_scores_gemma":[0.0005325528,0.0005713935,0.00060883065,0.068859674,0.00025921752,0.0006683217,0.023716057,0.007495738,0.00003241044],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000049794173,0.00038134432,0.0049061477,0.0011385245,0.00013542427,0.0012371053,0.00025896507,0.7167008,0.0000060596867,0.049752872,0.04718751,0.17824547],"study_design_scores_gemma":[0.00042331085,0.000514815,0.0076363287,0.00035703686,0.000011956651,0.00004100198,0.000059353042,0.8745436,0.0000070267038,0.07758165,0.03780043,0.0010234985],"about_ca_topic_score_codex":0.00011882424,"about_ca_topic_score_gemma":0.00002172162,"teacher_disagreement_score":0.8710564,"about_ca_system_score_codex":0.00037619044,"about_ca_system_score_gemma":0.00028409212,"threshold_uncertainty_score":0.9997627},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4280628334","doi":"10.21037/jtd-22-233","title":"A study of mechanical ventilation in the ICU after cardiac surgery: a bibliometric analysis","year":2022,"lang":"en","type":"article","venue":"Journal of Thoracic Disease","topic":"Scientific Research and Technology","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Science Citation Index; Bibliometrics; Mechanical ventilation; Intensive care unit; Citation; Impact factor; Cardiac surgery; Library science; Intensive care medicine; Surgery; Internal medicine; Political science","score_opus":0.03947840529724142,"score_gpt":0.35591367082379083,"score_spread":0.31643526552654944,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4280628334","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9923699,0.0006589701,0.006033292,0.000514365,0.0002480778,0.0001509698,0.000010132244,0.0000064783326,0.000007844592],"genre_scores_gemma":[0.99968827,0.000015636579,0.00020949979,0.0000246695,0.000017268783,0.000030175626,0.000001059889,0.0000023281048,0.0000110934925],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9967739,0.0009724622,0.00043911132,0.00017233788,0.00146266,0.00017955297],"domain_scores_gemma":[0.9985876,0.00033098605,0.0003043021,0.0005156043,0.00014660395,0.00011487225],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.005958198,0.000060026385,0.00027643293,0.048072714,0.00007638491,0.000105522704,0.0011443321,0.000013764039,0.000049066704],"category_scores_gemma":[0.0004450612,0.00004109853,0.00034445155,0.14568737,0.000025256773,0.00027448847,0.00036256088,0.00024167768,0.0000014114711],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001007772,0.0053633656,0.83080924,0.00004573983,0.0007851093,0.0028989068,0.0016693228,0.00093792414,0.00008891158,0.00037713005,0.002860338,0.15315625],"study_design_scores_gemma":[0.00012031159,0.00029932545,0.9933634,0.000008417869,0.00022018308,0.0000109817975,0.0010088775,0.003058586,0.000056702153,0.0013673345,0.0004120505,0.0000738456],"about_ca_topic_score_codex":0.000014182325,"about_ca_topic_score_gemma":0.0000056471754,"teacher_disagreement_score":0.16255416,"about_ca_system_score_codex":0.00004734781,"about_ca_system_score_gemma":0.00016635978,"threshold_uncertainty_score":0.9627166},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4281255308","doi":"10.1007/s11192-022-04387-6","title":"Measuring the disparity among scientific disciplines using Library of Congress Subject Headings","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Subject (documents); Library of congress; Computer science; Information retrieval; Bibliometrics; Scientific literature; Library science; Data science; Scientific communication","score_opus":0.6099867867029692,"score_gpt":0.5220149078759858,"score_spread":0.08797187882698343,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281255308","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9879879,0.0015075497,0.0013361184,0.00052365946,0.0053552277,0.00053315965,0.00029660552,0.000052499592,0.002407278],"genre_scores_gemma":[0.9935462,0.000035789883,0.00087497046,0.000054386677,0.00008843888,0.000023520164,0.000013053303,0.000027988659,0.005335636],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9656685,0.0009097184,0.0013897247,0.0015005269,0.029245835,0.0012857212],"domain_scores_gemma":[0.984528,0.007917204,0.0011270529,0.0023133033,0.0033797661,0.00073464296],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","open_science"],"category_scores_codex":[0.07053809,0.00027002013,0.00055330124,0.15632807,0.0039363354,0.0077411365,0.00913916,0.000080132384,0.0012392598],"category_scores_gemma":[0.044533387,0.0001722012,0.0003984683,0.67784804,0.0020639056,0.002198376,0.008683388,0.00074123207,0.000038542956],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000030800664,0.0003192126,0.9601827,0.000016292148,0.000021679974,0.000013142909,0.0005438753,0.0035693957,0.0020777837,0.0012293599,0.0125460075,0.019449735],"study_design_scores_gemma":[0.0009178443,0.00026934507,0.7697348,0.000025490062,0.00003962694,0.00003573169,0.0031927584,0.12977989,0.010902231,0.007883293,0.07650656,0.0007124588],"about_ca_topic_score_codex":0.00021839826,"about_ca_topic_score_gemma":0.000025655732,"teacher_disagreement_score":0.52151996,"about_ca_system_score_codex":0.00029372665,"about_ca_system_score_gemma":0.0007736293,"threshold_uncertainty_score":0.9996737},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4281647098","doi":"10.1002/leap.1463","title":"The <scp>AMSTAR</scp>‐2 critical appraisal tool and editorial decision‐making for systematic reviews: Retrospective, bibliometric study","year":2022,"lang":"en","type":"article","venue":"Learned Publishing","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto General Hospital; University of Toronto","funders":"","keywords":"Operationalization; Systematic review; Strengths and weaknesses; Medicine; Critical appraisal; Retrospective cohort study; MEDLINE; Family medicine; Psychology; Pathology; Alternative medicine; Social psychology; Political science; Physics; Law","score_opus":0.47601007212451607,"score_gpt":0.5261948034173058,"score_spread":0.050184731292789775,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281647098","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.33091587,0.07123648,0.24263431,0.0069349874,0.27506912,0.059564415,0.00036376205,0.00029338806,0.012987673],"genre_scores_gemma":[0.97578776,0.00008392514,0.0063761068,0.00034543945,0.010999858,0.0039877756,0.000009485772,0.00007628835,0.0023333605],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9113967,0.042888932,0.017584624,0.0031740274,0.023653258,0.0013024708],"domain_scores_gemma":[0.16093944,0.8021024,0.0154280765,0.012273969,0.008522968,0.00073313713],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.4386619,0.00049879646,0.0047253855,0.009805897,0.0031462798,0.035406303,0.0047950684,0.00014563015,0.0005735203],"category_scores_gemma":[0.926664,0.00023668376,0.0016025518,0.054499276,0.00010665521,0.0030390078,0.0015290745,0.0011217437,0.00016156668],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016248478,0.00018473419,0.015631301,0.001215737,0.0002658152,0.000006808829,0.0022015234,0.000063801985,0.000003903316,0.0077267187,0.9590498,0.013633574],"study_design_scores_gemma":[0.0010307828,0.00053801475,0.0083163185,0.0007571075,0.00079078315,0.000051561852,0.03651738,0.009127251,3.5619064e-7,0.028793644,0.9137969,0.00027989244],"about_ca_topic_score_codex":0.000014991397,"about_ca_topic_score_gemma":0.00002263611,"teacher_disagreement_score":0.7592135,"about_ca_system_score_codex":0.00018605567,"about_ca_system_score_gemma":0.000139757,"threshold_uncertainty_score":0.9981515},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4282004915","doi":"10.1177/20597002221086095","title":"Top cited articles in concussion: A bibliometric analysis of the state of the science","year":2022,"lang":"en","type":"article","venue":"Journal of Concussion","topic":"Traumatic Brain Injury Research","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; University Health Network; Mount Sinai Hospital; Sinai Health System; Toronto Rehabilitation Institute; McMaster University","funders":"","keywords":"Concussion; Citation; Medicine; Cohort; Citation analysis; Psychology; Computer science; Poison control; Environmental health; Injury prevention; Library science; Pathology","score_opus":0.05726624470480519,"score_gpt":0.36336507602712503,"score_spread":0.30609883132231985,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4282004915","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9957277,0.00026579245,0.000031192154,0.003407352,0.00016378678,0.00019434615,0.000008675696,0.0000020701436,0.00019907975],"genre_scores_gemma":[0.9995222,0.00002708423,0.00009887882,0.00012449911,0.000014205249,0.0000023571629,2.0422969e-7,0.0000054207,0.00020512093],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9962708,0.00036231065,0.0008124199,0.000109028675,0.0022492057,0.00019623464],"domain_scores_gemma":[0.99787045,0.00040169066,0.00075767294,0.00039816386,0.0004684259,0.000103593746],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0036485775,0.00007070634,0.00041413138,0.031009885,0.00013863956,0.000015559144,0.0006226805,0.000020497166,0.00039493528],"category_scores_gemma":[0.0017705568,0.000030030324,0.00026305948,0.20101488,0.00043473695,0.000115423376,0.0003497981,0.00049226504,4.1115453e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001020766,0.00048497188,0.47340104,0.000064831445,0.0003643704,0.000030320653,0.003385198,0.0026366515,0.43362767,0.00014020671,0.0006483746,0.084195614],"study_design_scores_gemma":[0.0012852085,0.00037596773,0.9400878,0.00016370165,0.00020311783,0.0000482019,0.00075589353,0.0040945923,0.052505262,0.00023374861,0.00020998377,0.00003652029],"about_ca_topic_score_codex":0.00004098018,"about_ca_topic_score_gemma":0.000004943988,"teacher_disagreement_score":0.4666868,"about_ca_system_score_codex":0.0001724468,"about_ca_system_score_gemma":0.0007872673,"threshold_uncertainty_score":0.9799728},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4282012046","doi":"10.1155/2022/3755460","title":"Research Trends around Exercise Rehabilitation among Cancer Patients: A Bibliometrics and Visualized Knowledge Graph Analysis","year":2022,"lang":"en","type":"article","venue":"BioMed Research International","topic":"Cancer survivorship and care","field":"Medicine","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Natural Science Foundation of China","keywords":"Bibliometrics; Rehabilitation; MEDLINE; Medicine; Data science; Computer science; Physical therapy; Library science; Biology","score_opus":0.08594269956322853,"score_gpt":0.487670211726531,"score_spread":0.4017275121633025,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4282012046","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.989496,0.0020688437,0.000028838076,0.0022634475,0.0010853515,0.00046903678,0.00037033524,0.00004366819,0.0041744965],"genre_scores_gemma":[0.9903069,0.0007856005,0.000062948006,0.000026220634,0.00026636125,0.0012094916,0.0006812784,0.000027381451,0.0066337786],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.993888,0.0008236887,0.0003880552,0.00062770356,0.0037209236,0.00055160496],"domain_scores_gemma":[0.9953641,0.0010154197,0.00007524894,0.00033411078,0.0028697562,0.00034134556],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0051209945,0.00013969785,0.00031401528,0.057018936,0.0005795851,0.00017593328,0.00038171865,0.00008534313,0.0026074913],"category_scores_gemma":[0.00079393876,0.00013255651,0.00020949334,0.0708057,0.0006108944,0.00020506346,0.0006331279,0.0008110143,0.000010720644],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017743909,0.0019228433,0.851814,0.00014972183,0.0011363773,0.000023942926,0.0045824395,0.000015529542,0.00046212532,0.0006953578,0.05359941,0.08382386],"study_design_scores_gemma":[0.0028850464,0.0010779483,0.8683832,0.00007764023,0.00016560352,0.0000017940388,0.0058835037,0.0021640388,0.00011511427,0.00044543744,0.11859735,0.00020331702],"about_ca_topic_score_codex":0.005873627,"about_ca_topic_score_gemma":0.0018088281,"teacher_disagreement_score":0.08362055,"about_ca_system_score_codex":0.0011912163,"about_ca_system_score_gemma":0.00027354917,"threshold_uncertainty_score":0.99830425},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4282052457","doi":"10.1016/j.burns.2022.06.003","title":"Who tells the story of burns in low-and-middle income countries? – A bibliometric study","year":2022,"lang":"en","type":"article","venue":"Burns","topic":"COVID-19 and healthcare impacts","field":"Medicine","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Middle income country; Low and middle income countries; Quarter (Canadian coin); Developing country; High income countries; Economic growth; Socioeconomics; Sociology; Geography","score_opus":0.05598908742653252,"score_gpt":0.36326289395023975,"score_spread":0.3072738065237072,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4282052457","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9905148,0.0032002795,0.0000065412014,0.004770312,0.00015408758,0.0009714563,0.000031332896,0.000024244233,0.00032694623],"genre_scores_gemma":[0.9968756,0.0003414024,0.00000601141,0.0018156315,0.000082019724,0.000054350578,0.0000026724251,0.000017783605,0.00080456474],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99824077,0.0002666887,0.0003398428,0.0002138178,0.00064266525,0.00029623145],"domain_scores_gemma":[0.99849254,0.0007373515,0.00013123694,0.000399609,0.000090725036,0.00014852837],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018871151,0.000121212295,0.00037814595,0.009677484,0.00018973922,0.000015536532,0.00016733159,0.00004028872,0.00036330166],"category_scores_gemma":[0.0005957721,0.00008875058,0.000051881805,0.019373648,0.00011107964,0.000055441375,0.00021363526,0.00059116766,0.000008769383],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037839683,0.0008000111,0.97948766,0.0008575018,0.0000657178,0.00020038002,0.014917898,0.000042581087,0.00005570992,0.000340249,0.0009793935,0.0018745136],"study_design_scores_gemma":[0.0022196353,0.0015704785,0.9655141,0.00019356536,0.000051286603,0.000033966873,0.0058293985,0.00006310027,0.00003180397,0.00005100962,0.024340803,0.00010084703],"about_ca_topic_score_codex":0.0012806783,"about_ca_topic_score_gemma":0.0002224622,"teacher_disagreement_score":0.02336141,"about_ca_system_score_codex":0.00027410474,"about_ca_system_score_gemma":0.0005000582,"threshold_uncertainty_score":0.9308396},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4283220805","doi":"10.1007/s40037-022-00717-9","title":"Joining the meta-research movement: A bibliometric case study of the journal &lt;em&gt;Perspectives on Medical Education&lt;/em&gt;","year":2022,"lang":"en","type":"article","venue":"Perspectives on Medical Education","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Ottawa","funders":"","keywords":"Citation; Library science; Metadata; Medical education; Medicine; Psychology; World Wide Web; Computer science","score_opus":0.3729390460331222,"score_gpt":0.5601110664533843,"score_spread":0.18717202042026204,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283220805","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9513281,0.012295006,0.00017407542,0.026831517,0.0028175423,0.0014873123,0.000022343207,0.000025593365,0.0050184885],"genre_scores_gemma":[0.9933635,0.0013788033,0.00008594294,0.001934489,0.00079710816,0.00044464064,0.0000028414295,0.000038597194,0.0019540675],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.9313866,0.0083170235,0.0016184289,0.0014615131,0.056153342,0.0010630989],"domain_scores_gemma":[0.93358725,0.036527496,0.0019413874,0.004396992,0.020294793,0.003252105],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.108470336,0.00040026417,0.00074671983,0.20523307,0.0040241904,0.0018215083,0.006988083,0.00020035153,0.021291632],"category_scores_gemma":[0.32135263,0.00020912448,0.0005431938,0.5120837,0.00090975,0.0005408703,0.0021496667,0.0039200312,0.000100391095],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035608868,0.052548196,0.00863595,0.000021646732,0.0012016399,0.00019940645,0.3066976,0.0003021719,0.00010257632,0.021526227,0.26668367,0.3417248],"study_design_scores_gemma":[0.0010482554,0.0022462504,0.025107639,0.000033532222,0.0001414611,0.00064542383,0.952462,0.0015438036,0.000029561576,0.0040325215,0.012439771,0.0002697492],"about_ca_topic_score_codex":0.00082503934,"about_ca_topic_score_gemma":0.0003132162,"teacher_disagreement_score":0.6457644,"about_ca_system_score_codex":0.0016875056,"about_ca_system_score_gemma":0.013238921,"threshold_uncertainty_score":0.9992147},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4283456040","doi":"10.21203/rs.3.rs-1759648/v1","title":"Bibliometric and Visual Analysis of Research on the Links between the Microbiome-Gut-Brain axis and Depression","year":2022,"lang":"en","type":"preprint","venue":"Research Square","topic":"Tryptophan and brain disorders","field":"Neuroscience","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"China Academy of Chinese Medical Sciences; National Natural Science Foundation of China; National Key Research and Development Program of China; University College Cork; McMaster University","keywords":"Depression (economics); Citation; Microbiome; Anxiety; Gut microbiome; Medicine; Psychology; Psychiatry; Bioinformatics; Biology; Library science; Computer science","score_opus":0.1737048388415651,"score_gpt":0.4724854897595097,"score_spread":0.2987806509179446,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283456040","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9873075,0.0019885094,0.00001558391,0.0072727157,0.00006887802,0.0015083115,0.00049629953,0.000037951544,0.0013042542],"genre_scores_gemma":[0.9966461,0.0021498897,0.000008122694,0.0002005988,0.00010809,0.00023270227,0.00005327198,0.000041337185,0.00055990624],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98702717,0.0068653063,0.00045100073,0.0013609879,0.0033630475,0.000932479],"domain_scores_gemma":[0.971134,0.026884582,0.00016199729,0.0011977298,0.00040985778,0.00021181763],"candidate_categories":["metaresearch","bibliometrics","sts","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.011509008,0.00028658286,0.0005341157,0.0648767,0.001635415,0.0005538982,0.0016545656,0.00043057796,0.0005898057],"category_scores_gemma":[0.011238356,0.00016829907,0.00023771242,0.1381179,0.0020017335,0.000076085555,0.006770358,0.005820064,0.000013794434],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016778426,0.0025240616,0.3359099,0.005131726,0.0025699276,0.00025144796,0.01414791,0.00085094024,0.27509046,0.0041782223,0.09647054,0.261197],"study_design_scores_gemma":[0.0007550374,0.0009621661,0.9354809,0.0004330084,0.00010950538,0.000003521144,0.006776528,0.002295145,0.035756353,0.0028434307,0.0140620265,0.00052236515],"about_ca_topic_score_codex":0.00046844463,"about_ca_topic_score_gemma":0.000035888886,"teacher_disagreement_score":0.599571,"about_ca_system_score_codex":0.00012126361,"about_ca_system_score_gemma":0.00025086335,"threshold_uncertainty_score":0.9996643},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4283704156","doi":"10.1177/01655515221097623","title":"Do retraction practices work effectively? Evidence from citations of psychological retracted articles","year":2022,"lang":"en","type":"article","venue":"Journal of Information Science","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"National Social Science Fund of China; Ministry of Education of the People's Republic of China","keywords":"Citation; Credibility; Quality (philosophy); Psychology; Inference; Web of science; Process (computing); Computer science; MEDLINE; World Wide Web; Political science; Law","score_opus":0.09565337181427233,"score_gpt":0.42321451935441334,"score_spread":0.327561147540141,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283704156","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9875714,0.00008106076,0.0038167692,0.0027504363,0.001033919,0.00011573362,0.000007217382,0.000011874991,0.00461158],"genre_scores_gemma":[0.9956241,0.00015307794,0.0038733813,0.00021250903,0.00011567967,0.0000030487686,7.5350175e-7,0.0000010831568,0.000016358095],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.9966473,0.00049805624,0.00062646565,0.00008125265,0.0019839916,0.00016292816],"domain_scores_gemma":[0.9939916,0.0020103322,0.0026815527,0.00009941132,0.0011162836,0.00010082949],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0142042,0.00005335016,0.00012461608,0.00037062084,0.0009486826,0.00016492944,0.0006367729,0.00013974709,0.00050804176],"category_scores_gemma":[0.02875283,0.000045041492,0.000061473205,0.0023187804,0.0004764455,0.009855596,0.000050929662,0.0015101377,0.000016556256],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002224311,0.00062007224,0.1937134,0.00003393719,0.00007904367,0.000015185299,0.49381346,0.0047389953,0.03737299,0.03401929,0.0231798,0.21018954],"study_design_scores_gemma":[0.0005088668,0.0008146914,0.857434,0.0002517309,0.000053292897,0.000056428464,0.10253419,0.0005857137,0.0031142775,0.010454503,0.023955097,0.00023723436],"about_ca_topic_score_codex":0.0003056212,"about_ca_topic_score_gemma":0.000008619556,"teacher_disagreement_score":0.6637206,"about_ca_system_score_codex":0.00023679914,"about_ca_system_score_gemma":0.00043435243,"threshold_uncertainty_score":0.9794284},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4283719256","doi":"10.5539/ass.v18n7p29","title":"Co-citation and Bibliographic Coupling Based on Connected Papers: Review of Public Opinion Research in a Broad Sense in the West","year":2022,"lang":"en","type":"article","venue":"Asian Social Science","topic":"Social Media and Politics","field":"Social Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Public opinion; Politics; Citation; Political science; Sociology; Law","score_opus":0.13811496094208575,"score_gpt":0.4426208273037746,"score_spread":0.30450586636168886,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283719256","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.70532775,0.00205944,0.000002522479,0.07502037,0.0006253149,0.0017593071,0.00004833961,0.000041866704,0.21511506],"genre_scores_gemma":[0.99716324,0.0016471174,0.000008835886,0.0009988503,0.00009494146,0.000074152566,0.000007048784,0.0000031762243,0.000002646956],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.99583554,0.0015456324,0.00021804683,0.00022841725,0.0016944091,0.00047794203],"domain_scores_gemma":[0.9984061,0.0011519414,0.0000911044,0.0001031365,0.00016514759,0.00008258522],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.011728197,0.000060513426,0.0001399554,0.0014970198,0.0016264393,0.00010012483,0.00037470672,0.000046647303,0.000065808315],"category_scores_gemma":[0.0022088292,0.000055584365,0.00003572908,0.020675154,0.0019639852,0.00017043145,0.000034141445,0.00036529973,0.0000018292459],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000052336654,0.00067144656,0.15247405,0.00074842677,0.0000075118533,0.00003486566,0.32784167,0.0000045421993,0.001362374,0.42261162,0.0027369421,0.0914542],"study_design_scores_gemma":[0.00093087996,0.00036220992,0.43939686,0.00093820476,0.000007661729,0.0000015018013,0.5132823,0.00015533366,0.000029060107,0.0065753497,0.037995584,0.00032508548],"about_ca_topic_score_codex":0.0025342142,"about_ca_topic_score_gemma":0.00080316706,"teacher_disagreement_score":0.41603628,"about_ca_system_score_codex":0.00018316951,"about_ca_system_score_gemma":0.00085279887,"threshold_uncertainty_score":0.9996733},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4283802616","doi":"10.5206/cjilsrcsib.v45i1.14149","title":"Measuring the prevalence of open access in Canada: A national comparison","year":2022,"lang":"en","type":"article","venue":"Canadian Journal of Information and Library Science","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"Université de Montréal","funders":"Université du Québec à Montréal","keywords":"Lagging; Political science; Compliance (psychology); Discipline; Momentum (technical analysis); Library science; Medicine; Business; Psychology; Computer science","score_opus":0.5376084805342523,"score_gpt":0.5023064140048592,"score_spread":0.03530206652939305,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283802616","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98193336,0.0005344467,0.00008430384,0.004769085,0.00040024484,0.00016189086,0.00009376124,5.991246e-7,0.012022306],"genre_scores_gemma":[0.99910307,0.000023971676,0.00010683007,0.0006566396,0.0000063825137,0.0000017223244,4.201792e-7,8.2815836e-7,0.000100108766],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9941161,0.00010755877,0.00074021047,0.000086455286,0.0047366074,0.00021301521],"domain_scores_gemma":[0.99732393,0.00060004863,0.0005658047,0.00013586585,0.00095707225,0.00041727207],"candidate_categories":["bibliometrics","scholarly_communication","open_science"],"consensus_categories":["bibliometrics","scholarly_communication"],"category_scores_codex":[0.014976857,0.000039901646,0.00011642354,0.019613521,0.00050581526,0.0043733413,0.008083475,0.0000074431005,0.000798042],"category_scores_gemma":[0.0065942663,0.000025645973,0.000017949405,0.06342222,0.00021942589,0.01602819,0.0011283635,0.00021900532,0.0000010910301],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000891439,0.0000044839517,0.9500354,0.0000068296604,0.0000012571198,0.0000026974772,0.00069537986,0.004087092,0.00000498595,0.002925512,0.015519097,0.026708364],"study_design_scores_gemma":[0.00014803381,0.0000319096,0.9424856,0.000008006652,5.09194e-7,0.000026243235,0.0021805211,0.011115415,0.00008972547,0.0013207501,0.042552426,0.000040913925],"about_ca_topic_score_codex":0.3814228,"about_ca_topic_score_gemma":0.28490916,"teacher_disagreement_score":0.096513614,"about_ca_system_score_codex":0.0003702256,"about_ca_system_score_gemma":0.042043403,"threshold_uncertainty_score":0.9977341},"labels":[{"model":"gemma","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4285083298","doi":"10.1108/mbr-01-2022-0001","title":"A bibliometric analysis and future research opportunities in <i>Multinational Business Review</i>","year":2022,"lang":"en","type":"article","venue":"Multinational Business Review","topic":"International Business and FDI","field":"Business, Management and Accounting","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Internationalization; Multinational corporation; Originality; International business; Internalization theory; Globalization; Value (mathematics); Bibliographic coupling; Conceptual framework; Scope (computer science); Business; Regional science; Marketing; Political science; Sociology; Economics; Management; International trade; Social science; Computer science; Qualitative research","score_opus":0.11673759832929219,"score_gpt":0.3530353205880919,"score_spread":0.2362977222587997,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285083298","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.025470432,0.70392096,0.00049917033,0.23566914,0.0018282197,0.0041399817,0.0003548631,0.00027183443,0.027845414],"genre_scores_gemma":[0.2439708,0.6882381,0.00089584343,0.04965315,0.003980763,0.0032915953,0.0069930893,0.00016267129,0.0028140356],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9948777,0.00018434486,0.0010662524,0.00075702305,0.002588613,0.0005260488],"domain_scores_gemma":[0.9922807,0.00032080765,0.00058548147,0.000378053,0.0063987104,0.000036242054],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.004602845,0.0003844902,0.00084044936,0.05652265,0.00078857446,0.00030053945,0.0007314358,0.00007050397,0.004124579],"category_scores_gemma":[0.0016959894,0.0003668083,0.00018821887,0.25677386,0.00015889978,0.0016929352,0.000996125,0.0005223964,0.000084966254],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001465713,0.0019837064,0.0981888,0.05192187,0.00056784466,0.00035092793,0.000046091875,0.0007733536,0.000050111055,0.1572936,0.08601006,0.60266703],"study_design_scores_gemma":[0.00041855458,0.0000029934208,0.45824212,0.0013074814,0.00018748002,0.00001829822,0.000038208498,0.0012358252,1.9409418e-7,0.00036960817,0.53788096,0.0002982609],"about_ca_topic_score_codex":0.00074974797,"about_ca_topic_score_gemma":0.00008850288,"teacher_disagreement_score":0.6023688,"about_ca_system_score_codex":0.0002354181,"about_ca_system_score_gemma":0.00022409693,"threshold_uncertainty_score":0.9998784},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4285083393","doi":"10.1002/cjas.1687","title":"<i>Canadian Journal of Administrative Sciences</i>: A Retrospective Review Using Bibliometric Analysis","year":2022,"lang":"en","type":"review","venue":"Canadian Journal of Administrative Sciences / Revue Canadienne des Sciences de l Administration","topic":"Corporate Social Responsibility Reporting","field":"Business, Management and Accounting","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Productivity; Knowledge management; Diversity (politics); Corporate governance; Bibliometrics; Human resources; Business; Political science; Management; Public relations; Library science; Computer science; Economics","score_opus":0.26735372469434215,"score_gpt":0.40115176149192633,"score_spread":0.13379803679758417,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285083393","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.048450246,0.93480283,0.00031722593,0.0006173369,0.0022628156,0.0014118176,0.0003079285,0.000018677427,0.011811091],"genre_scores_gemma":[0.47796455,0.5109933,0.006689392,0.0010267124,0.0027309724,0.000051204497,0.00007908775,0.00013105948,0.00033370496],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9899832,0.0007944119,0.004322443,0.0012676269,0.0014901429,0.0021421614],"domain_scores_gemma":[0.9806065,0.0010214073,0.013332592,0.00052090344,0.0029491452,0.0015694625],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics","sts"],"category_scores_codex":[0.02451976,0.0009660673,0.003412977,0.06699388,0.005925509,0.0033021136,0.0038484947,0.00037284795,0.0015236111],"category_scores_gemma":[0.018869782,0.00088194077,0.0018941059,0.19979924,0.008902316,0.004240435,0.000107000706,0.0015269063,0.000008347969],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.00024603965,0.00090949866,0.1650931,0.05976988,0.0074421177,0.03900049,0.015139926,0.0033556286,0.00005271422,0.09613109,0.0058693583,0.60699016],"study_design_scores_gemma":[0.0007163014,0.009008621,0.005941823,0.03536535,0.017713001,0.017022202,0.019692963,0.0012723522,0.000014867885,0.021733586,0.8672336,0.0042853495],"about_ca_topic_score_codex":0.07605619,"about_ca_topic_score_gemma":0.76929325,"teacher_disagreement_score":0.86136425,"about_ca_system_score_codex":0.006737218,"about_ca_system_score_gemma":0.14128666,"threshold_uncertainty_score":0.9993891},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4285156024","doi":"10.7202/1089191ar","title":"Bibliothèques universitaires et usage de la bibliométrie dans l’évaluation de la performance de la recherche","year":2022,"lang":"fr","type":"article","venue":"Documentation et bibliothèques","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"Université du Québec en Outaouais","funders":"","keywords":"Political science; Humanities; Philosophy","score_opus":0.46655375761439644,"score_gpt":0.573577256834977,"score_spread":0.10702349922058052,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285156024","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8255508,0.019552927,0.0509993,0.0060670124,0.0006088234,0.0006672779,0.00026461875,0.00022075194,0.096068524],"genre_scores_gemma":[0.77402806,0.18204089,0.023650397,0.0015840128,0.00014218705,0.00013040143,0.00006927263,0.0000764901,0.018278297],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9324786,0.05362441,0.0013060094,0.0014661029,0.0093639605,0.0017609183],"domain_scores_gemma":[0.93054724,0.06581346,0.0007013737,0.00082062726,0.0013488171,0.00076851295],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.19076061,0.00050766795,0.000525194,0.34932098,0.001413164,0.04266511,0.0028719937,0.000737353,0.018730234],"category_scores_gemma":[0.028203275,0.00052627403,0.00034328917,0.5576176,0.0013608892,0.020532316,0.0014621929,0.002564028,0.00012502578],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.00037707412,0.0016081049,0.4979153,0.0001771074,0.00022351564,0.00031099122,0.021900725,0.013039504,0.020929918,0.047644127,0.23988077,0.15599285],"study_design_scores_gemma":[0.0031855535,0.0012590961,0.3695856,0.00016217717,0.00016474588,0.00049740396,0.016608689,0.043358225,0.018102445,0.11721484,0.4286762,0.0011850315],"about_ca_topic_score_codex":0.0041996622,"about_ca_topic_score_gemma":0.00006510136,"teacher_disagreement_score":0.20829667,"about_ca_system_score_codex":0.003859483,"about_ca_system_score_gemma":0.006662601,"threshold_uncertainty_score":0.99996454},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4285317767","doi":"10.37108/shaut.v13i2.565","title":"ANALISIS BIBLIOMETRIC DARI PROGRAM HIBAH (BIBLIOMETRIC OF GRANTS PROGRAM)","year":2021,"lang":"en","type":"article","venue":"Shaut Al-Maktabah Jurnal Perpustakaan Arsip dan Dokumentasi","topic":"Mental Health and Well-being","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Library science; Bibliometrics; Citation; Political science; Computer science; MEDLINE","score_opus":0.04248360514523474,"score_gpt":0.3956574326587859,"score_spread":0.35317382751355114,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285317767","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94489837,0.0044857976,0.000038282535,0.0020562035,0.0014607627,0.0021317014,0.00012835462,0.00046542895,0.04433511],"genre_scores_gemma":[0.9899597,0.004230512,0.0028978558,0.0013432233,0.00043199526,0.00017374549,0.0002241722,0.00009248646,0.0006462752],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99103355,0.0008391985,0.001560944,0.0011517926,0.0032565566,0.002157962],"domain_scores_gemma":[0.99553144,0.00043408741,0.0007978734,0.0007763856,0.00090922316,0.0015510167],"candidate_categories":["metaepi_narrow","bibliometrics","sts","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0025967308,0.0006056966,0.0009830402,0.08668057,0.0014488488,0.0008103786,0.0010643313,0.0003271863,0.0012278936],"category_scores_gemma":[0.0006274655,0.0006169652,0.0006017984,0.35441482,0.00075089955,0.0012142732,0.00038167828,0.00080515747,0.000095999436],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013446531,0.0028179595,0.30782822,0.00039400696,0.00022918182,0.00038333196,0.0027320257,0.0000065258237,0.00089681544,0.003206316,0.011382279,0.6699889],"study_design_scores_gemma":[0.0052780677,0.0021666624,0.3767393,0.00086221244,0.0005313791,0.00021681827,0.0257847,0.00029833894,0.015314966,0.0011910789,0.56938684,0.002229666],"about_ca_topic_score_codex":0.0025920267,"about_ca_topic_score_gemma":0.00057715096,"teacher_disagreement_score":0.66775924,"about_ca_system_score_codex":0.00067222334,"about_ca_system_score_gemma":0.001274833,"threshold_uncertainty_score":0.9998511},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4285490980","doi":"10.1007/s11192-022-04466-8","title":"Scientometric portraits of recognized scientists: a structured literature review","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Tech University","funders":"Macquarie University","keywords":"Portrait; Scientometrics; Data science; Library science; Sociology; Information retrieval; History; Computer science; Art history","score_opus":0.4423497706215747,"score_gpt":0.5649995372104198,"score_spread":0.12264976658884513,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285490980","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.45691952,0.46987295,0.0047461623,0.003640614,0.016643472,0.004522085,0.0039129737,0.00030529828,0.039436907],"genre_scores_gemma":[0.963683,0.014189549,0.0064422474,0.0016464108,0.00013723849,0.00013118959,0.00013681436,0.00006159011,0.0135719525],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9337935,0.0012845664,0.0028235707,0.002651866,0.057525113,0.0019214149],"domain_scores_gemma":[0.9713778,0.00540963,0.0021901133,0.0031631733,0.016285114,0.0015741899],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.11765935,0.00045941147,0.0012589478,0.4470861,0.00156462,0.0028881703,0.010750653,0.00014708855,0.006618031],"category_scores_gemma":[0.19142734,0.00035857785,0.00076208904,0.92149866,0.0007629203,0.0013587951,0.0052786535,0.001158069,0.00018072451],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000100950754,0.0013442521,0.06375939,0.00061255647,0.0000837021,0.0002205746,0.000637607,0.00024680689,0.0036903657,0.0037525299,0.35143647,0.5741148],"study_design_scores_gemma":[0.0043587787,0.0018786506,0.20685683,0.00073398324,0.00013493065,0.0007431741,0.00093674136,0.0039687413,0.0029563599,0.014246623,0.76117736,0.0020077955],"about_ca_topic_score_codex":0.000022791544,"about_ca_topic_score_gemma":0.0000038039282,"teacher_disagreement_score":0.572107,"about_ca_system_score_codex":0.0007086142,"about_ca_system_score_gemma":0.0011621525,"threshold_uncertainty_score":0.99988663},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4285728602","doi":"10.3389/fpsyt.2022.884600","title":"A bibliometrics analysis and visualization of autism spectrum disorder","year":2022,"lang":"en","type":"article","venue":"Frontiers in Psychiatry","topic":"Autism Spectrum Disorder Research","field":"Neuroscience","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"State Administration of Traditional Chinese Medicine of the People's Republic of China; National Natural Science Foundation of China","keywords":"Autism spectrum disorder; Visualization; Bibliometrics; Autism; Psychology; Medicine; Clinical psychology; Psychiatry; Computer science; Artificial intelligence; Data mining","score_opus":0.012985797316245103,"score_gpt":0.2942900300952375,"score_spread":0.2813042327789924,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285728602","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8802556,0.003658793,0.10231128,0.007524713,0.0036221396,0.00050121144,0.00007944958,0.0000795306,0.0019673174],"genre_scores_gemma":[0.9972776,0.0005194277,0.0017854125,0.0001526991,0.000011004891,0.00003794422,0.000008606624,0.000023444492,0.00018381674],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9978655,0.0002813357,0.0003342048,0.00048821085,0.000682764,0.00034797264],"domain_scores_gemma":[0.9994007,0.00005339629,0.00015012201,0.00032666963,0.00000270821,0.00006644097],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0003766709,0.00012694292,0.0002910238,0.015856056,0.00021080906,0.0000348518,0.00037817153,0.000044533554,0.00029315494],"category_scores_gemma":[0.00008860534,0.00014378333,0.00009181432,0.053123243,0.00013307034,0.00014560208,0.00029060713,0.00025637285,0.0000017123534],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000046767083,0.00031653102,0.93138045,0.000027063308,0.000038022143,0.00000360267,0.0003125151,0.00052708515,0.00010417662,0.06399276,0.0023772612,0.00087377854],"study_design_scores_gemma":[0.0015036183,0.00034498348,0.77278507,0.0000071356053,0.00009933567,0.000009871778,0.0008786509,0.03428481,0.0004729991,0.18690246,0.0023270897,0.00038398916],"about_ca_topic_score_codex":0.00016567808,"about_ca_topic_score_gemma":0.00007967961,"teacher_disagreement_score":0.15859538,"about_ca_system_score_codex":0.00007386006,"about_ca_system_score_gemma":0.00009211726,"threshold_uncertainty_score":0.9952984},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4286434472","doi":"10.3390/su14148857","title":"International Research Progress in School Travel and Behavior: A Literature Review and Bibliometric Analysis","year":2022,"lang":"en","type":"review","venue":"Sustainability","topic":"Urban Transport and Accessibility","field":"Social Sciences","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Natural Science Foundation of China","keywords":"Citation; Bibliometrics; Travel behavior; Urban sprawl; Regional science; Publishing; Citation analysis; Sustainability; Business; Geography; Political science; Transport engineering; Urban planning; Computer science; Engineering; Library science","score_opus":0.0768624551240452,"score_gpt":0.48225612514755545,"score_spread":0.40539367002351023,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4286434472","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0010892116,0.99418604,5.9481454e-7,0.0005159595,0.00006738583,0.0031250268,0.00012802228,0.00002614147,0.00086162664],"genre_scores_gemma":[0.003375088,0.994504,0.0000236273,0.0000169352,0.00007520923,0.0012971319,0.00018956873,0.000015012554,0.0005034625],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99377775,0.0027099126,0.00078739855,0.0010461974,0.0010989353,0.00057979976],"domain_scores_gemma":[0.99745524,0.00059302466,0.00021356942,0.00053205667,0.00091728306,0.00028883337],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0126684625,0.00028244732,0.0014917924,0.03819367,0.0005597162,0.00044704668,0.00076229393,0.00031263952,0.0011234995],"category_scores_gemma":[0.0036670761,0.00024269694,0.00037773256,0.20490922,0.0010015302,0.00044676874,0.0003112835,0.0018309711,7.569416e-7],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000032864439,0.00020616197,0.29768798,0.034896083,0.00008510576,0.000059834732,0.00050702994,3.4081091e-9,2.3023235e-10,0.00017520058,0.00006141764,0.6663179],"study_design_scores_gemma":[0.00007674673,0.000020372749,0.27502263,0.0015680491,0.0016504009,0.0000011781973,0.0005389079,2.2400953e-7,1.2421528e-9,0.00057226146,0.72033435,0.00021486705],"about_ca_topic_score_codex":0.0010893067,"about_ca_topic_score_gemma":0.000892561,"teacher_disagreement_score":0.72027296,"about_ca_system_score_codex":0.001856096,"about_ca_system_score_gemma":0.0017197822,"threshold_uncertainty_score":0.9997896},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4286579620","doi":"10.4018/ijtee.305807","title":"A Bibliometric Analysis of Automated Writing Evaluation in Education Using VOSviewer and CitNetExplorer from 2008 to 2022","year":2022,"lang":"en","type":"article","venue":"International Journal of Technology-Enhanced Education","topic":"Online Learning and Analytics","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Beijing Language and Culture University","keywords":"LEAPS; Bibliometrics; China; Field (mathematics); Service (business); Library science; Computer science; Political science; Marketing","score_opus":0.019867654931889325,"score_gpt":0.36848020044218427,"score_spread":0.34861254551029497,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4286579620","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97954524,0.0014128095,0.013134175,0.0045493883,0.0011723853,0.00010314198,0.0000071115855,0.000043214135,0.000032558335],"genre_scores_gemma":[0.9674038,0.0000948183,0.032200515,0.00014774359,0.00008487032,0.000019767378,0.00002347609,0.000006109203,0.000018896524],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981025,0.00011919449,0.0006562556,0.000229832,0.0007764014,0.00011583495],"domain_scores_gemma":[0.9973891,0.00008934056,0.0007726539,0.00020279615,0.0015008745,0.000045239027],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0009850528,0.00009196208,0.00023758154,0.080392525,0.00006644746,0.00006304874,0.0007290065,0.000055530076,0.00005987739],"category_scores_gemma":[0.0007706739,0.00010114118,0.000073837226,0.06410894,0.000024560555,0.00039418135,0.00022249574,0.00029990743,0.0000010999682],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000022758353,0.0006952276,0.046149902,0.0000059357712,0.00042377232,0.0000019020055,0.0012482437,0.05843977,0.03148292,0.0007555347,0.00045518248,0.86031884],"study_design_scores_gemma":[0.000503597,0.0001801501,0.23296134,0.00017375725,0.00028386732,0.000044614728,0.0033591876,0.7451979,0.0039093294,0.012585973,0.0005762956,0.00022398225],"about_ca_topic_score_codex":0.00010759776,"about_ca_topic_score_gemma":0.000010976411,"teacher_disagreement_score":0.86009485,"about_ca_system_score_codex":0.00047858772,"about_ca_system_score_gemma":0.0008739459,"threshold_uncertainty_score":0.9557839},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4287307882","doi":"10.12688/f1000research.54437.2","title":"Autism spectrum disorder in architecture perspective: a review of the literature and bibliometric assessment of research indexed in Web of Science","year":2022,"lang":"en","type":"review","venue":"F1000Research","topic":"Autism Spectrum Disorder Research","field":"Neuroscience","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Web of science; Social science; Library science; Sociology; Medicine; Computer science; Pathology; Meta-analysis","score_opus":0.10207242834392906,"score_gpt":0.4849486522070718,"score_spread":0.38287622386314274,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4287307882","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021217016,0.98282665,0.0000016332083,0.0031165204,0.00009701677,0.0059760935,0.00030500634,0.000015850695,0.005539551],"genre_scores_gemma":[0.030682493,0.9681754,0.000045246186,0.000019404963,0.000015514497,0.00059833424,0.000007163237,0.00009035263,0.00036605704],"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","domain_scores_codex":[0.97768176,0.00788973,0.0017909949,0.0018849223,0.008974825,0.0017777912],"domain_scores_gemma":[0.9913584,0.005293375,0.0006745451,0.0022476178,0.00019137296,0.00023471408],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","open_science","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.01861772,0.00053806265,0.0024738768,0.09469097,0.00037326818,0.00012818619,0.0058422596,0.0003027807,0.00079012354],"category_scores_gemma":[0.012651997,0.00037998258,0.0004353629,0.36508104,0.007037698,0.00030638417,0.006880307,0.00735799,0.0000039159477],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011031535,0.0029502357,0.0063398965,0.43909618,0.000065288514,0.0002921162,0.003055646,0.00001763666,0.0011543427,0.21993697,0.00027692108,0.32670444],"study_design_scores_gemma":[0.0031247935,0.0022446224,0.04214797,0.42512208,0.00012329723,0.00041118296,0.0008954336,0.00037827253,0.0007276754,0.053486705,0.46963263,0.0017053253],"about_ca_topic_score_codex":0.000727381,"about_ca_topic_score_gemma":0.00027069572,"teacher_disagreement_score":0.4693557,"about_ca_system_score_codex":0.0013380289,"about_ca_system_score_gemma":0.0109796515,"threshold_uncertainty_score":0.99986523},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4289874625","doi":"10.30518/jav.1121377","title":"Contribution of Scientific Production to Air Logistics: A Bibliometric Analysis from the 70s to the Present","year":2022,"lang":"en","type":"article","venue":"Journal of Aviation","topic":"Aviation Industry Analysis and Trends","field":"Economics, Econometrics and Finance","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Popularity; Web of science; Subject (documents); Bibliometrics; China; Field (mathematics); Data science; Systematic review; Computer science; Library science; Political science; Geography; MEDLINE; Archaeology","score_opus":0.04653019373702898,"score_gpt":0.26469612130017167,"score_spread":0.21816592756314268,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4289874625","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8283014,0.0005852559,0.12999518,0.03759539,0.0017044705,0.0011466622,0.00043583184,0.0000075581056,0.00022821722],"genre_scores_gemma":[0.9988771,0.000007446961,0.0001241752,0.00013683815,0.00034665273,0.00008818422,0.000043886565,0.0000041572844,0.00037153947],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.9984884,0.00009101963,0.00087119447,0.00018409838,0.00025421137,0.00011106173],"domain_scores_gemma":[0.99763125,0.00012839968,0.0014432184,0.0003043814,0.0004411332,0.000051637468],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.002957167,0.000067212815,0.00026587376,0.013493071,0.00039307782,0.00009584139,0.00032601203,0.000028140643,0.0004371013],"category_scores_gemma":[0.0010827512,0.000047891626,0.0002283497,0.06553384,0.000023441396,0.00016975772,0.0000780527,0.0001848582,0.000029042156],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012368755,0.00027677845,0.3657406,0.0000027556707,0.0009803642,0.0000010207825,0.0009228218,0.5869899,0.000115202376,0.008685039,0.029868348,0.0062934905],"study_design_scores_gemma":[0.00025216554,0.00017496118,0.879367,0.0000037951143,0.0002732501,0.000001886961,0.00027992402,0.0043432005,0.00029581552,0.003511323,0.11141171,0.00008497174],"about_ca_topic_score_codex":0.0001979939,"about_ca_topic_score_gemma":0.0000517775,"teacher_disagreement_score":0.5826467,"about_ca_system_score_codex":0.00019797015,"about_ca_system_score_gemma":0.000032758147,"threshold_uncertainty_score":0.9976881},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4290723710","doi":"10.21203/rs.3.rs-1933790/v1","title":"Global Research Trends at the Intersection of Acupuncture and Headache Disorders: A Bibliometric Analysis","year":2022,"lang":"en","type":"preprint","venue":"Research Square","topic":"Complementary and Alternative Medicine Studies","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Impact","funders":"McMaster University","keywords":"Intersection (aeronautics); Acupuncture; Medicine; Alternative medicine; Geography; Cartography; Pathology","score_opus":0.19278944257630723,"score_gpt":0.5341937433659565,"score_spread":0.34140430078964923,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4290723710","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9666337,0.013880127,0.0000836573,0.006266665,0.00017758795,0.0010552544,0.0005247416,0.000024814097,0.011353458],"genre_scores_gemma":[0.9917521,0.0034299875,0.000029902429,0.000021003149,0.00031100767,0.0003111692,0.00055989716,0.000017313927,0.0035676532],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99346614,0.0016072881,0.00043042342,0.0007908986,0.0030571243,0.00064810645],"domain_scores_gemma":[0.9962391,0.0017392059,0.00012036568,0.00089110277,0.0008247247,0.00018546237],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0053158845,0.00031122347,0.00085978495,0.051797938,0.00079266925,0.000054520868,0.00043915346,0.00006274905,0.004473322],"category_scores_gemma":[0.0014214586,0.00015926064,0.00048278377,0.16651817,0.0010306693,0.000040705017,0.0061942865,0.002228591,0.000003401133],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00078055135,0.0003274717,0.85460824,0.001020047,0.0036892467,0.00003911953,0.0017354407,0.0000760748,0.000035280107,0.00019376558,0.016235383,0.1212594],"study_design_scores_gemma":[0.00087613176,0.0021915834,0.9522038,0.00024547137,0.00066616765,0.000010736041,0.013862531,0.0004749546,0.000022513148,0.0013316155,0.027983783,0.00013069811],"about_ca_topic_score_codex":0.0064230976,"about_ca_topic_score_gemma":0.009694001,"teacher_disagreement_score":0.1211287,"about_ca_system_score_codex":0.0012089164,"about_ca_system_score_gemma":0.00020845585,"threshold_uncertainty_score":0.9964367},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4291982736","doi":"","title":"A Very Long Embargo: Journal Choice Reveals Active Non-Compliance with Funder Open Access Policies by Australian and Canadian Neuroscientists","year":2018,"lang":"en","type":"article","venue":"Spectrum Research Repository (Concordia University)","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Concordia University","funders":"","keywords":"Compliance (psychology); Political science; Public economics; Psychology; Public administration; Public relations; Economics; Social psychology","score_opus":0.4513140322787154,"score_gpt":0.5260346446697138,"score_spread":0.07472061239099842,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4291982736","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9289218,0.000055487526,0.00025307309,0.0035135874,0.0006270278,0.000637174,0.00006555517,0.000015884283,0.065910414],"genre_scores_gemma":[0.9179739,0.00008678402,0.000034233868,0.000096276824,0.00027871062,0.00000229383,0.000002227934,0.00002205456,0.08150352],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98936415,0.0010139847,0.00042184882,0.0013749577,0.005640888,0.0021841899],"domain_scores_gemma":[0.9901372,0.0017493857,0.00030627675,0.0010565328,0.0036479717,0.0031026371],"candidate_categories":["bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.008500346,0.00027550763,0.000458903,0.04005377,0.004204301,0.024128351,0.010518026,0.00017003922,0.00027450605],"category_scores_gemma":[0.0036856392,0.00022566406,0.00009761263,0.09715174,0.0023350623,0.0053581665,0.0038710928,0.001195173,0.00008961201],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026846922,0.00009318191,0.9021432,0.000010982027,0.000061950814,0.0011592173,0.00019400638,0.0000033672109,0.0042910664,0.000259351,0.08823765,0.003277591],"study_design_scores_gemma":[0.0008442884,0.000769649,0.8986996,0.000047417067,0.00000949922,0.00022494174,0.0008847358,0.00011125845,0.005696054,0.0006508327,0.09175076,0.0003109933],"about_ca_topic_score_codex":0.38666764,"about_ca_topic_score_gemma":0.3215717,"teacher_disagreement_score":0.06509593,"about_ca_system_score_codex":0.0014679633,"about_ca_system_score_gemma":0.0034258387,"threshold_uncertainty_score":0.99709207},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W4293224259","doi":"10.2196/39365","title":"An Evaluation of Primary Studies Published in Predatory Journals Included in Systematic Reviews From High-Impact Dermatology Journals: Cross-sectional Study","year":2022,"lang":"en","type":"article","venue":"JMIR Dermatology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Systematic review; Publishing; MEDLINE; Impact factor; Medicine; Family medicine; Grey literature; Data extraction; Directory; Publication bias; Meta-analysis; Computer science; Pathology; Political science","score_opus":0.7449597743669143,"score_gpt":0.6623973011713971,"score_spread":0.08256247319551724,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4293224259","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.964667,0.03050346,0.00004287839,0.00033249517,0.0008670629,0.003407606,0.000070463684,0.000013577782,0.00009543519],"genre_scores_gemma":[0.9980654,0.00038479906,0.0000875321,0.00012696598,0.000039617713,0.0012387959,0.00002246772,0.000016228554,0.000018162711],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.93260604,0.036533084,0.008846189,0.0016202773,0.019230863,0.0011635636],"domain_scores_gemma":[0.97950107,0.010691037,0.0029268607,0.0013547319,0.00522611,0.0003001628],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.113814965,0.0003032121,0.0025955676,0.063987754,0.00040072636,0.0011970135,0.0033171722,0.00022440411,0.0027711086],"category_scores_gemma":[0.0687687,0.00021318652,0.00026452076,0.07393526,0.00023804723,0.002066005,0.001513511,0.0011309582,0.00004388716],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000987291,0.0014890261,0.99197805,0.00028474652,0.00009861171,0.00014511215,0.0016569282,0.0011982449,0.00031191445,0.000014341901,0.0018303295,0.0008939801],"study_design_scores_gemma":[0.00271722,0.00022326429,0.98061085,0.00011212908,0.000023976438,0.0005043672,0.0032576197,0.0049860617,0.000044723216,0.0072676074,0.00006359731,0.00018861107],"about_ca_topic_score_codex":0.00031797602,"about_ca_topic_score_gemma":0.00040201415,"teacher_disagreement_score":0.046895072,"about_ca_system_score_codex":0.001479972,"about_ca_system_score_gemma":0.0010420382,"threshold_uncertainty_score":0.99983984},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","research_integrity","scholarly_communication"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4294405730","doi":"10.1007/s43069-022-00156-6","title":"A Literature Review on Correlation Clustering: Cross-disciplinary Taxonomy with Bibliometric Analysis","year":2022,"lang":"en","type":"review","venue":"Operations Research Forum","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":9,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Natural Sciences and Engineering Research Council of Canada; Mitacs","keywords":"Cluster analysis; Taxonomy (biology); Discipline; Computer science; Data science; Set (abstract data type); Cross disciplinary; Bibliometrics; Information retrieval; Data mining; Management science; Artificial intelligence; Social science; Sociology; Engineering; Biology","score_opus":0.16849762507913518,"score_gpt":0.4663325407691728,"score_spread":0.2978349156900376,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4294405730","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000015509495,0.9735361,0.012901379,0.000118109725,0.000039135837,0.0031295265,0.00036024983,0.00007225281,0.009841701],"genre_scores_gemma":[0.00010067499,0.9759228,0.0012091249,0.0000290516,0.00022605651,0.007679549,0.008928642,0.00007759242,0.0058264886],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9949032,0.0011227021,0.0009223257,0.0010190716,0.0012836573,0.00074901554],"domain_scores_gemma":[0.99690056,0.0005452807,0.0002183143,0.0015755342,0.0005606442,0.00019967007],"candidate_categories":["metaepi_narrow","bibliometrics","sts","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0019607625,0.00054496044,0.0017856124,0.068598755,0.0017291866,0.0011970232,0.0009796082,0.00013217259,0.006044296],"category_scores_gemma":[0.00006411453,0.00040631279,0.0011570096,0.29313362,0.00013240922,0.0004003675,0.00088195136,0.0021537,0.000120290206],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011031907,0.00029864002,0.00040577754,0.0051746834,0.0022867958,0.000019569756,0.000018761237,0.00567462,2.1730514e-8,0.0022049034,0.017046733,0.96685845],"study_design_scores_gemma":[0.00008855455,0.00022783862,0.000019822153,0.015184993,0.0015710786,0.000005626716,0.000021006452,0.006728745,7.4484134e-8,0.000027634313,0.9757156,0.0004090551],"about_ca_topic_score_codex":0.00008361175,"about_ca_topic_score_gemma":0.00011812118,"teacher_disagreement_score":0.9664494,"about_ca_system_score_codex":0.00038689745,"about_ca_system_score_gemma":0.00053664745,"threshold_uncertainty_score":0.99983984},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4295234397","doi":"10.3389/fnhum.2022.981571","title":"A scientometric analysis of the 100 most cited articles on magnetic resonance guided focused ultrasound","year":2022,"lang":"en","type":"review","venue":"Frontiers in Human Neuroscience","topic":"Ultrasound and Hyperthermia Applications","field":"Engineering","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Magnetic resonance imaging; Medical physics; Medicine; Nuclear magnetic resonance; Physics; Radiology","score_opus":0.05917727419740901,"score_gpt":0.29538274094329786,"score_spread":0.23620546674588885,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4295234397","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0040716953,0.9924667,0.00048541726,0.0000056995173,0.00092951034,0.0007546051,0.0001386278,0.0000967001,0.0010510843],"genre_scores_gemma":[0.0068148784,0.99183404,0.000259681,0.0000755666,0.000025706531,0.0002608875,0.000017641138,0.00005383153,0.00065777946],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99777156,0.00013319535,0.00058749603,0.0005395155,0.00060363364,0.00036457117],"domain_scores_gemma":[0.9985697,0.0002448689,0.00015935433,0.0009409798,0.000019879208,0.00006520804],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00032809412,0.00027438186,0.0008095443,0.003603529,0.0003246507,0.00008576004,0.0015056191,0.0000745865,0.00008019534],"category_scores_gemma":[0.00033082892,0.00021634245,0.0003239542,0.038559955,0.00035784987,0.00006639911,0.000088223285,0.00040165812,9.0331844e-7],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000027922779,0.00036476812,0.004265713,0.0012115655,0.000103249105,0.000009584281,0.00051691104,0.03532123,0.0007969542,0.0008336719,0.009585741,0.9469878],"study_design_scores_gemma":[0.000096606476,0.00003919702,0.00414088,0.00022989418,0.00054325355,0.0000035950782,0.0000132882915,0.0055513186,0.000012399371,0.00007298635,0.98902154,0.00027503647],"about_ca_topic_score_codex":0.000014280059,"about_ca_topic_score_gemma":0.000010835344,"teacher_disagreement_score":0.9794358,"about_ca_system_score_codex":0.00021687556,"about_ca_system_score_gemma":0.000060959625,"threshold_uncertainty_score":0.9818759},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4296077872","doi":"10.21203/rs.3.rs-2057322/v1","title":"Global Trends and Prospects in Research of Artificial Cornea Over Past 20 Years: A Bibliometric and Visualized Analysis","year":2022,"lang":"en","type":"preprint","venue":"Research Square","topic":"Corneal surgery and disorders","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Natural Science Foundation of China","keywords":"Cornea; China; Perspective (graphical); Geography; Computer science; Artificial intelligence; Medicine; Ophthalmology; Archaeology","score_opus":0.1679558855161167,"score_gpt":0.5116451288237108,"score_spread":0.3436892433075941,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4296077872","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99036825,0.0052764188,0.0000059452323,0.00041713187,0.00005205882,0.00089544524,0.0003738817,0.000027271912,0.0025836132],"genre_scores_gemma":[0.99705625,0.0017755113,0.000041530493,0.000011824323,0.00010112389,0.00022231648,0.00027938944,0.000025602212,0.00048644198],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99338996,0.0012470054,0.0006032387,0.0010071624,0.0028914302,0.0008612121],"domain_scores_gemma":[0.9971264,0.0011306489,0.000111094225,0.00068912597,0.00054510013,0.0003976441],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.006594578,0.0002317561,0.0009948278,0.0774718,0.00015065743,0.00016007558,0.00023068515,0.00031158607,0.0018510374],"category_scores_gemma":[0.0019138225,0.00022981143,0.0002475738,0.16521716,0.0005652266,0.00006905347,0.0020198268,0.0017360613,0.0000038524354],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002073198,0.00080898736,0.93523884,0.000977165,0.0004259764,0.000370039,0.00045768978,0.00004562315,0.00008206498,0.00032435212,0.002074728,0.057121336],"study_design_scores_gemma":[0.0011229501,0.0006298992,0.99231285,0.00020760957,0.00013246648,0.0000053609283,0.000886466,0.0013636552,0.000013203055,0.0015409997,0.0016049182,0.00017961657],"about_ca_topic_score_codex":0.0046550916,"about_ca_topic_score_gemma":0.0018978966,"teacher_disagreement_score":0.08774535,"about_ca_system_score_codex":0.00041157895,"about_ca_system_score_gemma":0.0005826174,"threshold_uncertainty_score":0.9990614},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4297242861","doi":"10.3389/fpsyg.2022.969679","title":"Global trends of research on depression in breast cancer: A bibliometric study based on VOSviewer","year":2022,"lang":"en","type":"article","venue":"Frontiers in Psychology","topic":"Cancer survivorship and care","field":"Medicine","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Guizhou Medical University","keywords":"Breast cancer; Depression (economics); Cancer; Psychological intervention; Medicine; Bibliometrics; Quality of life (healthcare); Psychology; Oncology; Psychiatry; Family medicine; Gerontology; Internal medicine; Library science; Nursing","score_opus":0.05402308991387793,"score_gpt":0.4424264694889033,"score_spread":0.38840337957502535,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4297242861","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9754564,0.002318034,0.000053103166,0.0026419454,0.0040084985,0.00049722224,0.00017375182,0.000022382614,0.014828628],"genre_scores_gemma":[0.9981973,0.00014200328,0.00004624967,0.0010272829,0.00007078353,0.00035146737,0.00001590797,0.000016507931,0.0001325073],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9971945,0.0007402812,0.00031648207,0.0005352962,0.0008267947,0.00038664337],"domain_scores_gemma":[0.9991426,0.00006870168,0.00006766759,0.00055737,0.00007821947,0.00008545409],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0012977222,0.00012812081,0.00041983475,0.033452593,0.000052960495,0.000004466563,0.00027014426,0.00009178191,0.00075732486],"category_scores_gemma":[0.000037589223,0.00011958919,0.00006922749,0.07151259,0.000087675224,0.000020676813,0.0000764644,0.00062780426,0.0000021610024],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004743481,0.0019347689,0.7021047,0.000018251365,0.00001557448,0.000118968695,0.00018402345,0.00011855346,0.00000495214,0.0000028338704,0.06890928,0.22184463],"study_design_scores_gemma":[0.0056416723,0.0021157346,0.9834568,0.000081395774,0.000013126635,0.000011772,0.0017942971,0.0002669372,0.000004097071,0.00007450525,0.006447155,0.00009252115],"about_ca_topic_score_codex":0.0022270666,"about_ca_topic_score_gemma":0.0009427569,"teacher_disagreement_score":0.2813521,"about_ca_system_score_codex":0.00059434323,"about_ca_system_score_gemma":0.00011348528,"threshold_uncertainty_score":0.9775024},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4297313018","doi":"10.3389/fonc.2022.947445","title":"Bibliometric analysis of the global research development of bone metastases in prostate cancer: A 22-year study","year":2022,"lang":"en","type":"article","venue":"Frontiers in Oncology","topic":"Bone health and treatments","field":"Medicine","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Basic and Applied Basic Research Foundation of Guangdong Province; Sun Yat-sen University; National Natural Science Foundation of China; Guangdong Science and Technology Department; Bethune Charitable Foundation","keywords":"Prostate cancer; Medicine; Bone metastasis; Cancer; Bibliometrics; Metastasis; Oncology; Radiation therapy; Radiation oncology; Medical physics; Internal medicine; Library science","score_opus":0.0641479677968338,"score_gpt":0.43900494543265217,"score_spread":0.3748569776358184,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4297313018","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9945372,0.0031298199,0.000004868821,0.00043039184,0.0002679924,0.001152122,0.000091933616,0.0000036686488,0.0003820517],"genre_scores_gemma":[0.99648696,0.00017825486,0.002646713,0.000054752487,0.0000033497095,0.00047267746,0.000016689071,0.0000055039122,0.00013508066],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9975722,0.0007375248,0.00058185414,0.00023336352,0.0005441677,0.00033088386],"domain_scores_gemma":[0.9993005,0.000096527685,0.00017542156,0.00024223623,0.00011419448,0.000071162925],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0020488333,0.0000783583,0.00071445445,0.034923077,0.0000826634,0.000001619076,0.00013369191,0.000034157463,0.000095535965],"category_scores_gemma":[0.00019067565,0.00005881326,0.000066380024,0.17192009,0.000101659665,0.000017982795,0.00022841079,0.0002765652,4.7388178e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011013739,0.0028431301,0.9407741,0.000037930116,0.00076566974,0.00008817486,0.0015431613,0.000119509255,0.000004010548,0.000007948991,0.0016946025,0.05102041],"study_design_scores_gemma":[0.003975005,0.0011546056,0.9830762,0.000011089443,0.00033734646,0.0000026162377,0.005845294,0.00007490194,0.00003447536,0.000058349855,0.0053919666,0.00003813081],"about_ca_topic_score_codex":0.0049762544,"about_ca_topic_score_gemma":0.0072418973,"teacher_disagreement_score":0.13699701,"about_ca_system_score_codex":0.0021609976,"about_ca_system_score_gemma":0.0018590824,"threshold_uncertainty_score":0.97601527},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4298140383","doi":"10.3126/ijosh.v12i4.43125","title":"Impacts of Mercury Exposure on Human Health, Safety and Environment: Literature Review and Bibliometric Analysis (1995 to 2021)","year":2022,"lang":"en","type":"article","venue":"International Journal of Occupational Safety and Health","topic":"Mercury impact and mitigation studies","field":"Environmental Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Mercury (programming language); Human health; Environmental health; Geography; Medicine; Computer science","score_opus":0.04319916413411223,"score_gpt":0.38048197180888876,"score_spread":0.33728280767477653,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4298140383","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.62170535,0.2060295,0.0021841747,0.16433889,0.0005603431,0.001293655,0.0029236898,0.000014416997,0.00095000287],"genre_scores_gemma":[0.81858474,0.17370327,0.00045029263,0.007019667,0.000065202,0.0000037774303,0.0000937411,0.000006581002,0.00007273229],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9976499,0.00018594158,0.0008579947,0.00020487144,0.00093421055,0.00016704509],"domain_scores_gemma":[0.99862105,0.00012526702,0.000787742,0.000093017086,0.00006383042,0.00030911274],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0017532689,0.0001382527,0.00046195212,0.0038744751,0.00038957936,0.000027512131,0.0001592017,0.000025221358,0.0010077389],"category_scores_gemma":[0.0000688169,0.00011872945,0.00008821863,0.0054864725,0.00007746691,0.00022206138,0.00020545047,0.00023361586,0.0000028131938],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002319858,0.0005940331,0.7188622,0.00052333897,0.0015696117,0.000028273782,0.005377226,0.0036612079,0.00017599616,0.0031016178,0.0155939385,0.2481927],"study_design_scores_gemma":[0.0005219386,0.00077809155,0.9694884,0.00021264565,0.00006516891,0.0000783226,0.000092761096,0.000012611668,0.0000040224572,0.00012529475,0.028526518,0.000094254174],"about_ca_topic_score_codex":0.00017108768,"about_ca_topic_score_gemma":0.000057269783,"teacher_disagreement_score":0.25062618,"about_ca_system_score_codex":0.00034807858,"about_ca_system_score_gemma":0.00006727952,"threshold_uncertainty_score":0.99990547},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4299584860","doi":"","title":"Goldener, grüner und \"anderer\" Weg zu Open Access: Statistiken für Österreich","year":2015,"lang":"en","type":"article","venue":"E-LIS Repository (University of Naples Federico II)","topic":"Libraries and Information Services","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Political science","score_opus":0.10093271775872135,"score_gpt":0.2991196745373948,"score_spread":0.19818695677867343,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4299584860","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25071985,0.00048895535,0.0001951136,0.0011883965,0.0011129761,0.0003484203,0.00010696595,0.00014263744,0.7456967],"genre_scores_gemma":[0.8476296,0.00008096775,0.000818221,0.00067480607,0.0003808999,0.000001842469,0.00018168826,0.000026614214,0.15020536],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987874,0.00007035403,0.00028062356,0.00025894103,0.00036351613,0.00023915744],"domain_scores_gemma":[0.9986766,0.00004257646,0.0004071812,0.00035950687,0.00031677927,0.00019736393],"candidate_categories":["sts","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00014061214,0.0001926017,0.00033705914,0.00011051291,0.0013595783,0.0023824428,0.0018875394,0.00007779235,0.002418498],"category_scores_gemma":[0.000009710429,0.00019366045,0.00009258757,0.00014229825,0.00027653156,0.008432803,0.0019338375,0.00012173987,0.00009501158],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003972305,0.0002619496,0.0019388151,0.00018340505,0.00041186198,0.000061375016,0.1385346,0.000059738133,0.000070972346,0.12275148,0.7324475,0.002881072],"study_design_scores_gemma":[0.00096894416,0.0001831247,0.0011859434,0.000042025258,0.0000616593,0.0000071715695,0.06986525,0.00024035755,0.00012837222,0.00049444305,0.9265427,0.0002800265],"about_ca_topic_score_codex":0.01258066,"about_ca_topic_score_gemma":0.0030799562,"teacher_disagreement_score":0.59690976,"about_ca_system_score_codex":0.00008468914,"about_ca_system_score_gemma":0.00024221519,"threshold_uncertainty_score":0.9999405},"labels":[{"model":"gemma","categories":["open_science","bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4303516523","doi":"10.1136/bmjhci-2022-100594","title":"Three-year trends in literature on artificial intelligence in ophthalmology and vision sciences: a protocol for bibliometric analysis","year":2022,"lang":"en","type":"article","venue":"BMJ Health & Care Informatics","topic":"Retinal Imaging and Analysis","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; McMaster University","funders":"","keywords":"Scopus; Web of science; Protocol (science); Intersection (aeronautics); Scientific literature; Computer science; Data science; Information retrieval; MEDLINE; Medicine; Engineering; Alternative medicine; Political science","score_opus":0.08356999082243632,"score_gpt":0.4971852159518268,"score_spread":0.4136152251293905,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4303516523","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.81673914,0.00071613136,0.0108823,0.015623918,0.00024238914,0.14995508,0.00028148363,0.00013013674,0.005429389],"genre_scores_gemma":[0.93418634,0.000013428681,0.014888144,0.0014810325,0.00006619002,0.04913835,0.00017780175,0.0000145042995,0.00003420652],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99806297,0.00007257405,0.00089096965,0.0001859042,0.00045587227,0.0003317272],"domain_scores_gemma":[0.99914813,0.00014886234,0.00027993257,0.00021022238,0.00010661424,0.000106249434],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0015892214,0.00012886223,0.00046148614,0.059199505,0.00020143359,0.00007460986,0.00012647072,0.000051911407,0.000048368915],"category_scores_gemma":[0.00018716896,0.00010496003,0.00011403859,0.114555955,0.00006725481,0.00012440319,0.000075341944,0.00034799115,0.0000015215754],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0016244494,0.00056483276,0.16989449,0.0034399102,0.0001029468,0.000107097614,0.02135255,0.010447247,0.0000033443516,0.0011252484,0.0013960996,0.7899418],"study_design_scores_gemma":[0.0027791525,0.014555339,0.19575788,0.001356345,0.00022482738,0.00039488944,0.033572674,0.74431723,0.00003863816,0.0012145754,0.0051269163,0.00066151575],"about_ca_topic_score_codex":0.00015683864,"about_ca_topic_score_gemma":0.00007970173,"teacher_disagreement_score":0.7892803,"about_ca_system_score_codex":0.00022632083,"about_ca_system_score_gemma":0.00022822402,"threshold_uncertainty_score":0.95146364},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4304694384","doi":"10.1080/1331677x.2022.2125888","title":"Bibliometric analysis of the literature on critical thinking: an increasingly important competence for higher education students","year":2022,"lang":"en","type":"article","venue":"Economic Research-Ekonomska Istraživanja","topic":"Education and Critical Thinking Development","field":"Social Sciences","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Transformative learning; Competence (human resources); Critical thinking; Field (mathematics); Library science; Social science; Political science; Engineering ethics; Sociology; Psychology; Mathematics education; Pedagogy; Computer science; Engineering","score_opus":0.11040328929545042,"score_gpt":0.4795980049839319,"score_spread":0.3691947156884815,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4304694384","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.978261,0.0003102891,0.0000055940764,0.013471652,0.0020241144,0.00060921203,0.00011170291,0.000031746393,0.0051747006],"genre_scores_gemma":[0.99473006,0.000054548615,0.0005170987,0.0007498672,0.00027522029,0.000068078494,0.000070916896,0.000016020645,0.003518172],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99628574,0.0010445659,0.0005336064,0.0004984793,0.0011227895,0.00051480276],"domain_scores_gemma":[0.99716485,0.0015393059,0.00014406268,0.00055506645,0.00037245709,0.00022424084],"candidate_categories":["bibliometrics","sts","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0062790643,0.00013238136,0.0002776552,0.013088503,0.002257136,0.0005584561,0.00169076,0.000080633974,0.0015231428],"category_scores_gemma":[0.0010412289,0.000118449934,0.0002108459,0.029945895,0.0005850741,0.00024023901,0.00027278689,0.00051481015,0.000010148494],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006550123,0.0010391574,0.05904212,0.000023961704,0.00017036177,9.832175e-7,0.012572584,0.00007150742,0.000021876313,0.9210852,0.0047549997,0.0011517884],"study_design_scores_gemma":[0.00042577216,0.00043323578,0.57875526,0.00007701694,0.00019049413,0.0000017844874,0.029024059,0.00013449158,0.0000881788,0.23601894,0.1544131,0.00043766873],"about_ca_topic_score_codex":0.0006400499,"about_ca_topic_score_gemma":0.00014547698,"teacher_disagreement_score":0.6850662,"about_ca_system_score_codex":0.0010559706,"about_ca_system_score_gemma":0.002154656,"threshold_uncertainty_score":0.9993896},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4304809609","doi":"10.1139/er-2021-0128","title":"Does land use legacy matter for current soil functioning? A bibliometric study (2001–2020)","year":2022,"lang":"en","type":"article","venue":"Environmental Reviews","topic":"Soil erosion and sediment transport","field":"Agricultural and Biological Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Climate change; Context (archaeology); Psychological resilience; Environmental resource management; Vulnerability (computing); Land use; Proxy (statistics); Geography; Environmental science; Environmental planning; Natural resource economics; Ecology; Psychology","score_opus":0.0453965385247976,"score_gpt":0.25962900515269394,"score_spread":0.21423246662789633,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4304809609","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99463046,0.0030325076,0.0000048165975,0.0003641736,0.0006455441,0.0011879216,0.000075960226,0.000024161594,0.00003446297],"genre_scores_gemma":[0.9928985,0.0032330817,0.000021223492,0.00084567396,0.00017556544,0.00072301284,0.00029248098,0.0000024166393,0.0018079958],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99859715,0.0001320541,0.00032822418,0.00037158444,0.00032100902,0.00024998712],"domain_scores_gemma":[0.99961185,0.00009231375,0.000109195826,0.000086469336,0.000001128029,0.00009902726],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0004742611,0.00015930171,0.00023576414,0.00021182763,0.0004315779,0.00008380038,0.00019011035,0.00001847618,0.007247348],"category_scores_gemma":[0.000013489368,0.000050596813,0.00018042623,0.0027995096,0.000022875769,0.00020298033,0.00009028595,0.00013246699,0.00025966347],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000029540726,0.0008165731,0.85428053,0.000005612757,0.000007655171,0.0000018743663,0.000045897286,0.0000020662058,0.0019648238,2.3498605e-7,0.019082677,0.12376251],"study_design_scores_gemma":[0.00017683874,0.00021003855,0.47542486,0.0000034610232,0.000020046862,0.000001327002,0.00015030673,0.000005126614,0.000015016783,0.0000034474747,0.5238964,0.00009311691],"about_ca_topic_score_codex":0.000078983096,"about_ca_topic_score_gemma":0.000056115146,"teacher_disagreement_score":0.50481373,"about_ca_system_score_codex":0.000060872953,"about_ca_system_score_gemma":0.0000020659136,"threshold_uncertainty_score":0.99366015},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4306253448","doi":"10.1002/pra2.641","title":"Investigating Open Access Publishing Practices of Early and <scp>Mid‐Career</scp> Researchers in Humanities and Social Sciences Disciplines","year":2022,"lang":"en","type":"article","venue":"Proceedings of the Association for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University","funders":"","keywords":"Publishing; Audience measurement; Descriptive statistics; Library science; Scholarly communication; Open peer review; Citation; Psychology; Political science; Advertising; Computer science; Plant biology; Business","score_opus":0.6736601966620449,"score_gpt":0.5621824102859028,"score_spread":0.11147778637614214,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4306253448","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9845085,0.00007561501,0.000002727295,0.009097034,0.000066672284,0.00044471625,0.00003055118,0.000009425561,0.005764748],"genre_scores_gemma":[0.9993123,0.0000367114,0.00033726703,0.000110421315,0.000007272898,0.000058592163,6.6283513e-7,0.0000019617266,0.00013485369],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9939381,0.000031832595,0.00061811,0.0002904762,0.004754495,0.0003670017],"domain_scores_gemma":[0.986647,0.0028685096,0.0033156613,0.00010272363,0.007018161,0.000047958518],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.055342477,0.000069432586,0.00020345261,0.019848257,0.0020194498,0.012004744,0.0047194175,0.000078439094,0.0000014675404],"category_scores_gemma":[0.20841368,0.000049711474,0.000018826531,0.078638405,0.001560307,0.026061628,0.009099861,0.0002905632,1.5427531e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000030556946,0.0000134565325,0.91181886,0.000027194677,0.000003977937,8.112961e-9,0.0041540554,0.0000031996226,0.0016777928,0.06952808,0.0010350879,0.011735207],"study_design_scores_gemma":[0.00083286065,0.00033362902,0.7160398,0.000022925211,0.000007982082,0.000003475047,0.09521636,0.00335206,0.006828561,0.16305554,0.014211025,0.0000957852],"about_ca_topic_score_codex":0.00021603183,"about_ca_topic_score_gemma":0.000022163647,"teacher_disagreement_score":0.1957791,"about_ca_system_score_codex":0.00015647139,"about_ca_system_score_gemma":0.00042697717,"threshold_uncertainty_score":0.9992798},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4306667375","doi":"10.1177/23779608221132164","title":"Hypersensitivity Reaction to Metal: A Bibliometric Study","year":2022,"lang":"en","type":"review","venue":"SAGE Open Nursing","topic":"Contact Dermatitis and Allergies","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Scopus; Portuguese; Medicine; Bibliometrics; Library science; Web of science; Family medicine; MEDLINE; Political science; Pathology; Computer science","score_opus":0.16534965866685514,"score_gpt":0.4404986649862657,"score_spread":0.2751490063194106,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4306667375","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00040954378,0.9582905,0.000009706318,0.00014649959,0.00083278655,0.004814821,0.000026346821,0.00006686694,0.035402894],"genre_scores_gemma":[0.0010236619,0.99586916,0.00015444407,0.0001182082,0.0001749351,0.00024000091,0.00009288433,0.000084403815,0.0022422927],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977941,0.00036013042,0.00049202715,0.000541967,0.00051879435,0.00029301338],"domain_scores_gemma":[0.9987259,0.00020435825,0.00023154613,0.00061895396,0.00003944058,0.00017980846],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0007564955,0.00034100763,0.001985115,0.016092237,0.00025263117,0.00026133517,0.00027780648,0.00008286056,0.00056248525],"category_scores_gemma":[0.00023605133,0.00029277935,0.0003418422,0.033164944,0.0000165105,0.0002453118,0.0003036777,0.0005253266,0.00011879377],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000018015511,0.00058750686,0.000002712369,0.0009037157,0.00023631312,0.0004600639,0.00036523232,3.9878742e-8,0.0000054282664,0.000003294197,0.000916377,0.9965013],"study_design_scores_gemma":[0.00043258243,0.0004788348,0.00067438383,0.010970252,0.0019150802,0.0005673613,0.0017892233,9.238724e-7,6.711246e-7,0.0000029611163,0.9828794,0.0002883313],"about_ca_topic_score_codex":0.00037011463,"about_ca_topic_score_gemma":0.000031682477,"teacher_disagreement_score":0.99621296,"about_ca_system_score_codex":0.00060596876,"about_ca_system_score_gemma":0.0002606262,"threshold_uncertainty_score":0.99995244},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4308297793","doi":"10.1016/j.joi.2022.101344","title":"Detecting emerging technologies and their evolution using deep learning and weak signal analysis","year":2022,"lang":"en","type":"article","venue":"Journal of Informetrics","topic":"Innovation Diffusion and Forecasting","field":"Decision Sciences","cited_by":36,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Defence Research and Development Canada; National Research Council Canada","funders":"","keywords":"Emerging technologies; Prosperity; Computer science; Data science; Field (mathematics); Identification (biology); Emerging markets; Domain (mathematical analysis); Risk analysis (engineering); Management science; Artificial intelligence; Business; Engineering; Political science; Mathematics","score_opus":0.062262075052657365,"score_gpt":0.3336898294886794,"score_spread":0.2714277544360221,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4308297793","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8011331,0.00077280717,0.19743516,0.00006890304,0.00011780206,0.000029940342,9.4763465e-7,0.000018663872,0.00042268343],"genre_scores_gemma":[0.9946579,0.000035871628,0.0052023744,0.000023407218,0.000026795911,4.7657343e-7,2.857891e-7,0.000004652689,0.000048277674],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9977144,0.00008478769,0.00096107984,0.00012791784,0.0009315129,0.00018030478],"domain_scores_gemma":[0.9967723,0.0009725893,0.0015058153,0.000081891434,0.00062472187,0.00004266522],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005930937,0.000099100725,0.0002974665,0.0066507435,0.00093877007,0.00024038444,0.00024160244,0.000050168863,0.00009188963],"category_scores_gemma":[0.008182838,0.00007066798,0.00011361664,0.012625715,0.000058842168,0.0005296748,0.00040726725,0.0005974965,4.183141e-7],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024003624,0.000010486785,0.1638108,0.0000031119841,0.00007825355,0.00000394933,0.0009805617,0.023491653,0.0014216321,0.00032196974,0.0000125035285,0.8098411],"study_design_scores_gemma":[0.00040917686,0.0002535004,0.013093046,0.000010316723,0.000081983446,0.0004332488,0.16170351,0.815938,0.0003394486,0.0027837385,0.0047834376,0.00017057672],"about_ca_topic_score_codex":0.0000064835585,"about_ca_topic_score_gemma":0.0000034309612,"teacher_disagreement_score":0.8096705,"about_ca_system_score_codex":0.00013109487,"about_ca_system_score_gemma":0.000051757426,"threshold_uncertainty_score":0.97962123},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4309214453","doi":"10.3389/fpsyt.2022.1022472","title":"The schizophrenia and gut microbiota: A bibliometric and visual analysis","year":2022,"lang":"en","type":"article","venue":"Frontiers in Psychiatry","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"China Postdoctoral Science Foundation","keywords":"Schizophrenia (object-oriented programming); Gut flora; Psychiatry; Autism; Microbiome; Bipolar disorder; Gut microbiome; Medicine; Psychology; Bioinformatics; Biology; Mood; Immunology","score_opus":0.002943017685195765,"score_gpt":0.2395603169183777,"score_spread":0.23661729923318195,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4309214453","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9600853,0.03674705,0.00064248545,0.0006902719,0.0013904091,0.00014182412,0.000022441305,0.0000072639523,0.00027299786],"genre_scores_gemma":[0.98871803,0.0024448314,0.006863249,0.0005087876,0.00010586855,0.000021756203,0.000040735868,0.000025522504,0.0012711886],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9991684,0.00008317787,0.00015726185,0.00030392676,0.000065466076,0.00022176716],"domain_scores_gemma":[0.9996961,0.0000055347277,0.000063096406,0.00016526565,0.00000965612,0.00006034156],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00025102784,0.00010352002,0.00013996981,0.0037011188,0.00031866287,0.000038240025,0.000121651115,0.00005899795,0.000016052669],"category_scores_gemma":[0.000010237996,0.00008939664,0.000053572185,0.008661425,0.00008689655,0.0000020197392,0.00018854672,0.0001398278,7.398302e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023487651,0.00009141538,0.75058097,0.00002118927,0.00033934283,0.0000013205442,0.000037966216,0.000009058474,0.023864271,0.00010240099,0.21365131,0.011065854],"study_design_scores_gemma":[0.0011329788,0.00023974838,0.87642795,0.0000032273147,0.00011172681,0.000024802035,0.00055765465,0.00015658094,0.00006804521,0.0001771023,0.12088492,0.00021527462],"about_ca_topic_score_codex":0.000031748423,"about_ca_topic_score_gemma":0.00012013989,"teacher_disagreement_score":0.12584695,"about_ca_system_score_codex":0.000013518431,"about_ca_system_score_gemma":0.00006761265,"threshold_uncertainty_score":0.41615272},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4310238085","doi":"10.5539/ibr.v15n12p86","title":"Collaborative Consumption and Institutional Theory: Bibliometric Analysis of Scientific Production from 2011 to 2021","year":2022,"lang":"en","type":"article","venue":"International Business Research","topic":"Sharing Economy and Platforms","field":"Business, Management and Accounting","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Scopus; Relevance (law); Consumption (sociology); Portfolio; Bibliometrics; Field (mathematics); Sample (material); Process (computing); Computer science; Sociology; Political science; Social science; Business; Library science; MEDLINE","score_opus":0.07917602596275185,"score_gpt":0.3417925361965752,"score_spread":0.2626165102338233,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4310238085","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99421585,0.00014928827,0.00042271812,0.0013471186,0.0011545132,0.0002328962,0.00017182431,0.000013258109,0.0022925506],"genre_scores_gemma":[0.99788624,0.00002556854,0.00010009648,0.00006847447,0.0003467998,0.000091643924,0.0007817108,0.0000058726914,0.0006936067],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99834573,0.000030409969,0.00021499523,0.00038276665,0.0008741813,0.00015189507],"domain_scores_gemma":[0.9976421,0.00018008403,0.000113022696,0.00017099942,0.0018791248,0.000014676832],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0021166198,0.00007489311,0.00014070355,0.059644878,0.0005811706,0.00047313375,0.00033199936,0.000024160407,0.0077771293],"category_scores_gemma":[0.0008388978,0.00007393025,0.00003259171,0.08297158,0.00018844818,0.0010731212,0.0005763334,0.00015010402,0.00019354785],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00066410576,0.00032945548,0.8923364,0.00004979624,0.0009357791,0.000010496022,0.00020759151,0.01747304,0.0031203872,0.042484384,0.0075436374,0.034844916],"study_design_scores_gemma":[0.0001775751,0.00000512177,0.9385797,0.0000117477375,0.000055058543,7.2576967e-7,0.00017070962,0.0029616766,0.00011622223,0.0051079346,0.052721474,0.00009205468],"about_ca_topic_score_codex":0.0012253933,"about_ca_topic_score_gemma":0.0003064257,"teacher_disagreement_score":0.046243288,"about_ca_system_score_codex":0.00010564008,"about_ca_system_score_gemma":0.00008233584,"threshold_uncertainty_score":0.9931299},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4311046123","doi":"10.1007/s11192-022-04565-6","title":"Retraction of Predatory publishing in Scopus: evidence on cross-country differences lacks justification","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":12,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Scopus; Publishing; Bibliometrics; Library science; Data science; Political science; Regional science; Computer science; Geography; MEDLINE; Law","score_opus":0.7729591827362152,"score_gpt":0.6206454994995081,"score_spread":0.15231368323670713,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311046123","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98859817,0.0014403409,0.0007649863,0.0004990958,0.0038719669,0.00041493453,0.00016691037,0.000033950568,0.004209634],"genre_scores_gemma":[0.9975622,0.00019991281,0.0002498348,0.00013407263,0.000082184,0.000044307388,0.000012112484,0.000012527898,0.0017027968],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9628152,0.0007973398,0.0015642622,0.0014876176,0.032456428,0.00087914756],"domain_scores_gemma":[0.97594756,0.01508291,0.0012771401,0.0017986144,0.005421689,0.0004720582],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.09781933,0.00019763832,0.00040519002,0.21557449,0.00068646536,0.0067095743,0.005449575,0.00020236606,0.0011561909],"category_scores_gemma":[0.27711844,0.00016431129,0.00013692804,0.6009262,0.00039972298,0.004809255,0.0015294885,0.0013916155,0.0000779208],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008790491,0.00045448792,0.9269142,0.00002293511,0.0000046827026,0.0000074609156,0.00032113513,0.0016653131,0.0012292454,0.0013475316,0.00771228,0.060232814],"study_design_scores_gemma":[0.00038789544,0.0003916381,0.9791538,0.000032638396,0.0000036056194,0.0000030835872,0.0012208901,0.008783312,0.0012112206,0.0018374604,0.00676536,0.00020910651],"about_ca_topic_score_codex":0.00031571757,"about_ca_topic_score_gemma":0.000034142344,"teacher_disagreement_score":0.38535172,"about_ca_system_score_codex":0.0010373471,"about_ca_system_score_gemma":0.00064940355,"threshold_uncertainty_score":0.9999314},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4311287923","doi":"10.1016/j.wneu.2022.12.013","title":"The 100 Most Cited Articles in Endoscopic Endonasal Skull Base Surgery: A Bibliometric Analysis","year":2022,"lang":"en","type":"article","venue":"World Neurosurgery","topic":"Head and Neck Surgical Oncology","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Medicine; Neurosurgery; Skull; Endoscopic endonasal surgery; Web of science; Citation; General surgery; Surgery; Pathology; Meta-analysis; Library science","score_opus":0.04346182143724634,"score_gpt":0.296329549928294,"score_spread":0.25286772849104766,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311287923","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9888166,0.000949606,0.0000040586337,0.0076955096,0.00067648676,0.00027986034,0.000026172427,0.00010801777,0.001443658],"genre_scores_gemma":[0.9914028,0.00028303106,0.000012283754,0.005019051,0.00013753948,0.00018302481,0.00004261964,0.000034848665,0.0028847728],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9962204,0.0008821311,0.0008653687,0.00052274886,0.0007343792,0.00077494787],"domain_scores_gemma":[0.98908365,0.009809867,0.00018329122,0.00055696734,0.000060755483,0.00030550125],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.002002994,0.00022282917,0.0008774169,0.07882428,0.0004196682,0.000056768295,0.0001733897,0.000044087104,0.0016179485],"category_scores_gemma":[0.0016786358,0.00017187187,0.00047038315,0.32410374,0.00012745256,0.00007459105,0.00020078532,0.00066566555,0.00003933246],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00043308764,0.00059793587,0.95913553,0.000013811698,0.00013481404,0.002698087,0.000029179126,0.00010821639,0.002215971,0.00010143395,0.017397856,0.017134085],"study_design_scores_gemma":[0.0007051904,0.000106632935,0.79006886,0.000010462813,0.00017507715,0.000058667105,0.000024815603,0.0010899649,0.0007718853,0.00003545209,0.2068073,0.0001456829],"about_ca_topic_score_codex":0.00017411844,"about_ca_topic_score_gemma":0.00024617676,"teacher_disagreement_score":0.24527946,"about_ca_system_score_codex":0.00015047191,"about_ca_system_score_gemma":0.00025989147,"threshold_uncertainty_score":0.9992947},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4311884401","doi":"10.3390/buildings12122242","title":"Towards a Data-Rich Era: A Bibliometric Analysis of Construction Management from 2000 to 2020","year":2022,"lang":"en","type":"article","venue":"Buildings","topic":"Construction Project Management and Performance","field":"Decision Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Social Science Fund of China; National Office for Philosophy and Social Sciences; National Natural Science Foundation of China","keywords":"Data management; Data science; Big data; Field (mathematics); Computer science; Domain (mathematical analysis); Technology management; Knowledge management; Emerging technologies; Engineering management; Management science; Engineering; Data mining","score_opus":0.0820736274263793,"score_gpt":0.3692760089674105,"score_spread":0.28720238154103117,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311884401","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9709495,0.0002038472,0.013517686,0.0010160272,0.0008852604,0.00035314087,0.00058827235,0.000056622393,0.012429664],"genre_scores_gemma":[0.96502066,0.000111540976,0.031666722,0.00041912665,0.000071357,0.000052484047,0.00012865262,0.000010451381,0.0025190364],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9958492,0.00011652451,0.0007734908,0.00083269324,0.0021890164,0.00023908935],"domain_scores_gemma":[0.9978275,0.00021690401,0.000389479,0.0012875781,0.00018561314,0.000092981034],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0025082985,0.0001461852,0.0004203314,0.07995962,0.0002818299,0.00023347714,0.0019276427,0.000026225895,0.011985525],"category_scores_gemma":[0.00027876082,0.00013461504,0.00013806504,0.33568177,0.000060384642,0.00049450126,0.0024723988,0.00014576003,0.000091836904],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009143234,0.00005371154,0.073104136,0.0000055847845,0.0008228662,0.0000079192205,0.00044776854,0.0023478337,0.00008544127,0.002018615,0.057977885,0.8630368],"study_design_scores_gemma":[0.0005604363,0.00009697568,0.26695028,0.0000056532667,0.000993777,0.0000060474104,0.0030714055,0.015970932,0.00013963867,0.003088131,0.7087898,0.00032689865],"about_ca_topic_score_codex":0.00058858603,"about_ca_topic_score_gemma":0.000038804505,"teacher_disagreement_score":0.8627099,"about_ca_system_score_codex":0.00010387111,"about_ca_system_score_gemma":0.00006513882,"threshold_uncertainty_score":0.98891765},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4312155589","doi":"10.1017/s0266462322003051","title":"PD46 Multi-Criteria Decision Analysis In Healthcare: Scientometric And Bibliometric Analysis","year":2022,"lang":"en","type":"article","venue":"International Journal of Technology Assessment in Health Care","topic":"Healthcare Systems and Reforms","field":"Economics, Econometrics and Finance","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Multiple-criteria decision analysis; Bibliometrics; Analytic hierarchy process; Health care; Decision analysis; Health technology; Management science; Library science; Operations research; Computer science; Political science; Engineering; Mathematics; Statistics","score_opus":0.03691137667741144,"score_gpt":0.39749978030312766,"score_spread":0.3605884036257162,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312155589","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94753397,0.026462525,0.0058220467,0.018130103,0.0014957429,0.00026159125,0.00023168519,0.000017498154,0.000044846856],"genre_scores_gemma":[0.991358,0.0023625714,0.0056939134,0.00046036218,0.00003581364,0.000028150806,0.00003216627,0.000012132556,0.00001689672],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9961854,0.000119587836,0.0024505008,0.0004390157,0.0004012899,0.00040415488],"domain_scores_gemma":[0.99741995,0.00008129931,0.0017242469,0.0002647633,0.00035550876,0.00015423348],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.004312449,0.00015400746,0.0010056518,0.36684164,0.00016304087,0.00007461635,0.00077607285,0.00015211616,0.00012282019],"category_scores_gemma":[0.00028511812,0.00015307973,0.00025446937,0.20629518,0.000053399937,0.00024360465,0.00033543486,0.000917026,0.0000023166706],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024609073,0.00017770234,0.9467823,0.000043397886,0.00045594436,0.00007239097,0.0002715564,0.0016663057,9.84086e-7,0.0039978884,0.00001956071,0.04648737],"study_design_scores_gemma":[0.0011797822,0.00033225404,0.987882,0.000033315046,0.00001676554,0.000051933628,0.0018047793,0.0037788744,9.2747985e-7,0.0014829558,0.0032989753,0.0001374197],"about_ca_topic_score_codex":0.006836981,"about_ca_topic_score_gemma":0.0013426775,"teacher_disagreement_score":0.16054647,"about_ca_system_score_codex":0.0031063082,"about_ca_system_score_gemma":0.0003139591,"threshold_uncertainty_score":0.9997766},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4312379132","doi":"10.31436/iiumlj.v30is1.697","title":"POLICY, LEGAL AND REGULATION RESEARCH IN THE SHARING ECONOMY: A BIBLIOMETRIC ANALYSIS AND SYSTEMATIC LITERATURE REVIEW","year":2022,"lang":"en","type":"article","venue":"IIUM Law Journal","topic":"Sharing Economy and Platforms","field":"Business, Management and Accounting","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Scopus; Sharing economy; Popularity; Bibliometrics; Political science; Systematic review; Business; Law; Computer science","score_opus":0.04719949876983625,"score_gpt":0.3095124755151847,"score_spread":0.2623129767453484,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4312379132","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6358178,0.21129087,0.000114861454,0.029713312,0.0004867722,0.0023028955,0.000012425675,0.000070231894,0.12019079],"genre_scores_gemma":[0.99521846,0.0018567557,0.000013910286,0.0022355444,0.0005268609,0.00003707987,0.000011082419,0.0000071236814,0.00009320867],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9989446,0.0000634275,0.00035853157,0.00017646716,0.0002398226,0.00021719029],"domain_scores_gemma":[0.9993613,0.00013967414,0.0002131813,0.00018315345,0.00008739269,0.000015338363],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0065087266,0.00008839958,0.00026418857,0.026647432,0.0008752724,0.0022418797,0.00028984513,0.00002635231,0.00008701713],"category_scores_gemma":[0.000104655366,0.000060760376,0.000068065616,0.05833139,0.0000282619,0.001639586,0.00025897368,0.00054802955,0.0000046775876],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000049968665,0.00012632733,0.12211816,0.079911955,0.0006125644,0.00036732183,0.00063730817,0.00044630654,0.0000068219424,0.78632444,0.0034789315,0.00591989],"study_design_scores_gemma":[0.0029454462,0.00014613017,0.4029283,0.043048285,0.0023081908,0.005523317,0.0029387716,0.0495616,0.0000033862677,0.26009417,0.2290731,0.0014293077],"about_ca_topic_score_codex":0.0006367827,"about_ca_topic_score_gemma":0.00028011363,"teacher_disagreement_score":0.5262303,"about_ca_system_score_codex":0.0000500827,"about_ca_system_score_gemma":0.000017288156,"threshold_uncertainty_score":0.9987939},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4313118920","doi":"10.20414/light.v2i1.4914","title":"Identifying and Mapping Study of the Information Professional in Library with Scientometric Analysis","year":2022,"lang":"en","type":"article","venue":"THE LIGHT Journal of Librarianship and Information Science","topic":"Blockchain Technology in Education and Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Badan Riset dan Inovasi Nasional","keywords":"Theme (computing); Digital library; Quarter (Canadian coin); The Internet; Library science; Psychology; Computer science; World Wide Web; Geography","score_opus":0.01585043951901023,"score_gpt":0.24566459544800662,"score_spread":0.2298141559289964,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313118920","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9843729,0.000058345417,0.006835393,0.0074903457,0.00027487485,0.00016619064,4.876288e-7,0.0000123112995,0.0007891628],"genre_scores_gemma":[0.9983977,0.000008093638,0.001098178,0.00046881574,0.000006189715,0.000005026246,2.2049228e-7,8.8432705e-7,0.000014891599],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9983805,0.00015634249,0.0004887261,0.00006662832,0.0007872337,0.00012060075],"domain_scores_gemma":[0.99876523,0.00010046489,0.00077431154,0.00022354015,0.00009447744,0.000041973315],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0025612195,0.000057702953,0.00011693398,0.0035855693,0.0007459473,0.00046941708,0.0013301637,0.000015974347,0.000010997729],"category_scores_gemma":[0.000124619,0.000030639945,0.000021580146,0.017110258,0.00018317634,0.021067558,0.0006723856,0.00036375935,3.0250428e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000037314632,0.0001509644,0.61502683,0.00003197542,0.00006544895,0.0000010317584,0.2484189,0.0036567545,0.0000891875,0.108659275,0.00015431378,0.023707986],"study_design_scores_gemma":[0.00059268763,0.00018674196,0.9139328,0.000031313644,0.000022813256,0.00008388305,0.06727664,0.013871972,0.00024751906,0.00097189046,0.0026973342,0.000084411455],"about_ca_topic_score_codex":0.00000295938,"about_ca_topic_score_gemma":3.2775898e-7,"teacher_disagreement_score":0.29890594,"about_ca_system_score_codex":0.000020140467,"about_ca_system_score_gemma":0.00032524212,"threshold_uncertainty_score":0.9926243},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4313208965","doi":"10.4103/0973-3698.364675","title":"Disseminating Biomedical Research: Predatory Journals and Practices","year":2022,"lang":"en","type":"article","venue":"Indian Journal of Rheumatology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; Ottawa Hospital; University of Ottawa","funders":"","keywords":"Medicine; Dissemination; Engineering","score_opus":0.6830509861330616,"score_gpt":0.6488471419017912,"score_spread":0.03420384423127043,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313208965","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9833119,0.007029027,0.00014075382,0.007916777,0.0009472963,0.00008728595,0.000019135565,0.0000039060565,0.00054395513],"genre_scores_gemma":[0.9980003,0.0009494791,0.00074206723,0.000052357635,0.000046559024,0.000004319109,5.642319e-7,0.000009548221,0.00019477459],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.981538,0.0032074107,0.0015722213,0.0003794171,0.012593251,0.0007097161],"domain_scores_gemma":[0.9654472,0.025972348,0.0032099613,0.00042474899,0.003988719,0.0009570552],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.12487538,0.000101748556,0.0004821823,0.067568064,0.0010347668,0.0012925814,0.0027757373,0.00010626692,0.002487135],"category_scores_gemma":[0.20049366,0.00007244194,0.000115665694,0.061971772,0.0006393072,0.00084342744,0.0015036359,0.002121441,0.000032023694],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000119018536,0.0003369398,0.7015039,0.000019291914,0.00007604732,0.0018359275,0.0036791025,0.000015968422,0.00066429545,0.0009404521,0.09676135,0.19404773],"study_design_scores_gemma":[0.0026481857,0.004397966,0.12911253,0.00014236593,0.00001872927,0.10573468,0.098282784,0.0012601323,0.00029738262,0.05457896,0.6030451,0.00048118908],"about_ca_topic_score_codex":0.0000344341,"about_ca_topic_score_gemma":0.0000046612304,"teacher_disagreement_score":0.57239133,"about_ca_system_score_codex":0.00013682978,"about_ca_system_score_gemma":0.0008589899,"threshold_uncertainty_score":0.9997442},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity","bibliometrics"],"domain":"evaluation","study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","research_integrity"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4313415928","doi":"10.24193/cbb.2022.26.13","title":"Global research trends on psychosocial rehabilitation in patients with cardiovascular diseases: A bibliometric analysis using CiteSpace","year":2022,"lang":"en","type":"article","venue":"Cognition Brain Behavior An Interdisciplinary Journal","topic":"Cardiac Health and Mental Health","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Psychosocial; Citation; Rehabilitation; Institution; Medicine; Popularity; Gerontology; Citation analysis; Web of science; Science Citation Index; Bibliometrics; Library science; Family medicine; Psychology; Social science; Sociology; Physical therapy; Psychiatry; Pathology; Computer science","score_opus":0.04957504233603674,"score_gpt":0.4524050432404831,"score_spread":0.4028300009044463,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313415928","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9968973,0.00013632896,0.00013651044,0.0008546396,0.00033259433,0.0007598106,0.0004417312,0.00002733281,0.00041374855],"genre_scores_gemma":[0.9983793,0.000006642597,0.00024734033,0.00022328382,0.00023556103,0.00019763125,0.000656259,0.000028891624,0.000025102807],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.994864,0.0014507953,0.00057200156,0.00052473013,0.0019825008,0.00060598843],"domain_scores_gemma":[0.99799645,0.0001778373,0.0001681429,0.00036737818,0.0005344212,0.0007557988],"candidate_categories":["bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.002476117,0.00020026002,0.0005768699,0.051576685,0.0014281148,0.000081396785,0.00013357375,0.000075648626,0.00040577655],"category_scores_gemma":[0.00011587016,0.00018837143,0.00063307164,0.098771,0.0001267479,0.00030048378,0.00018048238,0.00092358765,0.0000032929668],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0051609497,0.0039789975,0.90117854,0.000048316087,0.00020468458,0.0002110572,0.001155361,0.00040723605,0.000004364747,0.0000039671613,0.00084427197,0.08680224],"study_design_scores_gemma":[0.0059232037,0.008696038,0.97879803,0.00011463527,0.0012314507,0.0001844109,0.0045502773,0.00012731682,3.9350542e-7,0.00011395767,0.00008384305,0.000176424],"about_ca_topic_score_codex":0.000051058174,"about_ca_topic_score_gemma":0.000043078384,"teacher_disagreement_score":0.086625814,"about_ca_system_score_codex":0.0028413702,"about_ca_system_score_gemma":0.0003988276,"threshold_uncertainty_score":0.9998719},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4313589464","doi":"10.2147/jpr.s375063","title":"Bibliometric Analysis of Post-Stroke Pain Research Published from 2012 to 2021","year":2023,"lang":"en","type":"article","venue":"Journal of Pain Research","topic":"Stroke Rehabilitation and Recovery","field":"Medicine","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Jiangxi Provincial Department of Science and Technology","keywords":"Medicine; Stroke (engine); Bibliometrics; Physical medicine and rehabilitation; Library science","score_opus":0.11445782345271628,"score_gpt":0.4557716988108319,"score_spread":0.34131387535811564,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313589464","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97641814,0.0015750316,0.00047246413,0.019492514,0.00022552535,0.00041196903,0.00011474699,0.000012119322,0.0012775208],"genre_scores_gemma":[0.97759604,0.0011786787,0.0018888955,0.00027214226,0.00086248637,0.0000218026,0.0000718703,0.00003770411,0.01807036],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9860067,0.00634587,0.0010962061,0.000339886,0.0053733904,0.000837998],"domain_scores_gemma":[0.9605849,0.028433194,0.00021822513,0.00065772183,0.00924818,0.00085780286],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.10441031,0.00012868572,0.0008256774,0.34848532,0.00014595283,0.00016525286,0.00054299715,0.00020110949,0.004584794],"category_scores_gemma":[0.10620532,0.00009574239,0.0007163188,0.3504578,0.000226938,0.00032802636,0.00021917213,0.0015825861,0.00020808933],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00097504706,0.0005049836,0.39462134,0.00015177272,0.0021748391,0.00023451645,0.0008359986,0.00015109283,0.06343905,0.000018720762,0.43440434,0.1024883],"study_design_scores_gemma":[0.0010125787,0.002994742,0.93572897,0.00024734234,0.00016926858,0.0000040528616,0.006667319,0.0021778068,0.0008843337,0.00014362844,0.04986965,0.00010031459],"about_ca_topic_score_codex":0.00066985236,"about_ca_topic_score_gemma":0.0001015926,"teacher_disagreement_score":0.54110765,"about_ca_system_score_codex":0.00032807433,"about_ca_system_score_gemma":0.0009169001,"threshold_uncertainty_score":0.99632514},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4313627686","doi":"10.1016/j.amjsurg.2023.01.005","title":"Bibliometric analysis of the classic cited papers in the American Journal of Surgery: Citation recapitulates surgical history","year":2023,"lang":"en","type":"article","venue":"The American Journal of Surgery","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Paul's Hospital; University of British Columbia","funders":"","keywords":"Citation; Citation analysis; Medicine; Bibliometrics; History; Library science; Computer science","score_opus":0.14872894825991914,"score_gpt":0.37712358614666014,"score_spread":0.228394637886741,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313627686","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98962605,0.000860744,0.00003198397,0.008264095,0.00093099824,0.00005358325,0.000009178428,0.000007348818,0.00021602509],"genre_scores_gemma":[0.99704015,0.0018331911,0.000025816666,0.00075167185,0.00022279785,0.0000017183877,0.0000019430338,0.0000161426,0.000106567866],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9874532,0.004653116,0.002963034,0.00024288782,0.0042512603,0.00043646106],"domain_scores_gemma":[0.93071294,0.05621758,0.011018605,0.00070642086,0.0011750978,0.0001693405],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.05345309,0.00021402379,0.0018787631,0.08190239,0.00015932233,0.00030423087,0.0033846165,0.000056096636,0.00013452653],"category_scores_gemma":[0.021158211,0.000092989176,0.0016171446,0.34146446,0.0015638798,0.00087709987,0.00016772699,0.0011047847,0.0000065441077],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037110946,0.000067170964,0.73296803,0.0000024989452,0.0005033658,0.00007803507,0.001574767,0.0020946956,0.00014154543,0.000019061756,0.049146127,0.21303362],"study_design_scores_gemma":[0.00011420426,0.00005772376,0.9807265,0.00006303568,0.00029596445,0.00014563336,0.00916537,0.00053921796,0.000017215743,0.000401921,0.008349929,0.00012329883],"about_ca_topic_score_codex":0.00033488142,"about_ca_topic_score_gemma":0.000013402992,"teacher_disagreement_score":0.25956208,"about_ca_system_score_codex":0.000220798,"about_ca_system_score_gemma":0.00084186444,"threshold_uncertainty_score":0.987087},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4313641121","doi":"10.3389/fsurg.2022.1056732","title":"Bibliometric analysis of research trends in relationship between sarcopenia and surgery","year":2023,"lang":"en","type":"article","venue":"Frontiers in Surgery","topic":"Nutrition and Health in Aging","field":"Medicine","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Sarcopenia; Medicine; Bibliometrics; Science Citation Index; Cluster (spacecraft); Cancer cachexia; Malnutrition; Gerontology; Citation; Cachexia; Cancer; Library science; Pathology; Internal medicine; Computer science","score_opus":0.30363323547944876,"score_gpt":0.46058431508797115,"score_spread":0.1569510796085224,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313641121","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99479103,0.0019337938,0.000040638326,0.0024247388,0.0002267521,0.0001016739,0.000016694523,0.00003991648,0.00042478257],"genre_scores_gemma":[0.9973629,0.0011831254,0.00048363686,0.00006066217,0.000059569276,0.000020280017,0.0002339618,0.000015084333,0.0005807542],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99766546,0.0003665975,0.000726758,0.00027479997,0.000527142,0.00043926472],"domain_scores_gemma":[0.9933266,0.0060805436,0.00009365022,0.00024068865,0.00010001996,0.00015851036],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.007825646,0.000086123175,0.0008720092,0.49280933,0.000059797152,0.000014667747,0.000048896934,0.0001329425,0.000060971688],"category_scores_gemma":[0.0022006154,0.00009007694,0.00014825418,0.4959149,0.00013139816,0.00010400734,0.000037571324,0.00040954206,0.0000067088527],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005795143,0.000035759353,0.8646235,0.00019392287,0.00005812856,0.00005144387,0.000106762986,0.0000028926197,0.0000024016956,0.000009982502,0.10694203,0.027915226],"study_design_scores_gemma":[0.00024943805,0.000008663464,0.99665266,0.00023212643,0.00006564009,9.0545075e-7,0.00023742123,0.00031852056,0.000012970954,0.00032935324,0.0018206942,0.0000716207],"about_ca_topic_score_codex":0.0002399742,"about_ca_topic_score_gemma":0.000040668812,"teacher_disagreement_score":0.13202915,"about_ca_system_score_codex":0.00012712338,"about_ca_system_score_gemma":0.000126429,"threshold_uncertainty_score":0.51479965},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4313856802","doi":"10.1016/j.heliyon.2023.e12896","title":"Health literacy in childhood and adolescence. A bibliometric analysis of scientific publications and professionals’ involvement","year":2023,"lang":"en","type":"review","venue":"Heliyon","topic":"School Health and Nursing Education","field":"Health Professions","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Inclusion (mineral); Bibliometrics; Promotion (chess); Publishing; Health professionals; Medical education; Scientific literature; Literacy; Web of science; Psychology; Medicine; MEDLINE; Social science; Library science; Health care; Political science; Pedagogy; Sociology; Politics","score_opus":0.180851579561156,"score_gpt":0.5314387577605512,"score_spread":0.35058717819939517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313856802","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005328455,0.98878783,0.000006996089,0.0019003018,0.0010297414,0.0026363237,0.00020979313,0.00006498096,0.000035556986],"genre_scores_gemma":[0.00035217078,0.9960617,0.00023565812,0.0002866369,0.00011516591,0.00076461106,0.0011194203,0.000037722413,0.0010269424],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99510044,0.0012001091,0.0018343184,0.0007454902,0.0004714791,0.0006481375],"domain_scores_gemma":[0.99623287,0.0010890836,0.0013141831,0.00059874635,0.0003397427,0.00042537024],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0038568336,0.0002745206,0.0016996999,0.13006707,0.0007470056,0.00006329373,0.00022125385,0.00034553523,0.000057526293],"category_scores_gemma":[0.0007082907,0.000227578,0.0001409155,0.2143023,0.00009525095,0.00021109391,0.00015678241,0.0009502622,0.000048417554],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000003884654,0.00033811267,0.03182344,0.11424096,0.00010754393,2.0216095e-7,0.0029745288,8.127559e-8,4.3492896e-8,0.0001346247,0.0031654085,0.8472112],"study_design_scores_gemma":[0.0002486509,0.000051721334,0.63483965,0.10378359,0.0005021253,7.7293606e-7,0.00053216744,0.000018156128,1.8244972e-8,0.00005203783,0.25977752,0.00019358801],"about_ca_topic_score_codex":0.0003210737,"about_ca_topic_score_gemma":0.000683347,"teacher_disagreement_score":0.8470176,"about_ca_system_score_codex":0.00040765156,"about_ca_system_score_gemma":0.0037714217,"threshold_uncertainty_score":0.92803615},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4317618843","doi":"10.3390/su15031994","title":"A Bibliometric Analysis of Forest Gap Research during 1980–2021","year":2023,"lang":"en","type":"article","venue":"Sustainability","topic":"Forest Management and Policy","field":"Environmental Science","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Intact forest landscape; China; Science Citation Index; Forest ecology; Forest management; Certified wood; Environmental resource management; Web of science; Geography; Forest inventory; Sustainable forest management; Citation; Forest dynamics; Ecology; Forestry; Environmental science; Political science; Library science; Ecosystem; Computer science","score_opus":0.039958651793453315,"score_gpt":0.3626600476683026,"score_spread":0.3227013958748493,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4317618843","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98558575,0.00001078374,0.0000042335214,0.00057080376,0.000023674047,0.00031590706,0.000005970634,0.000043040614,0.013439833],"genre_scores_gemma":[0.984881,0.000026833668,0.000017783384,0.000005261132,0.000020784146,0.000036680143,0.00001363735,0.000008087138,0.014989932],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9978675,0.000183869,0.00024986637,0.00037207082,0.0007282399,0.0005984068],"domain_scores_gemma":[0.9988374,0.00023185459,0.000052305062,0.00066136854,0.00011393043,0.00010311125],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0024886138,0.00009280176,0.00020178157,0.052085083,0.00018455434,0.000036168312,0.00038280126,0.00005054073,0.0034493888],"category_scores_gemma":[0.0017600262,0.00008601433,0.0001574787,0.45436317,0.000437751,0.00018185031,0.0008656767,0.00013793631,0.00064884016],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002021267,0.000065700515,0.98663956,0.0000993779,0.0000685919,0.000014201962,0.00034461793,0.0055730185,0.000049956132,0.0013173824,0.0022090487,0.0035983315],"study_design_scores_gemma":[0.00011270348,0.00003622641,0.98876435,0.0000015600274,0.00005639129,8.19502e-8,0.00023490518,0.0027106644,0.000046810284,0.0047245235,0.0032289647,0.000082834835],"about_ca_topic_score_codex":0.0054176315,"about_ca_topic_score_gemma":0.00090029335,"teacher_disagreement_score":0.40227807,"about_ca_system_score_codex":0.0006394098,"about_ca_system_score_gemma":0.000035672612,"threshold_uncertainty_score":0.9974616},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4319264300","doi":"10.36227/techrxiv.21990410.v1","title":"Distributed Energy Resources, Transmission Power System Control, Energy Transition: An Exploratory Bibliometric Analysis","year":2023,"lang":"en","type":"preprint","venue":"","topic":"Smart Grid Security and Resilience","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hydro-Québec","funders":"Hydro-Québec","keywords":"Context (archaeology); Computer science; Electric power system; Reliability (semiconductor); Distributed generation; Electricity; Systems engineering; Reliability engineering; Power (physics); Renewable energy; Engineering; Electrical engineering","score_opus":0.012568478825145557,"score_gpt":0.21078112398064242,"score_spread":0.19821264515549686,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4319264300","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.024134327,0.004587434,0.9653632,0.00012661397,0.001110343,0.0001331412,0.00075704255,0.0031492517,0.00063862524],"genre_scores_gemma":[0.9960846,0.0014980267,0.00020445173,0.00007500551,0.0003111492,0.00013271118,0.0015066541,0.000100562735,0.00008688737],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9967642,0.000271799,0.0007803951,0.0008383848,0.00078082754,0.0005644304],"domain_scores_gemma":[0.9981766,0.00014947173,0.00009892597,0.00092326716,0.0001721968,0.00047957202],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0004351839,0.0005854896,0.0010033909,0.04616626,0.00017062298,0.00023514635,0.0006731986,0.0008318436,0.00016371519],"category_scores_gemma":[0.000009853629,0.00053656974,0.0006365596,0.08110298,0.000070172806,0.00023131564,0.00007562725,0.00049956405,0.000016616099],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000061074345,0.000097303644,0.00004880294,0.000554513,0.0018704396,0.00018209647,0.0008532533,0.9896073,0.0004558264,0.0006288323,0.0038481967,0.0017924133],"study_design_scores_gemma":[0.0006703582,0.000078411205,0.001889137,0.00025276715,0.0011313142,0.000008908084,0.00082004024,0.9607258,0.0007472977,0.00008499602,0.0325604,0.0010305558],"about_ca_topic_score_codex":0.00068644516,"about_ca_topic_score_gemma":0.0003383583,"teacher_disagreement_score":0.97195023,"about_ca_system_score_codex":0.0002071288,"about_ca_system_score_gemma":0.000055982448,"threshold_uncertainty_score":0.9997086},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4320000384","doi":"10.5225/jjsm.2023-001","title":"Research Promotion Measures based on a Bibliometric Review of Sport Management Research","year":2023,"lang":"en","type":"article","venue":"Japanese Journal of Sport Management","topic":"Diverse Approaches in Healthcare and Education Studies","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Promotion (chess); Bibliometrics; Sport management; Management science; Sociology; Psychology; Political science; Computer science; Engineering; Public relations; Library science","score_opus":0.3907481524376887,"score_gpt":0.4958524012183109,"score_spread":0.10510424878062224,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4320000384","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9249041,0.0022915674,0.00000996924,0.011402679,0.00038086073,0.0023399934,0.000001022506,0.000045897938,0.058623895],"genre_scores_gemma":[0.9547231,0.040710557,0.00083726906,0.00036861008,0.00016480166,0.00012231308,0.000013087241,0.000021543648,0.0030386816],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9934651,0.00006672342,0.0010434751,0.00031604548,0.0045436486,0.0005649776],"domain_scores_gemma":[0.9964026,0.000074953416,0.00033410348,0.0006014692,0.0023325349,0.0002543147],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.022555573,0.00015279705,0.0005693545,0.034792304,0.0002009495,0.000024389226,0.00036170732,0.00004824582,0.00018359024],"category_scores_gemma":[0.00013008738,0.000114995055,0.00019930744,0.052838244,0.0001572269,0.00013561257,0.00020526805,0.00050932204,0.00016573445],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021948407,0.0059463466,0.53444785,0.07838663,0.0015328396,0.0043785893,0.0019324985,0.00024331262,0.000032337688,0.0069255987,0.17020868,0.19377045],"study_design_scores_gemma":[0.0012780966,0.0015008729,0.9041181,0.011511418,0.00021651233,0.00004701649,0.012455062,0.00005861406,0.0000923273,0.00017671389,0.068412244,0.00013301552],"about_ca_topic_score_codex":0.000038669143,"about_ca_topic_score_gemma":7.81719e-7,"teacher_disagreement_score":0.36967024,"about_ca_system_score_codex":0.00032241837,"about_ca_system_score_gemma":0.000119528006,"threshold_uncertainty_score":0.97614753},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"incentives","study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4321096183","doi":"10.5539/jms.v13n1p58","title":"Plant Waste in the Production of New Materials in Brazil: A Scientometric Analysis from 19912021","year":2023,"lang":"en","type":"article","venue":"Journal of Management and Sustainability","topic":"Agricultural and Food Sciences","field":"Agricultural and Biological Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Conselho Nacional de Desenvolvimento Científico e Tecnológico; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior","keywords":"Reuse; Scopus; Agriculture; Agricultural waste; Biomass (ecology); Environmental science; Waste management; Agroforestry; Geography; Biology; Engineering; Agronomy; Archaeology","score_opus":0.021976747324995206,"score_gpt":0.25473527541713764,"score_spread":0.23275852809214242,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4321096183","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9950263,0.00013811467,8.7532504e-7,0.0045230845,0.00008779491,0.00017998254,0.0000047797594,0.0000031133445,0.000035955756],"genre_scores_gemma":[0.9995471,0.00016611702,0.000013278245,0.000019786636,0.000077879304,0.000002022084,0.0000072641337,1.4095482e-7,0.00016640766],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9988112,0.00014620615,0.00037156718,0.00015822925,0.00035647693,0.00015634182],"domain_scores_gemma":[0.999531,0.00013432454,0.0001853336,0.000041504525,0.00007330338,0.000034553486],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0023074169,0.0000654208,0.00020965163,0.00027367787,0.000056349512,0.00006367449,0.0002551661,0.000026305812,0.000028752425],"category_scores_gemma":[0.00017999088,0.000018448647,0.00006804589,0.0076446263,0.000053337713,0.00019690869,0.00007937582,0.00006463252,3.8656987e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001853249,0.00025435243,0.9240237,0.000070666116,0.00010454641,0.00006153753,0.0026493447,0.00051971216,0.0031965107,0.00033380662,0.0017513892,0.06684914],"study_design_scores_gemma":[0.00007855583,0.00010431095,0.9748358,0.000012385147,0.00004287957,0.0000010937634,0.021179425,0.000024473182,0.00020017258,0.0030382103,0.00043877668,0.000043904343],"about_ca_topic_score_codex":0.0009289743,"about_ca_topic_score_gemma":0.0011797579,"teacher_disagreement_score":0.06680524,"about_ca_system_score_codex":0.00003550363,"about_ca_system_score_gemma":0.000009344048,"threshold_uncertainty_score":0.36729896},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4321611345","doi":"10.1108/raf-11-2022-0306","title":"A bibliometric analysis of political connections literature","year":2023,"lang":"en","type":"article","venue":"Review of Accounting and Finance","topic":"Political Influence and Corporate Strategies","field":"Business, Management and Accounting","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Politics; Originality; Value (mathematics); Political communication; Corporate governance; Sociology; Political geography; Social science; Political science; Computer science; Economics; Management; Law","score_opus":0.029592577355047743,"score_gpt":0.29669092923306245,"score_spread":0.26709835187801473,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4321611345","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94995266,0.03849161,0.000028023554,0.0010066885,0.00008568745,0.000128144,0.0000322397,0.00006990083,0.010205057],"genre_scores_gemma":[0.9688279,0.029791119,0.000033811968,0.0011335483,0.00011300636,0.000008548831,0.000034617104,0.0000060669436,0.000051401345],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99899644,0.0000063221614,0.00038928608,0.00017337842,0.00017623235,0.00025834053],"domain_scores_gemma":[0.99910563,0.00008143124,0.00026003877,0.00017760831,0.00036767835,0.0000075851594],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00042502637,0.00010400737,0.00044932865,0.016938863,0.000062511244,0.00008131399,0.00012495645,0.00004622124,0.000021342294],"category_scores_gemma":[0.0004993475,0.00008587827,0.00015184542,0.16155852,0.000103059574,0.0005580769,0.00007054838,0.000083416184,0.000024802686],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000029986827,0.000028494911,0.03097307,0.008892612,0.00011367082,0.0000046407717,0.000006761325,0.000013068505,0.00007617454,0.94956326,0.0011481646,0.009177082],"study_design_scores_gemma":[0.00018060119,0.00001857884,0.92289615,0.0095390985,0.0013339765,0.0000028849147,0.000113048736,0.0035748545,0.00004779607,0.019072225,0.04290097,0.0003198465],"about_ca_topic_score_codex":0.00020550001,"about_ca_topic_score_gemma":0.000003918637,"teacher_disagreement_score":0.93049103,"about_ca_system_score_codex":0.000004381053,"about_ca_system_score_gemma":0.000018624998,"threshold_uncertainty_score":0.99420327},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4323567636","doi":"10.3389/fnins.2023.1109510","title":"Research trends in the field of the gut-brain interaction: Functional dyspepsia in the spotlight – An integrated bibliometric and science mapping approach","year":2023,"lang":"en","type":"review","venue":"Frontiers in Neuroscience","topic":"Gastrointestinal motility and disorders","field":"Medicine","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Natural Science Foundation of China","keywords":"Citation; Metadata; Web of science; Library science; China; Computer science; Field (mathematics); Data science; Information retrieval; World Wide Web; Geography; Political science; MEDLINE; Archaeology","score_opus":0.23405695670620572,"score_gpt":0.44140264638534016,"score_spread":0.20734568967913444,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4323567636","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0467307,0.88561344,0.005200656,0.02986683,0.008420864,0.008909584,0.00006985797,0.00014367882,0.015044374],"genre_scores_gemma":[0.22212805,0.7669296,0.0030597975,0.0047459695,0.00031932897,0.0007614611,0.000059791633,0.00010769297,0.0018882835],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9956915,0.0011320965,0.0005693654,0.0007512833,0.0013519966,0.0005038048],"domain_scores_gemma":[0.99774975,0.0012130929,0.00016972516,0.0007020192,0.00009816361,0.00006723446],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.007868769,0.00022046185,0.0005478524,0.02917828,0.00024789994,0.00016844936,0.0015591424,0.00010168541,0.0000055222054],"category_scores_gemma":[0.0063949735,0.00010892225,0.00010790411,0.1810209,0.0016356934,0.0003166078,0.00024728008,0.0019617605,8.734667e-7],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012415966,0.0010146514,0.037439115,0.0034564275,0.000006762677,0.000117294534,0.0024758345,0.00008676939,0.00004409307,0.00054392003,0.022995366,0.9316956],"study_design_scores_gemma":[0.00075398287,0.0008535422,0.6328171,0.0074985684,0.000049439306,0.0007507069,0.016703386,0.0066271583,0.0000017436082,0.00035794327,0.33327624,0.00031017401],"about_ca_topic_score_codex":0.00018046278,"about_ca_topic_score_gemma":0.000058194186,"teacher_disagreement_score":0.93138546,"about_ca_system_score_codex":0.00013422583,"about_ca_system_score_gemma":0.00049112795,"threshold_uncertainty_score":0.9818252},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4323772350","doi":"10.3138/jsp-2022-0049","title":"What Can Be Learned about Predatory Journals from a Failed Study? Possible Motivations behind Predatory Journals","year":2023,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"","keywords":"Scrutiny; Publishing; Value (mathematics); Grammar; Documentation; Compromise; Public relations; Psychology; Sociology; Internet privacy; Computer science; Social science; Political science; Law; Linguistics","score_opus":0.62953521345765,"score_gpt":0.5478280282914975,"score_spread":0.0817071851661525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4323772350","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9597394,0.0080348,0.00066556345,0.025814287,0.004251598,0.0004516387,0.00020354243,0.00005953562,0.0007796513],"genre_scores_gemma":[0.98731506,0.00318616,0.0010432081,0.0009504287,0.0015914084,0.000010998026,0.000012197156,0.00006315212,0.0058273952],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9574059,0.0024365005,0.004229961,0.0010643782,0.03347124,0.0013920157],"domain_scores_gemma":[0.946077,0.010681242,0.0045458535,0.0015790977,0.034657437,0.0024593882],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.15450838,0.00041983527,0.001150375,0.1055856,0.0015892718,0.44051355,0.009997955,0.00057859335,0.0026389195],"category_scores_gemma":[0.32009065,0.00030651334,0.0006763601,0.12845936,0.00017817496,0.23200504,0.0021211545,0.0063811317,0.00014850804],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000089726804,0.00056756666,0.8114066,0.0000065481477,0.0003731189,0.0004195255,0.007446118,0.0005256999,0.0039040959,0.00003153453,0.119450256,0.055779204],"study_design_scores_gemma":[0.0022656105,0.00043633542,0.91334844,0.00029206031,0.00005695101,0.00006988568,0.042533323,0.0005816566,0.00036745227,0.0097383,0.029946173,0.00036382163],"about_ca_topic_score_codex":0.00025494196,"about_ca_topic_score_gemma":0.00019103172,"teacher_disagreement_score":0.20850852,"about_ca_system_score_codex":0.0006784449,"about_ca_system_score_gemma":0.002418506,"threshold_uncertainty_score":0.9999387},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","research_integrity"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4323924096","doi":"10.1186/s13104-023-06294-2","title":"Insights from using an outcomes measurement properties search filter and conducting citation searches to locate psychometric articles of tools used to measure context attributes","year":2023,"lang":"en","type":"article","venue":"BMC Research Notes","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; Canadian Agency for Drugs and Technologies in Health; Ottawa Hospital","funders":"Canadian Institutes of Health Research; Barwon Health Foundation","keywords":"Measure (data warehouse); Context (archaeology); Citation; Data science; Computer science; Information retrieval; Filter (signal processing); Medicine; World Wide Web; Data mining; Biology","score_opus":0.9924253076182565,"score_gpt":0.5832305564598512,"score_spread":0.4091947511584053,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4323924096","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9831136,0.0010150581,0.011958896,0.0018378068,0.00007869258,0.0019208556,0.000034787303,0.000016738133,0.00002358793],"genre_scores_gemma":[0.9951926,0.0000062574295,0.0043799086,0.000058348236,0.00006247009,0.0001059081,0.0000060210014,0.000023592642,0.0001649008],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.96088773,0.017646661,0.0039158273,0.0013682878,0.015476488,0.00070500566],"domain_scores_gemma":[0.97399664,0.01445953,0.00053831993,0.0021897352,0.008244745,0.0005710422],"candidate_categories":["metaresearch","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.11854617,0.00028539367,0.0019858016,0.0033348575,0.00042362476,0.0020194808,0.0013145973,0.000087808156,0.00025090278],"category_scores_gemma":[0.23282705,0.00014866322,0.00036760094,0.009713655,0.00018730019,0.000763314,0.000470721,0.0002601811,0.0008956001],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012357987,0.00017197749,0.45739803,0.00026236146,0.00035403855,0.0000037908217,0.016099196,0.00206331,0.46237686,0.00029210508,0.0014610954,0.059393656],"study_design_scores_gemma":[0.00094895274,0.0005921933,0.71888953,0.0008036508,0.00011274978,0.0000017322997,0.054622892,0.03748316,0.1810531,0.004165087,0.0007097922,0.0006171668],"about_ca_topic_score_codex":0.0008276974,"about_ca_topic_score_gemma":0.0017080194,"teacher_disagreement_score":0.28132376,"about_ca_system_score_codex":0.00011492819,"about_ca_system_score_gemma":0.00026460912,"threshold_uncertainty_score":0.99988234},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["metaresearch"],"domain":"methods","study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4324336900","doi":"10.18280/ijsdp.180214","title":"Cross-Border Tourism: What We Can Learn So Far from Eminent Scientific Publications Using Bibliometric Analysis","year":2023,"lang":"en","type":"article","venue":"International Journal of Sustainable Development and Planning","topic":"International Business and FDI","field":"Business, Management and Accounting","cited_by":9,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Tourism; Regional science; Geography; Data science; Political science; Archaeology; Computer science","score_opus":0.03280458451246968,"score_gpt":0.331625370327785,"score_spread":0.2988207858153153,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4324336900","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9919259,0.0012154484,0.002681556,0.0022389127,0.0014599558,0.000082415056,0.0000061246483,0.000042807067,0.0003468955],"genre_scores_gemma":[0.991387,0.00020594012,0.00085293374,0.00034756053,0.001124172,0.0000064383294,0.00032552183,0.000021599257,0.005728827],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99771184,0.000009212939,0.00064514234,0.0002725958,0.0010079894,0.00035323823],"domain_scores_gemma":[0.99544156,0.00011195887,0.0006537516,0.000102634614,0.0036501922,0.000039907463],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00096541934,0.00018218716,0.00025751453,0.05404305,0.00061310467,0.00716691,0.00046650897,0.00006998217,0.0004718528],"category_scores_gemma":[0.00036358452,0.0001676224,0.000117694726,0.035829466,0.00008251488,0.003924099,0.0003407518,0.00018485318,0.000024309027],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001996131,0.0002480805,0.815771,0.00015741098,0.0041939956,0.002647801,0.0059213527,0.048767824,0.00046756837,0.008530276,0.00816713,0.10492797],"study_design_scores_gemma":[0.000620605,0.000003906791,0.44086772,0.00016137792,0.00017450396,0.000016570184,0.01693643,0.013135669,0.00008526066,0.0009349316,0.52675515,0.0003079023],"about_ca_topic_score_codex":0.00057086285,"about_ca_topic_score_gemma":0.000025731635,"teacher_disagreement_score":0.518588,"about_ca_system_score_codex":0.00018193724,"about_ca_system_score_gemma":0.00026500967,"threshold_uncertainty_score":0.99386376},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4360785484","doi":"10.4108/eetpht.v8i5.3171","title":"The 100 most cited articles on wearable technology in the area of Medical Informatics: A bibliometric analysis using Web of Science","year":2022,"lang":"en","type":"article","venue":"EAI Endorsed Transactions on Pervasive Health and Technology","topic":"Scientific Research and Technology","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Wearable computer; Bibliometrics; Informatics; China; Wearable technology; Health informatics; Web of science; Smartwatch; Health technology; MEDLINE; Medical education; Computer science; Medicine; Data science; Health care; Political science; Library science","score_opus":0.03450227182355079,"score_gpt":0.3229173432682072,"score_spread":0.2884150714446564,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4360785484","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.91956806,0.0012535854,0.046569906,0.031680357,0.00014262438,0.0005254177,0.000022775652,0.00013205063,0.000105249885],"genre_scores_gemma":[0.9969449,0.0009697061,0.0016892996,0.0002616303,0.0000019536863,0.00011745453,7.6343696e-7,0.0000041784,0.000010115152],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9966868,0.00017427251,0.0006629491,0.0003734078,0.0014215914,0.00068097527],"domain_scores_gemma":[0.9977235,0.00059517694,0.00028109443,0.0010487684,0.00022574145,0.00012568402],"candidate_categories":["bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0031831192,0.00013309451,0.00035206974,0.09473686,0.0014709579,0.000057884594,0.0026915828,0.00014359337,0.00003096561],"category_scores_gemma":[0.00058819074,0.00008930157,0.000067204,0.31637824,0.00248925,0.0001678549,0.00015442497,0.00097333925,0.0000016569584],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011451655,0.0011062174,0.004684585,0.000067732464,0.00019210253,0.000038605747,0.0022981742,0.002421591,0.0020814303,0.13448106,0.0000960318,0.85241795],"study_design_scores_gemma":[0.0032524574,0.0066559706,0.0026420958,0.00013402812,0.000056394834,0.0004881597,0.04868851,0.88119775,0.023355758,0.02953482,0.0035722565,0.00042177315],"about_ca_topic_score_codex":0.00016913482,"about_ca_topic_score_gemma":0.00018682604,"teacher_disagreement_score":0.8787762,"about_ca_system_score_codex":0.00012496652,"about_ca_system_score_gemma":0.0014905292,"threshold_uncertainty_score":0.999829},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4360797522","doi":"10.1101/2023.03.22.533841","title":"Expanding Interdisciplinarity: A bibliometric study of medical education using the MEJ-24","year":2023,"lang":"en","type":"preprint","venue":"bioRxiv (Cold Spring Harbor Laboratory)","topic":"Innovations in Medical Education","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; Université de Montréal; Bibliothèque et Archives nationales du Québec","funders":"","keywords":"Variety (cybernetics); Diversity (politics); CLARITY; Bibliometrics; Web of science; Library science; Medical education; Social science; Data science; MEDLINE; Sociology; Engineering ethics; Computer science; Political science; Medicine; Biology","score_opus":0.06255270871918815,"score_gpt":0.3770019123117695,"score_spread":0.31444920359258133,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4360797522","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98550195,0.00043080567,0.0021791405,0.0032261494,0.0066096657,0.0017809548,0.000012330115,0.00025269808,0.000006285145],"genre_scores_gemma":[0.9948724,0.00012765329,0.002725193,0.00038181795,0.0013431162,0.00040212765,0.0000010949228,0.00013407467,0.000012539432],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.994963,0.00023002681,0.0012131543,0.0008330335,0.0023220167,0.00043878087],"domain_scores_gemma":[0.99551636,0.0001607201,0.0007766809,0.0017757626,0.0015374182,0.00023307293],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0037298126,0.0004185434,0.0007433032,0.01666235,0.00027624718,0.00014573315,0.0008877711,0.00066190603,0.00012597081],"category_scores_gemma":[0.0061061624,0.00034260214,0.00014584238,0.04009899,0.00023843649,0.00012292017,0.0012951087,0.0018439206,0.00002561807],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00050519715,0.033266455,0.73352396,0.0130723845,0.004899924,0.0003515107,0.0044747717,0.0002704771,0.15285975,0.0034489515,0.05224421,0.0010824085],"study_design_scores_gemma":[0.003514166,0.00075215095,0.9458521,0.009536525,0.0022712809,7.8553336e-7,0.003781827,0.013436068,0.017588567,0.000039604824,0.0017314295,0.0014955148],"about_ca_topic_score_codex":0.0003254386,"about_ca_topic_score_gemma":0.0000060287707,"teacher_disagreement_score":0.21232812,"about_ca_system_score_codex":0.00066292286,"about_ca_system_score_gemma":0.0044333595,"threshold_uncertainty_score":0.9999026},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4361276962","doi":"10.47260/amae/1335","title":"Financial Literacy: A Bibliometric Literature Review","year":2023,"lang":"en","type":"article","venue":"Advances in Management and Applied Economics","topic":"Financial Literacy and Behavior","field":"Business, Management and Accounting","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Rimouski","funders":"","keywords":"Scopus; Financial literacy; Web of science; Literacy; Bibliometrics; Maturity (psychological); Sociology; Library science; Political science; Computer science; Finance; Business; Pedagogy","score_opus":0.008532661400452354,"score_gpt":0.2419898776343379,"score_spread":0.23345721623388554,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4361276962","genre_codex":"empirical","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.74179053,0.10866152,0.00009862961,0.001768693,0.0018311485,0.0026070396,0.000035312667,0.00062479964,0.14258231],"genre_scores_gemma":[0.30878457,0.67094886,0.001466636,0.014018121,0.001260703,0.000595537,0.00044541172,0.00007520947,0.0024049291],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99870527,0.0000031813015,0.00040171374,0.00044555217,0.000083413644,0.00036086712],"domain_scores_gemma":[0.9995406,0.000038819504,0.00016028549,0.00022444874,0.000024810115,0.00001102441],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00039125507,0.0002203273,0.00030482048,0.0073556886,0.00011361527,0.00035306954,0.00024300184,0.000058917907,0.000043941673],"category_scores_gemma":[0.000027727401,0.00021988897,0.000056739605,0.01890033,0.00003654284,0.0021379357,0.00029863743,0.00014485623,0.00038489245],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000046128673,0.000064058266,0.009991268,0.0059409845,0.000006620629,0.000042234005,0.000032666907,0.00018503048,0.000001952656,0.3636877,0.007814787,0.6121866],"study_design_scores_gemma":[0.00055867544,0.0000054986886,0.06138206,0.0011769665,0.000036413898,0.0000010921767,0.000015579843,0.00021570086,0.0000026194941,0.032243494,0.904009,0.00035290737],"about_ca_topic_score_codex":0.0000026026094,"about_ca_topic_score_gemma":0.000009684637,"teacher_disagreement_score":0.8961942,"about_ca_system_score_codex":0.000020415391,"about_ca_system_score_gemma":0.0000037328086,"threshold_uncertainty_score":0.9080982},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4362590789","doi":"10.1016/j.heliyon.2023.e15184","title":"COVID-19 retracted publications on retraction watch: A systematic survey of their pre-prints and citations","year":2023,"lang":"en","type":"article","venue":"Heliyon","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; St. Joseph’s Healthcare Hamilton; Impact","funders":"","keywords":"Misinformation; Coronavirus disease 2019 (COVID-19); Citation; 2019-20 coronavirus outbreak; Scientific misconduct; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Medicine; Library science; Alternative medicine; Computer science; Pathology; Disease; Infectious disease (medical specialty)","score_opus":0.11558438210485338,"score_gpt":0.3766997325126453,"score_spread":0.26111535040779194,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4362590789","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9578506,0.00021138386,0.0007229226,0.0242541,0.0005720075,0.0011747679,0.00024370766,0.00040080215,0.014569715],"genre_scores_gemma":[0.9938141,0.0011817496,0.00003237,0.0002867444,0.000056945475,0.000048330487,0.000094569354,0.0000071032546,0.0044780606],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.99801165,0.0009729969,0.00029691972,0.00017701907,0.00036476422,0.00017666903],"domain_scores_gemma":[0.9963572,0.0028264772,0.00019358542,0.00017734367,0.0002730702,0.00017231253],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0049234508,0.000073875875,0.00017344163,0.00020922454,0.00048603176,0.000048250007,0.00016170637,0.0005033267,0.00014418016],"category_scores_gemma":[0.021647802,0.000064062064,0.00003199652,0.001263946,0.00013061255,0.00019683268,0.000019748362,0.0006084456,0.00010219963],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018924427,0.0005305627,0.16532812,0.029438559,0.00034009857,0.000006287825,0.36030394,0.00017556253,0.0033491014,0.35154188,0.087217346,0.0015793184],"study_design_scores_gemma":[0.0003379039,0.000087355336,0.9555527,0.0022343483,0.000050081366,0.0000040739014,0.015616864,0.00070191856,0.0003336221,0.0054555666,0.019320855,0.00030468617],"about_ca_topic_score_codex":0.0031687915,"about_ca_topic_score_gemma":0.0031784712,"teacher_disagreement_score":0.7902246,"about_ca_system_score_codex":0.00011655824,"about_ca_system_score_gemma":0.00030192325,"threshold_uncertainty_score":0.98659325},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4362700715","doi":"10.1101/2023.04.05.535734","title":"Second-order Citations in Altmetrics: A Case Study Analyzing the Audiences of COVID-19 Research in the News and on Social Media","year":2023,"lang":"en","type":"preprint","venue":"bioRxiv (Cold Spring Harbor Laboratory)","topic":"Misinformation and Its Impacts","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; Simon Fraser University","funders":"Social Sciences and Humanities Research Council of Canada; Queensland University of Technology","keywords":"Altmetrics; Social media; Order (exchange); CITES; Coronavirus disease 2019 (COVID-19); Public relations; Sociology; Political science; Data science; Computer science; World Wide Web; Medicine; Business","score_opus":0.22443319868326883,"score_gpt":0.4181630226410769,"score_spread":0.19372982395780808,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4362700715","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99427557,0.0001797246,0.000021645481,0.0037715232,0.00027540416,0.001243181,0.000064149506,0.00004366842,0.00012515418],"genre_scores_gemma":[0.9990557,0.00031594056,0.00007647595,0.00025738214,0.00018472246,0.0000902006,1.1853724e-7,0.00001567618,0.0000038062578],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.99560004,0.0019988234,0.0005597976,0.00038637134,0.0009891383,0.0004658026],"domain_scores_gemma":[0.9955871,0.0031003773,0.00030349742,0.0004295402,0.00041043357,0.00016903861],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.013435406,0.00018195758,0.0003032774,0.0014499517,0.00095393514,0.0004088965,0.00066395267,0.00021612193,0.000047827532],"category_scores_gemma":[0.010944738,0.00013155646,0.000047858073,0.0060538854,0.0005206516,0.00020726467,0.00023900696,0.0009468375,0.000010744383],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009617773,0.001237346,0.11645126,0.0007075785,0.00021846416,0.0011682408,0.8392832,0.000578347,0.00030620384,0.032745667,0.0070312605,0.00017622976],"study_design_scores_gemma":[0.0012908061,0.00012834811,0.6247889,0.00018476148,0.000051419534,7.207386e-8,0.36813375,0.00028563145,0.000054885106,0.00014684329,0.004376096,0.00055850315],"about_ca_topic_score_codex":0.008388991,"about_ca_topic_score_gemma":0.03647442,"teacher_disagreement_score":0.5083376,"about_ca_system_score_codex":0.0003204631,"about_ca_system_score_gemma":0.0020408279,"threshold_uncertainty_score":0.99821424},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4366210629","doi":"10.14221/ajte.2022v47n10.3","title":"Teacher Educators: A Bibliometric Mapping of an Emerging Research Area","year":2022,"lang":"en","type":"article","venue":"The Australian journal of teacher education","topic":"Teacher Education and Leadership Studies","field":"Social Sciences","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Educational research; Teacher education; Professional development; Bibliometrics; Publishing; Field (mathematics); Faculty development; Sociology; Pedagogy; Medical education; Library science; Political science; Medicine; Computer science","score_opus":0.3262156526225317,"score_gpt":0.4843248323347498,"score_spread":0.1581091797122181,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366210629","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92358536,0.0013682551,0.000011234084,0.069046706,0.0020888033,0.00027942448,0.000001910335,0.00001933352,0.00359896],"genre_scores_gemma":[0.9822859,0.00004565394,0.0002837133,0.00014134278,0.001739679,0.000048294383,0.000003849193,0.000022383985,0.0154291615],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9941656,0.002967879,0.00066446944,0.00021717968,0.001494153,0.0004907409],"domain_scores_gemma":[0.99770635,0.00025470712,0.00068649807,0.00033909036,0.0007503403,0.00026302884],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.011699226,0.00014347098,0.00025182776,0.010058716,0.0011442685,0.00013919652,0.0010519174,0.00007324108,0.0024716144],"category_scores_gemma":[0.0006218574,0.00012361405,0.00016360793,0.02162336,0.0004188048,0.00047914946,0.00007110436,0.001379572,0.000011123568],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000033013454,0.0025079944,0.08607768,0.00003104649,0.00010682538,0.0000016064927,0.6729504,0.000023382954,0.00038606577,0.0007547011,0.16112258,0.07600472],"study_design_scores_gemma":[0.0001226295,0.00010143174,0.06605511,0.000027814278,0.000028094588,0.00001680071,0.70050645,0.0000014888516,0.000025377727,0.0007959628,0.23221503,0.00010381822],"about_ca_topic_score_codex":0.0022240998,"about_ca_topic_score_gemma":0.00018896977,"teacher_disagreement_score":0.075900905,"about_ca_system_score_codex":0.0006881675,"about_ca_system_score_gemma":0.002578089,"threshold_uncertainty_score":0.9991725},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4366783310","doi":"10.1186/s41073-023-00128-2","title":"Scientific sinkhole: estimating the cost of peer review based on survey data with snowball sampling","year":2023,"lang":"en","type":"article","venue":"Research Integrity and Peer Review","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Prince Edward Island; Agricultural Research Institute of Ontario","funders":"","keywords":"Snowball sampling; Scopus; Peer review; Psychology; Demography; Medicine; Actuarial science; Geography; Statistics; MEDLINE; Political science; Business; Sociology; Mathematics","score_opus":0.952932832349722,"score_gpt":0.7128552951760203,"score_spread":0.24007753717370173,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366783310","genre_codex":"commentary","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008900476,0.45563105,0.01418203,0.4871189,0.0015437815,0.010934273,0.0070580426,0.00015732733,0.014474101],"genre_scores_gemma":[0.27519447,0.60402906,0.018777374,0.009009569,0.0006369592,0.0006736663,0.007404114,0.00020898547,0.08406579],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9388121,0.0055411663,0.0014558879,0.0018160795,0.05110046,0.0012743281],"domain_scores_gemma":[0.7242428,0.12653145,0.0011073444,0.012871005,0.13354589,0.0017015286],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.5547979,0.00021518701,0.0007798712,0.010945799,0.0010654031,0.0035327594,0.0065204636,0.00009799213,0.0006333195],"category_scores_gemma":[0.63858366,0.00010254955,0.000114888164,0.19269064,0.0012004178,0.0005447088,0.0023484963,0.0021607645,0.0005419625],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000020875057,0.00009121836,0.0071694455,0.0025214362,0.000013243253,0.000009521886,0.000026165071,0.000029473898,0.000006837244,0.00018360357,0.691698,0.2982302],"study_design_scores_gemma":[0.00024597286,0.0002283798,0.031462498,0.0154436305,0.000021467285,0.000009838818,0.000074488285,0.06817354,0.000014715915,0.00070851773,0.883397,0.00021997251],"about_ca_topic_score_codex":0.0006840166,"about_ca_topic_score_gemma":0.00030211522,"teacher_disagreement_score":0.47810936,"about_ca_system_score_codex":0.0000745736,"about_ca_system_score_gemma":0.0012074243,"threshold_uncertainty_score":0.99885476},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4367181851","doi":"10.3390/brainsci13050724","title":"Sensor-Based Rehabilitation in Neurological Diseases: A Bibliometric Analysis of Research Trends","year":2023,"lang":"en","type":"review","venue":"Brain Sciences","topic":"Stroke Rehabilitation and Recovery","field":"Medicine","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Rehabilitation; Bibliometrics; Web of science; Citation; Field (mathematics); Citation analysis; Data science; MEDLINE; Computer science; Medical education; Medicine; Library science; Political science; Physical therapy","score_opus":0.24912227028645742,"score_gpt":0.5245159785685666,"score_spread":0.2753937082821092,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4367181851","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.020676557,0.9754809,0.000007030649,0.0021205468,0.00018056932,0.00078071805,0.00011670663,0.00007227974,0.00056467496],"genre_scores_gemma":[0.0017926841,0.99608654,0.00079937855,0.000116984236,0.0000796205,0.0002012018,0.0001261459,0.000022094668,0.00077538064],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9948967,0.0014995295,0.0008968492,0.00075513724,0.0015689973,0.00038279142],"domain_scores_gemma":[0.9687464,0.030243058,0.00024234453,0.00038940329,0.00022405681,0.00015468488],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0057853926,0.0001904282,0.0017506528,0.36876068,0.00008503659,0.00004529928,0.0003196255,0.00022169533,0.00017210338],"category_scores_gemma":[0.024918571,0.00012175866,0.0011195168,0.6238513,0.0013381471,0.00006867183,0.00006380045,0.00036270483,0.00002803162],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003316431,0.00037199055,0.09900473,0.009168953,0.00019941086,0.000029367153,0.00007212987,0.00012824165,5.479173e-7,0.000106915606,0.0015565138,0.889328],"study_design_scores_gemma":[0.0004177636,0.0021359094,0.7699609,0.0045676837,0.0014661676,0.0000030952267,0.0003337478,0.002021514,5.4240726e-8,0.00007583268,0.21880706,0.00021030882],"about_ca_topic_score_codex":0.0000666361,"about_ca_topic_score_gemma":0.000016853306,"teacher_disagreement_score":0.8891177,"about_ca_system_score_codex":0.00007331146,"about_ca_system_score_gemma":0.0007341758,"threshold_uncertainty_score":0.98329496},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4367298421","doi":"10.55755/deparch.2023.18","title":"Affect, Architecture and Water: Bibliometric Analysis of the Literature","year":2023,"lang":"en","type":"article","venue":"Journal of Design Planning and Aesthetics Research","topic":"Urban Green Space and Health","field":"Environmental Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Mental Health; National Key Research and Development Program of China; Medical Research Council; National Institute of Child Health and Human Development; Natural Sciences and Engineering Research Council of Canada; Horizon 2020 Framework Programme; Office of Science; Ministerio de Ciencia e Innovación; Russian Science Foundation; National Institutes of Health; Türkiye Bilimsel ve Teknolojik Araştırma Kurumu; Chinese Academy of Sciences; National Natural Science Foundation of China; Eunice Kennedy Shriver National Institute of Child Health and Human Development; Economic and Social Research Council; U.S. Department of Energy; European Commission; Deutsche Forschungsgemeinschaft; National Aeronautics and Space Administration; National Science Foundation","keywords":"Scopus; Affect (linguistics); Architecture; Space (punctuation); Intersection (aeronautics); Perception; Computer science; Psychology; Art; Geography; Visual arts; Cartography; Communication","score_opus":0.07917268993996318,"score_gpt":0.3638443978792217,"score_spread":0.2846717079392585,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4367298421","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99666226,0.0013075231,0.00030474033,0.0015444705,0.00002071374,0.00008100127,0.0000030248875,0.0000037585273,0.00007251053],"genre_scores_gemma":[0.9983423,0.0007949063,0.00040740782,0.000027897013,0.00001942521,7.860426e-7,8.30313e-7,0.0000064871892,0.00039996894],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99830127,0.0003608851,0.00019305058,0.00011525061,0.0007457816,0.00028375478],"domain_scores_gemma":[0.9992656,0.00033188582,0.00008477017,0.00013818002,0.000050292238,0.00012924393],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0033267797,0.000070893075,0.0002018506,0.013609046,0.00019518922,0.00008116512,0.00021764266,0.00006395645,0.000018312616],"category_scores_gemma":[0.000093442024,0.000035367564,0.00006774973,0.047164366,0.00021864768,0.00006216931,0.00016030815,0.0005245729,0.0000037717923],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000085741194,0.00003524676,0.9511173,0.00004180099,0.00010500366,0.000089559,0.007364895,0.010415326,0.011158177,0.000008986518,0.0063026887,0.013275307],"study_design_scores_gemma":[0.000324704,0.00076357974,0.99006975,0.00020438088,0.00012847407,0.00014995695,0.00038232293,0.0032231964,0.0015893502,0.00089487195,0.0021696805,0.000099719466],"about_ca_topic_score_codex":0.000028579714,"about_ca_topic_score_gemma":0.000006886696,"teacher_disagreement_score":0.038952496,"about_ca_system_score_codex":0.000021383403,"about_ca_system_score_gemma":0.00001741972,"threshold_uncertainty_score":0.9975709},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4367322153","doi":"10.3138/jsp-2022-0043","title":"Reflection on ResearchGate’s Terminating ResearchGate Score, and Interest Score, as Social Media Altmetrics and Academic Evaluation Tools","year":2023,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"Social Media in Health Education","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Altmetrics; Metric (unit); Reputation; Computer science; Social media; F1 score; Weighting; Ranking (information retrieval); Remuneration; Reflection (computer programming); Sociology; Psychology; Information retrieval; Data science; World Wide Web; Business; Marketing; Artificial intelligence; Social science; Medicine; Finance","score_opus":0.7096541819502548,"score_gpt":0.5536927636685833,"score_spread":0.15596141828167143,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4367322153","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95161045,0.00056787883,0.0000051080315,0.041549325,0.0032102934,0.0006442164,0.000004591844,0.000049154132,0.002359],"genre_scores_gemma":[0.99234974,0.002180387,0.00024380551,0.0002636138,0.004802327,0.000065344626,0.000010884652,0.000028605707,0.00005526098],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99183035,0.0023856459,0.0008004194,0.00032457316,0.004000751,0.00065825623],"domain_scores_gemma":[0.983281,0.011012933,0.00086187094,0.00012221465,0.004106898,0.0006150536],"candidate_categories":["metaresearch","sts","scholarly_communication","research_integrity"],"consensus_categories":["metaresearch","scholarly_communication"],"category_scores_codex":[0.056828808,0.00013748065,0.0002764353,0.002494691,0.0017859667,0.012227685,0.00057844125,0.00050745805,0.000034931127],"category_scores_gemma":[0.44328305,0.0001371918,0.00005633342,0.0039707436,0.00035102735,0.034711663,0.00019016328,0.0040768515,0.00001452542],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001487201,0.000037994963,0.054068353,0.000080486025,0.000028416176,0.000018868786,0.06922568,0.00000228728,0.00095752906,0.0075309095,0.008515616,0.85938513],"study_design_scores_gemma":[0.0014801815,0.00048294084,0.8721478,0.0010051772,0.00006256093,0.00002684934,0.056887366,0.00016622627,0.00029488688,0.05799782,0.009136031,0.00031215878],"about_ca_topic_score_codex":0.0008275385,"about_ca_topic_score_gemma":0.0005231944,"teacher_disagreement_score":0.859073,"about_ca_system_score_codex":0.0012683047,"about_ca_system_score_gemma":0.0020206901,"threshold_uncertainty_score":0.99951357},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W4367322216","doi":"10.3138/jsp-2022-0066","title":"Who is Publishing in Biomedical Predatory Journals? A Study on Chinese Scholars","year":2023,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"China; Publishing; Misconduct; Political science; Law","score_opus":0.5387497138985565,"score_gpt":0.5629177425169369,"score_spread":0.02416802861838041,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4367322216","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9622217,0.0010842714,0.00018825018,0.030107241,0.0023754511,0.0003104643,0.000041858322,0.000039820923,0.0036309436],"genre_scores_gemma":[0.9957267,0.00023774352,0.00032111272,0.0014989092,0.0010114078,0.0000056420467,0.0000022468023,0.000039637474,0.0011565909],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9436076,0.0020527872,0.0040545696,0.001110475,0.047687076,0.0014874911],"domain_scores_gemma":[0.9649983,0.009916127,0.00245083,0.0013677373,0.019077055,0.0021899445],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.27564022,0.0003575288,0.0010303566,0.215598,0.0005998862,0.41802508,0.011789939,0.0005984516,0.0005024198],"category_scores_gemma":[0.6091589,0.00023265283,0.00048532564,0.33289507,0.00014354134,0.26619807,0.0023719463,0.010185017,0.00030350714],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008474617,0.00069341104,0.7460919,0.0000051804213,0.00006145326,0.00078621146,0.0021434047,0.000050495208,0.00039131084,0.000043655902,0.19635178,0.053296465],"study_design_scores_gemma":[0.0026800565,0.000826095,0.93627834,0.00016793041,0.000008700473,0.00010859203,0.007101155,0.00078460324,0.000022015472,0.008076262,0.043694712,0.000251548],"about_ca_topic_score_codex":0.00008155908,"about_ca_topic_score_gemma":0.000018205597,"teacher_disagreement_score":0.33351862,"about_ca_system_score_codex":0.00066279684,"about_ca_system_score_gemma":0.0012323802,"threshold_uncertainty_score":0.99355674},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4367856932","doi":"10.1016/j.sleep.2023.05.002","title":"The Italian contribution to pediatric sleep medicine: A scientometric analysis","year":2023,"lang":"en","type":"article","venue":"Sleep Medicine","topic":"Sleep and related disorders","field":"Psychology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital; University of Ottawa","funders":"","keywords":"Sleep medicine; Citation; Web of science; Sleep (system call); Bibliometrics; Psychopathology; Sleep patterns; Psychology; Psychiatry; Medicine; Sleep disorder; Library science; Computer science; Cognition; Meta-analysis; Pathology; Electroencephalography","score_opus":0.016957001446036256,"score_gpt":0.3364399052407403,"score_spread":0.31948290379470407,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4367856932","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.816247,0.021561306,0.004755243,0.08986468,0.016211864,0.001826502,0.000031829124,0.0010309577,0.04847066],"genre_scores_gemma":[0.9912114,0.0006806607,0.0000064456813,0.001474134,0.001293833,0.00011543457,0.0001647624,0.00003179815,0.0050215386],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99642664,0.00023347184,0.0006851326,0.000651879,0.0010390076,0.00096386287],"domain_scores_gemma":[0.99722254,0.0009568277,0.0002044803,0.00082900835,0.0003078177,0.00047930388],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0026374771,0.00028615596,0.0005298352,0.0056703785,0.00055870303,0.000024266645,0.0006258144,0.00021910516,0.0022961346],"category_scores_gemma":[0.0022832963,0.00016538423,0.00018039378,0.048040472,0.00040779336,0.000055824028,0.000097843025,0.00035932625,0.0032105958],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004203305,0.00035217975,0.17825572,0.000029747325,0.008603389,0.00026561334,0.010904488,0.0009012428,0.0001725977,0.018909913,0.6436076,0.13757715],"study_design_scores_gemma":[0.011405501,0.0018018623,0.7399567,0.00005150186,0.013178495,0.000023899673,0.01240115,0.0026556784,0.000011709746,0.0016112255,0.2160838,0.0008184822],"about_ca_topic_score_codex":0.0003583278,"about_ca_topic_score_gemma":0.00009077477,"teacher_disagreement_score":0.561701,"about_ca_system_score_codex":0.0000980975,"about_ca_system_score_gemma":0.00001986178,"threshold_uncertainty_score":0.9986159},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4372348940","doi":"10.18280/ijsdp.180423","title":"Identifying the Absence of Knowledge Using Bibliometric Analysis: The Case of the “Bad Behavior Index”","year":2023,"lang":"en","type":"article","venue":"International Journal of Sustainable Development and Planning","topic":"Corruption and Economic Development","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Index (typography); Statistics; Econometrics; Computer science; Mathematics","score_opus":0.072848639121632,"score_gpt":0.375874906200116,"score_spread":0.303026267078484,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4372348940","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9974615,0.00046467877,0.0005643434,0.0002207154,0.0005588408,0.00011931335,0.0000011700988,0.000003495538,0.0006059383],"genre_scores_gemma":[0.9986535,0.00008259766,0.00014629099,0.000019220524,0.000071463604,0.000004837877,7.3852976e-7,0.0000032401808,0.0010181444],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99882245,0.00009645226,0.00048963015,0.000075032985,0.00034692843,0.00016949546],"domain_scores_gemma":[0.99833846,0.00029514142,0.0005366426,0.000063337655,0.00072768383,0.000038721053],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0029243247,0.000065453954,0.0001382919,0.006943641,0.0005029561,0.00012350718,0.00047431147,0.000032921118,0.000043373362],"category_scores_gemma":[0.0002989208,0.000038150196,0.00008688088,0.014757577,0.00012961155,0.00022556158,0.00019878853,0.00012934927,0.0000010717138],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004717723,0.00006412444,0.74762315,0.000038649257,0.0011188278,0.0011716685,0.19422331,0.0037973223,0.00009244661,0.0034470106,0.0002369675,0.048139367],"study_design_scores_gemma":[0.00019603538,0.000004702989,0.7321571,0.00008894879,0.00009447857,0.00012340913,0.26053774,0.00048293968,0.000121432444,0.00017106844,0.0059451377,0.0000769987],"about_ca_topic_score_codex":0.00044626102,"about_ca_topic_score_gemma":0.00012505152,"teacher_disagreement_score":0.066314444,"about_ca_system_score_codex":0.00015556956,"about_ca_system_score_gemma":0.00057734014,"threshold_uncertainty_score":0.7090527},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4376128787","doi":"10.2337/ds22-0085","title":"Trends in Time in Range–Related Publications and Clinical Trials: A Bibliometric Review","year":2023,"lang":"en","type":"review","venue":"Diabetes Spectrum","topic":"Diabetes Treatment and Management","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Diabetes Canada","funders":"Abbott Diabetes Care; Abbott Laboratories","keywords":"Medicine; Clinical trial; Observational study; MEDLINE; Trial registration; Family medicine; Internal medicine","score_opus":0.20441163780054222,"score_gpt":0.4697637114838882,"score_spread":0.26535207368334596,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4376128787","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00013944345,0.9895162,7.972993e-8,0.0022277362,0.00018906867,0.0029382,0.000056612895,0.00020535378,0.004727314],"genre_scores_gemma":[0.0000061303713,0.9843892,0.000027421987,0.00021329835,0.000121650504,0.0007958396,0.0014107015,0.000101268,0.012934511],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99390006,0.0011112358,0.0030238305,0.00087188586,0.00040786812,0.0006851345],"domain_scores_gemma":[0.9959575,0.0022488222,0.00075257476,0.0007567082,0.000027468452,0.00025696162],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.009228255,0.00054906966,0.006161773,0.06175073,0.00002686509,0.00009220825,0.00030364544,0.00043005683,0.0009950969],"category_scores_gemma":[0.003170419,0.00041339797,0.0011191136,0.10038867,0.00008354299,0.00012020432,0.00023115311,0.00071141165,0.001092177],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000028348568,0.00045682144,0.009641639,0.018677723,0.0008496551,0.000030903007,0.0000030619565,8.021528e-9,9.779146e-9,0.00009996069,0.01659996,0.9536374],"study_design_scores_gemma":[0.0021304903,0.00027115154,0.018440701,0.06492896,0.005563202,0.0000022223603,0.0000011461083,0.00002372864,3.6690178e-8,0.00013024056,0.90814054,0.00036759448],"about_ca_topic_score_codex":0.000014480322,"about_ca_topic_score_gemma":0.000026848546,"teacher_disagreement_score":0.95326984,"about_ca_system_score_codex":0.0002551376,"about_ca_system_score_gemma":0.0001418314,"threshold_uncertainty_score":0.9999181},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4377047099","doi":"10.1177/23259671221134117","title":"Bibliometric Analysis of the Top 50 Most Cited Studies on Achilles Tendon Pathology","year":2023,"lang":"en","type":"article","venue":"Orthopaedic Journal of Sports Medicine","topic":"Tendon Structure and Treatment","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Achilles tendon; Medicine; Bibliometrics; MEDLINE; Analytics; Achilles tendon rupture; Tendinopathy; Tendon; Pathology; Library science; Data science; Computer science","score_opus":0.028861024678751603,"score_gpt":0.32525718808835974,"score_spread":0.2963961634096081,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4377047099","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9821302,0.0095074065,0.000012107991,0.0063338475,0.001195582,0.00021226525,0.000015433952,0.000028352173,0.0005647974],"genre_scores_gemma":[0.99049985,0.0076137865,0.00006293659,0.0007896719,0.00051019684,0.0000028094698,0.000029499748,0.000020961048,0.0004702619],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99666554,0.00007260431,0.0012302889,0.00023387105,0.0015039395,0.00029373175],"domain_scores_gemma":[0.99708235,0.00037716757,0.0011966683,0.0004889779,0.0006483371,0.00020651492],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.001284943,0.00025255987,0.0014351873,0.06343296,0.00009471867,0.000003192522,0.00017208533,0.00010113119,0.00023210778],"category_scores_gemma":[0.0011757965,0.00011655609,0.0005261714,0.12439192,0.0003713974,0.000041626587,0.00006230367,0.00040767304,0.000002449508],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003961039,0.00015105949,0.89478153,0.000072135015,0.0048702075,0.0026993705,0.00075435947,0.00013081908,0.0006205675,0.00018694212,0.0051859836,0.09015089],"study_design_scores_gemma":[0.0029246286,0.0009988657,0.9828708,0.00064852944,0.007723957,0.00055305474,0.001202393,0.000009441158,0.00027932887,0.00021573152,0.0024909975,0.00008223808],"about_ca_topic_score_codex":0.000010744398,"about_ca_topic_score_gemma":0.0000038180424,"teacher_disagreement_score":0.09006865,"about_ca_system_score_codex":0.000068764195,"about_ca_system_score_gemma":0.00013094043,"threshold_uncertainty_score":0.94718224},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4377138204","doi":"10.7759/cureus.39224","title":"Recent Trend in Artificial Intelligence-Assisted Biomedical Publishing: A Quantitative Bibliometric Analysis","year":2023,"lang":"en","type":"review","venue":"Cureus","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Google; Microsoft","keywords":"Medicine; Publishing; MEDLINE; Artificial intelligence; Library science; Computer science","score_opus":0.681906452458046,"score_gpt":0.5656427110911867,"score_spread":0.11626374136685935,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4377138204","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0001373581,0.994631,0.0004684669,0.0015593127,0.0015451172,0.0010189924,0.00010063244,0.00015494009,0.00038416474],"genre_scores_gemma":[0.00014228726,0.995406,0.0006061201,0.000075552474,0.0006419665,0.00027832034,0.0025843193,0.000073768795,0.00019169762],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9944179,0.00040619072,0.002623367,0.00092720444,0.0008976458,0.00072770886],"domain_scores_gemma":[0.9957523,0.001921208,0.00067606877,0.0007192469,0.00040396035,0.0005272685],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0021036323,0.00047838115,0.002568212,0.122311145,0.00011016226,0.00031602627,0.00045163574,0.0009942885,0.00091802963],"category_scores_gemma":[0.008794272,0.00039092533,0.000820637,0.40295678,0.00019151477,0.00031652927,0.000110538036,0.0012355162,0.0006420778],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024525523,0.0005498731,0.00007187043,0.0030601313,0.0009753527,0.00009522259,0.0005601781,0.00000115878,1.6714839e-7,0.00057636277,0.002865031,0.9912201],"study_design_scores_gemma":[0.000016838027,0.0003783148,0.00023205217,0.0051853927,0.00450737,0.000030529434,0.0023801972,0.0003561933,0.0000028852685,0.001276391,0.98519593,0.00043792586],"about_ca_topic_score_codex":0.001697067,"about_ca_topic_score_gemma":0.007379342,"teacher_disagreement_score":0.9907822,"about_ca_system_score_codex":0.0010872333,"about_ca_system_score_gemma":0.0016536125,"threshold_uncertainty_score":0.9999953},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4377821117","doi":"10.2196/41388","title":"Assessment of the Dissemination of COVID-19–Related Articles Across Social Media: Altmetrics Study","year":2023,"lang":"en","type":"article","venue":"JMIR Formative Research","topic":"Misinformation and Its Impacts","field":"Social Sciences","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Altmetrics; Citation; Social media; Bibliometrics; Scopus; Coronavirus disease 2019 (COVID-19); Dimension (graph theory); Citation analysis; Information Dissemination; MEDLINE; Library science; Medicine; Computer science; World Wide Web; Political science; Mathematics","score_opus":0.18063523789001237,"score_gpt":0.5805738851175438,"score_spread":0.3999386472275315,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4377821117","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98675674,0.0000051650595,0.000033252512,0.0019310289,0.000119006974,0.0007100429,0.000053248215,0.00003551115,0.010356025],"genre_scores_gemma":[0.9995825,0.000032090255,0.0000050554786,0.0000126968835,0.000024033081,0.000021310609,0.0000080432865,0.000004455696,0.00030978414],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99569327,0.0010263115,0.00041810976,0.000083201616,0.0023793306,0.00039978986],"domain_scores_gemma":[0.9963926,0.0025120429,0.0002282066,0.00014862535,0.00057552446,0.0001429942],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00970821,0.000061122395,0.00015039863,0.0003660844,0.0011657916,0.0000747252,0.00041145293,0.00008006824,0.00022569693],"category_scores_gemma":[0.007831557,0.0000425449,0.00006487294,0.005748107,0.0006499001,0.0005243938,0.00021871857,0.00027644995,0.000029992745],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000092114005,0.0002442104,0.005418642,0.000051513594,0.000021674961,4.6284435e-7,0.96358144,0.000014891805,0.00011959054,0.012952195,0.0071485243,0.010437611],"study_design_scores_gemma":[0.00027237803,0.00007288974,0.4770223,0.0000129859445,0.0000020909497,8.8336094e-8,0.5203957,0.00048486333,0.00018272096,0.0009579234,0.0005581767,0.000037909907],"about_ca_topic_score_codex":0.00018363248,"about_ca_topic_score_gemma":0.00032801254,"teacher_disagreement_score":0.47160366,"about_ca_system_score_codex":0.00028815438,"about_ca_system_score_gemma":0.0005898685,"threshold_uncertainty_score":0.937567},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","scholarly_communication"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4377940628","doi":"10.1007/s00404-023-07081-x","title":"Exploring the use of ChatGPT in OBGYN: a bibliometric analysis of the first ChatGPT-related publications","year":2023,"lang":"en","type":"article","venue":"Archives of Gynecology and Obstetrics","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":40,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Jewish General Hospital","funders":"","keywords":"Bibliometrics; Medicine; Obstetrics and gynaecology; Library science; Pregnancy; Computer science","score_opus":0.3059902072136145,"score_gpt":0.37763018089054096,"score_spread":0.07163997367692648,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4377940628","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9947345,0.0005368096,0.000024529036,0.0039050402,0.00033793977,0.00027002697,0.000010243247,0.000011139269,0.00016976465],"genre_scores_gemma":[0.994926,0.0047307084,0.00006778461,0.00008060066,0.000008341458,0.000037189566,0.000015689031,0.0000053408203,0.00012838536],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9989169,0.00008648737,0.0005437601,0.00014466111,0.00013938833,0.00016880497],"domain_scores_gemma":[0.9857377,0.0135021815,0.0002457055,0.00035631913,0.00011611882,0.00004196462],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00029755852,0.00006667877,0.00028416276,0.027448857,0.00009996393,0.0000050890503,0.00015123568,0.000058584057,0.000017733328],"category_scores_gemma":[0.010668793,0.00004328584,0.000114781986,0.115726136,0.0003220763,0.000104331826,0.00008581298,0.0001773513,0.000001772655],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001631608,0.00013472875,0.88652086,0.00008576776,0.00019029273,7.179888e-7,0.0074357605,0.0007000712,0.00009056406,0.0016135253,0.00004649493,0.10316488],"study_design_scores_gemma":[0.00005819542,0.00009637937,0.98939925,0.000031618583,0.0002126514,8.2856815e-7,0.0011735038,0.006254636,0.00078532397,0.00044636484,0.0015070246,0.0000342472],"about_ca_topic_score_codex":0.00046180023,"about_ca_topic_score_gemma":0.000489988,"teacher_disagreement_score":0.10313064,"about_ca_system_score_codex":0.000015598916,"about_ca_system_score_gemma":0.000097578486,"threshold_uncertainty_score":0.99766475},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4378717516","doi":"10.54452/jrb.1239959","title":"IS IT A MAN’S WORLD? A BIBLIOMETRIC ANALYSIS OF WOMEN’S ENTREPRENEURSHIP LITERATURE FROM A GENDER PERSPECTIVE","year":2023,"lang":"en","type":"article","venue":"Journal of Research in Business","topic":"Entrepreneurship Studies and Influences","field":"Business, Management and Accounting","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Entrepreneurship; Citation; Perspective (graphical); Gender gap; Political science; Sociology; Gender studies; Social science; Demographic economics; Economics; Law","score_opus":0.1354580341183166,"score_gpt":0.3920419981871948,"score_spread":0.2565839640688782,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4378717516","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9848874,0.0028496855,0.0000047047006,0.010207562,0.00027747377,0.00014176332,0.000023667006,0.00001804165,0.0015896742],"genre_scores_gemma":[0.99583125,0.002488744,0.000033912245,0.00057853665,0.0007648115,0.00001920362,0.000010706759,0.000019698024,0.00025316444],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99622357,0.000112438895,0.0007419668,0.00039132003,0.001751782,0.00077892415],"domain_scores_gemma":[0.9938101,0.00075115706,0.00056295685,0.00035695863,0.0044646338,0.000054170752],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.003563564,0.00020277276,0.00072353036,0.18018897,0.00013041706,0.0005468491,0.00073690974,0.00008584872,0.0007896994],"category_scores_gemma":[0.002101262,0.00015854857,0.00028006293,0.44468114,0.00017096793,0.001375742,0.00044256894,0.00057376357,0.0000569415],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037261032,0.00025914726,0.9725511,0.0002901783,0.0012586056,0.00046059262,0.0054409774,0.0008807426,0.00016164049,0.00085804035,0.01567705,0.0017892864],"study_design_scores_gemma":[0.00066371495,0.000019409288,0.9752497,0.00042722916,0.00018358121,0.000001835751,0.010268469,0.00044423467,0.000024909812,0.0047938954,0.007765719,0.00015728558],"about_ca_topic_score_codex":0.0017909948,"about_ca_topic_score_gemma":0.00032091784,"teacher_disagreement_score":0.26449215,"about_ca_system_score_codex":0.0003514277,"about_ca_system_score_gemma":0.00010705833,"threshold_uncertainty_score":0.86466545},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4378900866","doi":"10.15354/sief.23.or249","title":"Trends and Issues in Science Education in the New Millennium: A Bibliometric Analysis of the JRST","year":2023,"lang":"en","type":"article","venue":"Science Insights Education Frontiers","topic":"Education and Critical Thinking Development","field":"Social Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Context (archaeology); Publishing; Web of science; Descriptive statistics; Bibliometrics; Science education; Sociology; Social science; Political science; Mathematics education; Library science; Psychology; Computer science; Geography; MEDLINE; Mathematics","score_opus":0.027042430556658045,"score_gpt":0.3749138997333006,"score_spread":0.3478714691766425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4378900866","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93593353,0.0014422066,0.000015705193,0.043586772,0.002566086,0.00020629895,5.5462885e-7,0.000018696319,0.016230132],"genre_scores_gemma":[0.99261194,0.0008446794,0.0007416772,0.0007164461,0.00006323494,0.000021908394,0.0000023104171,0.0000026694643,0.004995159],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9976543,0.0001695353,0.00028071803,0.00035950684,0.0012022917,0.00033363295],"domain_scores_gemma":[0.999044,0.00012078704,0.00009912872,0.00031872882,0.00027192596,0.00014540048],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0034011523,0.0000815015,0.00013894134,0.05937444,0.0008160619,0.0003196182,0.0010934658,0.000040347353,0.000033659977],"category_scores_gemma":[0.0015443016,0.000052993248,0.00004129509,0.42035973,0.0023816838,0.0006879187,0.00006587167,0.00011868531,0.0000052790733],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000026951097,0.00032518708,0.23142725,0.000007619436,0.000010539816,1.1839875e-7,0.2625693,0.000030007286,0.0001926903,0.11254021,0.041853484,0.3510409],"study_design_scores_gemma":[0.00003944276,0.0000056632525,0.8481052,0.000022852908,0.000019620315,1.3409723e-7,0.09985198,0.00009988276,0.000076809076,0.008657468,0.043043043,0.00007790632],"about_ca_topic_score_codex":0.0042984523,"about_ca_topic_score_gemma":0.0027419513,"teacher_disagreement_score":0.61667794,"about_ca_system_score_codex":0.00033736485,"about_ca_system_score_gemma":0.013330463,"threshold_uncertainty_score":0.992263},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4379364963","doi":"10.1002/asi.24758","title":"A comparative study of the coverage of African journals in Web of Science, Scopus, and CrossRef","year":2023,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":63,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Council of Prairie and Pacific University Libraries; Western University","funders":"","keywords":"Scopus; Web of science; Directory; Library science; Citation; Bibliometrics; World Wide Web; Computer science; MEDLINE; Political science","score_opus":0.28533283705545054,"score_gpt":0.536237433066305,"score_spread":0.25090459601085446,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379364963","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9972012,0.00004301766,0.000026777769,0.0014185997,0.00023109077,0.00029564416,0.00001796197,0.000001968682,0.0007637553],"genre_scores_gemma":[0.9998111,0.000082287595,0.000042444568,0.000015191428,0.00000414487,0.000002215195,3.632737e-8,7.9705285e-7,0.00004176566],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9920731,0.00007773766,0.0012167448,0.00010592225,0.006293956,0.00023249432],"domain_scores_gemma":[0.9843748,0.0012741241,0.0032394358,0.00024278711,0.010815318,0.00005349106],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.04947075,0.000053283635,0.00031891782,0.03845652,0.0003966199,0.00037365788,0.002044581,0.000059360354,0.0000021285502],"category_scores_gemma":[0.055764697,0.000026755806,0.000051665404,0.18819375,0.0014406062,0.0022447044,0.00068144675,0.00019951239,6.576706e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000029977215,0.000115017276,0.965319,0.0000079594565,0.0000150061405,1.0362247e-7,0.005175968,0.00029777482,0.010932993,0.0032833884,0.00056850776,0.014254318],"study_design_scores_gemma":[0.0011263799,0.0004335156,0.94875145,0.00003057155,0.000007590939,0.0000060180973,0.022925429,0.0037460744,0.012902012,0.0089925565,0.0010343724,0.00004403226],"about_ca_topic_score_codex":0.000011209703,"about_ca_topic_score_gemma":0.000020974448,"teacher_disagreement_score":0.14973724,"about_ca_system_score_codex":0.00016990052,"about_ca_system_score_gemma":0.001097817,"threshold_uncertainty_score":0.9787699},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4379930290","doi":"10.7759/cureus.40143","title":"Plain Packaging in Tobacco Control: A Bibliometric Analysis","year":2023,"lang":"en","type":"review","venue":"Cureus","topic":"Consumer Attitudes and Food Labeling","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Tobacco control; Packaging and labeling; Medicine; Tobacco industry; Bibliometrics; Library science; Environmental health; Public health; Marketing; Business","score_opus":0.10590408474609861,"score_gpt":0.40130031947162315,"score_spread":0.29539623472552456,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379930290","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000046743626,0.99817866,0.00007854878,0.000046348203,0.00018200769,0.00068758323,0.00006051076,0.00012984329,0.0005897539],"genre_scores_gemma":[0.0010648858,0.9979478,0.000094227726,0.000059012167,0.0001584704,0.0001047057,0.00016611422,0.00007600571,0.00032875492],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.997662,0.00016645656,0.00082848564,0.0005366586,0.00035295912,0.00045343558],"domain_scores_gemma":[0.9980737,0.00081000867,0.00023946808,0.00062652095,0.00007385589,0.00017642626],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0011302178,0.00037360218,0.002961939,0.090420656,0.00005858327,0.000064224216,0.00022865974,0.00029163153,0.00021038433],"category_scores_gemma":[0.00080042885,0.0002987005,0.0010111053,0.24101712,0.00002542364,0.000048013488,0.000096282514,0.0008556656,0.00027014912],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000072142575,0.00007637154,0.008211544,0.0063306475,0.009393669,0.0003129428,0.00002906675,0.0000026259654,1.4416234e-7,0.000027020033,0.00019828934,0.97541046],"study_design_scores_gemma":[0.00089193205,0.000060541795,0.0033588405,0.010531429,0.046012774,0.000031873144,0.00002306809,0.000110245426,3.744371e-8,0.000007823312,0.9385495,0.00042191558],"about_ca_topic_score_codex":0.00030554045,"about_ca_topic_score_gemma":0.00064507086,"teacher_disagreement_score":0.9749885,"about_ca_system_score_codex":0.00024124644,"about_ca_system_score_gemma":0.00023274096,"threshold_uncertainty_score":0.99994653},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4380303151","doi":"10.1108/medar-09-2022-1809","title":"Mapping the state of expanded audit reporting: a bibliometric view","year":2023,"lang":"en","type":"article","venue":"Meditari Accountancy Research","topic":"Auditing, Earnings Management, Governance","field":"Business, Management and Accounting","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Accounting; Audit; Originality; Corporate governance; Business; Shareholder; Transparency (behavior); Citation; Political science; Sociology; Public relations; Law; Finance","score_opus":0.12192921496022187,"score_gpt":0.351107099280899,"score_spread":0.22917788432067715,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4380303151","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95795035,0.0012683877,0.004495136,0.005636414,0.0015139685,0.0014734989,0.000010273712,0.00054791226,0.027104078],"genre_scores_gemma":[0.9903778,0.0007698091,0.000059162154,0.00047400364,0.0020896222,0.00023752569,0.000041444127,0.00009317162,0.0058574597],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.992584,0.00009324152,0.0015379007,0.0007398328,0.0036891366,0.001355926],"domain_scores_gemma":[0.9799122,0.0008728383,0.01662289,0.0011009284,0.0014528893,0.000038205897],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.016274698,0.0003027709,0.0005331311,0.021557916,0.0007534424,0.0006228071,0.0016065567,0.00008510139,0.0007043127],"category_scores_gemma":[0.0751463,0.00023180169,0.00017486024,0.122474305,0.00035148082,0.0014296543,0.0017889612,0.0009961682,0.0025751675],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000058516216,0.00012764668,0.10199703,0.0018263541,0.00017567544,0.000381556,0.00082324044,0.00019554913,0.0014371849,0.0018379813,0.29834205,0.5927972],"study_design_scores_gemma":[0.00054211763,0.000024077566,0.34751016,0.00047395853,0.000022766048,0.0000056884364,0.001310729,0.0018243389,0.00013997484,0.009421315,0.6383763,0.00034858557],"about_ca_topic_score_codex":0.0020988977,"about_ca_topic_score_gemma":0.00031937155,"teacher_disagreement_score":0.59244865,"about_ca_system_score_codex":0.00014115898,"about_ca_system_score_gemma":0.00015416266,"threshold_uncertainty_score":0.99820143},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4380364391","doi":"10.3176/earth.2023.56","title":"Ordovician scientometrics","year":2023,"lang":"en","type":"article","venue":"Proceedings of the Estonian Academy of Sciences Geology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Ordovician; Scientometrics; Paleontology; Geology; Computer science; Library science","score_opus":0.5186620442457053,"score_gpt":0.5528916263581309,"score_spread":0.03422958211242566,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4380364391","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9661598,0.00027958435,0.000015764745,0.013142963,0.000434382,0.00026812856,0.0000111230665,0.000041568306,0.019646654],"genre_scores_gemma":[0.9949108,0.00016469811,0.0010527688,0.00027434592,0.000058334288,0.0000057549664,1.0360748e-7,0.00000793835,0.0035252958],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9862663,0.00005595732,0.0011457996,0.00094877294,0.010534666,0.0010484974],"domain_scores_gemma":[0.99356526,0.0028305799,0.0011513393,0.00019136934,0.00195961,0.00030182098],"candidate_categories":["metaresearch","bibliometrics","sts","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.042427246,0.00018115237,0.0005106112,0.070001386,0.0005756867,0.0004118583,0.010776695,0.00024402328,0.0001391794],"category_scores_gemma":[0.047562882,0.00010499151,0.00025708464,0.46512,0.0050317035,0.0008231851,0.0030768614,0.00045265508,0.000290216],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000109348275,0.00006519198,0.8734533,0.000021285017,0.000010617019,2.3084588e-7,0.00036518878,0.000121958,0.016112538,0.06856273,0.012275656,0.029000325],"study_design_scores_gemma":[0.0002493371,0.00021188405,0.86546236,0.000019680816,0.000007646155,0.000012610932,0.00087745144,0.0063000643,0.018912602,0.09181129,0.01596008,0.00017501193],"about_ca_topic_score_codex":0.000037281345,"about_ca_topic_score_gemma":0.0000016914589,"teacher_disagreement_score":0.39511862,"about_ca_system_score_codex":0.00005375551,"about_ca_system_score_gemma":0.00019267686,"threshold_uncertainty_score":0.997676},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4381166507","doi":"10.3389/fpsyg.2023.1175856","title":"From criminal interrogations to investigative interviews: a bibliometric study","year":2023,"lang":"en","type":"review","venue":"Frontiers in Psychology","topic":"Deception detection and forensic psychology","field":"Psychology","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Criminal investigation; Psychology; Field (mathematics); Dialog box; Variety (cybernetics); Criminology; Computer science; World Wide Web","score_opus":0.2660349935338887,"score_gpt":0.49634444218704726,"score_spread":0.23030944865315855,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4381166507","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0003261748,0.92287284,0.007996258,0.0002790522,0.054383457,0.0035262187,0.00032134767,0.0004632882,0.009831392],"genre_scores_gemma":[0.000073938754,0.9841389,0.0035111813,0.0021393797,0.0011750123,0.0044137696,0.00031101087,0.00032071705,0.003916061],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9913396,0.0025769072,0.0023338133,0.0024022886,0.0003558851,0.0009914924],"domain_scores_gemma":[0.9963071,0.0003796748,0.0007634469,0.0019803604,0.00011780113,0.0004516371],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0012104093,0.00094468263,0.0033547415,0.08333375,0.000096567295,0.00007578096,0.0016619064,0.001159747,0.0031646185],"category_scores_gemma":[0.00044962286,0.0009149368,0.00066072267,0.07379517,0.00049871637,0.000106616324,0.000339011,0.0018232537,0.016018309],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000055017204,0.00057894026,0.00014531921,0.00016495762,0.0006459454,0.00016872816,0.0064951344,8.4523954e-8,6.846646e-8,0.000030058241,0.34029454,0.6514212],"study_design_scores_gemma":[0.001400823,0.0010628215,0.0022804448,0.0015916656,0.000648512,0.00007533643,0.011805173,0.0000011957582,2.8899644e-8,0.004001551,0.9763823,0.000750143],"about_ca_topic_score_codex":0.00027390054,"about_ca_topic_score_gemma":0.00036435336,"teacher_disagreement_score":0.65067106,"about_ca_system_score_codex":0.00029810908,"about_ca_system_score_gemma":0.00009192776,"threshold_uncertainty_score":0.9993301},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4382050189","doi":"10.1108/mrr-09-2022-0646","title":"A multi-method bibliometric review of value co-creation research","year":2023,"lang":"en","type":"article","venue":"Management Research Review","topic":"Service and Product Innovation","field":"Business, Management and Accounting","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Francis Xavier University","funders":"","keywords":"Knowledge management; Scopus; Value (mathematics); Extant taxon; Resource (disambiguation); Co-creation; Citation; Data science; Computer science; Strategic management; Sociology; Business; Marketing; Political science; World Wide Web","score_opus":0.3664722735043016,"score_gpt":0.5451507605715484,"score_spread":0.17867848706724682,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382050189","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0007634725,0.62654877,0.0012130542,0.10092927,0.00054230273,0.0134582175,0.0000124474345,0.00059520226,0.25593728],"genre_scores_gemma":[0.00044620835,0.9795398,0.0027611647,0.010777919,0.0005970742,0.0008039309,0.0004949478,0.000059450478,0.0045194705],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9936921,0.0006564151,0.0009355809,0.0006984453,0.0031400898,0.0008773615],"domain_scores_gemma":[0.99513483,0.00045039563,0.00028995713,0.0011630494,0.002932632,0.000029124345],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.05566521,0.00020968521,0.0005501448,0.039694317,0.00038960163,0.0002297861,0.001065553,0.00006261558,0.0009943468],"category_scores_gemma":[0.0032802536,0.00018107769,0.00014999241,0.2596971,0.00011012075,0.00090686645,0.000977342,0.00053425395,0.006253188],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000008954297,0.00011361098,0.0004070493,0.28502312,0.000056525463,0.000018342891,0.000006091074,0.0000015796554,0.00012586155,0.028351882,0.3932026,0.29268438],"study_design_scores_gemma":[0.0002728554,0.000024812725,0.009909549,0.048307143,0.00007214282,6.4859756e-7,0.00007536923,0.00060554076,0.000036704343,0.0017704798,0.9387461,0.00017863984],"about_ca_topic_score_codex":0.0003999388,"about_ca_topic_score_gemma":0.0000045160878,"teacher_disagreement_score":0.5455435,"about_ca_system_score_codex":0.000090133544,"about_ca_system_score_gemma":0.000044608398,"threshold_uncertainty_score":0.9999189},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4382704124","doi":"10.1504/ijbbm.2023.10057330","title":"A topic modelling based bibliometric exploration of international business research","year":2023,"lang":"en","type":"article","venue":"International Journal of Bibliometrics in Business and Management","topic":"Big Data and Business Intelligence","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Data science; Bibliometrics; Regional science; Computer science; Management science; Geography; Engineering; Library science","score_opus":0.32154758182946624,"score_gpt":0.40990671870098067,"score_spread":0.08835913687151442,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4382704124","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.45996448,0.0018989558,0.4947903,0.016343508,0.013129079,0.0008411653,0.000039567487,0.00011823964,0.012874714],"genre_scores_gemma":[0.9751665,0.020781508,0.0025257994,0.0002189159,0.001031554,0.000021562792,0.000074306445,0.000029391276,0.00015045665],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9959695,0.00002555754,0.0009829291,0.00030878044,0.0023959167,0.000317331],"domain_scores_gemma":[0.9920842,0.0002517492,0.00065156445,0.00022307859,0.006763915,0.000025488758],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0032825908,0.00018814176,0.00031533578,0.3821204,0.00007764763,0.00073221256,0.0012641193,0.00009556793,0.000111705136],"category_scores_gemma":[0.00091954286,0.00017206818,0.000072759096,0.3633323,0.00011888398,0.0039675306,0.00071158324,0.00023121088,0.000032992364],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010520472,0.0019180145,0.15747051,0.00260035,0.00057830004,0.0011226498,0.00014569037,0.19968097,0.0005217932,0.07041641,0.023540525,0.54095274],"study_design_scores_gemma":[0.0028896045,0.000036570826,0.6001021,0.0013623254,0.00007490412,0.00002125634,0.00056786765,0.10399656,0.00014196956,0.013451123,0.27684867,0.000507025],"about_ca_topic_score_codex":0.0003915105,"about_ca_topic_score_gemma":0.000022204844,"teacher_disagreement_score":0.5404457,"about_ca_system_score_codex":0.000095119765,"about_ca_system_score_gemma":0.000056403092,"threshold_uncertainty_score":0.7060743},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4383198701","doi":"10.3389/fonc.2023.1197168","title":"Global trends in BRCA-related breast cancer research from 2013 to 2022: A scientometric analysis","year":2023,"lang":"en","type":"article","venue":"Frontiers in Oncology","topic":"PARP inhibition in cancer therapy","field":"Medicine","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Breast cancer; Web of science; Bibliometrics; Oncology; Medicine; Cancer; Incidence (geometry); Demography; Internal medicine; Library science; Meta-analysis; Computer science","score_opus":0.058921481928579086,"score_gpt":0.46580727643543135,"score_spread":0.40688579450685225,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4383198701","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96600354,0.0019084447,0.0002539657,0.023841282,0.0023079666,0.00038128113,0.0005641273,0.000097667165,0.0046417527],"genre_scores_gemma":[0.98973805,0.0012610324,0.0015294905,0.0009702498,0.00025408377,0.00042450574,0.0002337687,0.000027232341,0.0055616032],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9967667,0.00043569878,0.0005172443,0.000716228,0.00073108554,0.00083305925],"domain_scores_gemma":[0.99883646,0.00015777664,0.0000789048,0.00043137302,0.0001945574,0.000300909],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0017092387,0.0001629653,0.00075178547,0.015785273,0.00006891171,0.000019503048,0.00030794472,0.00032747304,0.0028999972],"category_scores_gemma":[0.00012884341,0.00016435132,0.00014120068,0.09022275,0.00024282101,0.00007856133,0.00013864618,0.0005114892,0.00018404858],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005158963,0.00015435297,0.58164924,0.0000040414016,0.00028114842,0.00018983752,0.00048604194,0.0013476053,0.00016415614,0.000012612027,0.32780588,0.08738916],"study_design_scores_gemma":[0.0040101125,0.00023180449,0.95048875,0.000049531376,0.00009842016,0.000009244893,0.0006444257,0.0041414853,0.000029565183,0.0007278523,0.039413266,0.00015556214],"about_ca_topic_score_codex":0.011920744,"about_ca_topic_score_gemma":0.00991522,"teacher_disagreement_score":0.36883947,"about_ca_system_score_codex":0.005741473,"about_ca_system_score_gemma":0.0005291233,"threshold_uncertainty_score":0.9980753},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4383227206","doi":"10.1163/15685306-bja10136","title":"Publications Trends in Society &amp; Animals from 2009–2019: A Bibliometric Analysis","year":2023,"lang":"en","type":"article","venue":"Society and Animals","topic":"Human-Animal Interaction Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia, Okanagan Campus; University of British Columbia","funders":"","keywords":"Publishing; Context (archaeology); Portrait; Library science; Bibliometrics; Animal welfare; Sociology; Social science; Political science; Media studies; History; Law; Art history; Biology; Ecology; Archaeology","score_opus":0.04883117263400211,"score_gpt":0.3828756642950114,"score_spread":0.3340444916610093,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4383227206","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9928999,0.0041017006,0.0001674695,0.0011842301,0.000039783696,0.00007289192,0.00037692665,0.000069770496,0.0010873147],"genre_scores_gemma":[0.98781526,0.0047913785,0.000987287,0.00036703693,0.00017279849,0.000045241406,0.0005518662,0.00001915758,0.0052499627],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9985512,0.00004834996,0.00031486776,0.00055658387,0.00018932033,0.0003396805],"domain_scores_gemma":[0.9993047,0.00007208048,0.000117963566,0.00027743535,0.00013565515,0.000092171445],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00040981956,0.00018571188,0.00029769985,0.0037338336,0.00021760723,0.00012553671,0.00016750375,0.00019634221,0.00033306744],"category_scores_gemma":[0.00009496178,0.00018243087,0.00056359015,0.051673256,0.00012288,0.000015501364,0.00023036731,0.00014650318,0.000069677575],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000030479134,0.000111173984,0.15151407,0.0000136730005,0.0021743102,8.1097636e-7,0.0013224047,0.000031998075,0.39140213,0.00013084507,0.45056203,0.002706064],"study_design_scores_gemma":[0.00036090304,0.000073291194,0.83367854,0.0000052379924,0.00017332537,0.000001024508,0.0012435194,0.00024466807,0.0017713073,0.0000880133,0.16210863,0.00025156513],"about_ca_topic_score_codex":0.00021246962,"about_ca_topic_score_gemma":0.000402408,"teacher_disagreement_score":0.68216443,"about_ca_system_score_codex":0.00002283804,"about_ca_system_score_gemma":0.00002556538,"threshold_uncertainty_score":0.96848387},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4383681840","doi":"10.15379/ijmst.v10i3.1499","title":"Analysis of Continuity of Care Research Developments Based on the Scopus Database 20 20-2023: A Bibliometric Study","year":2023,"lang":"en","type":"article","venue":"International Journal of Membrane Science and Technology","topic":"Health Systems, Economic Evaluations, Quality of Life","field":"Economics, Econometrics and Finance","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"University of Toronto; Monash University; University of Technology Sydney; University of Alberta; McMaster University; University of Warwick; McGill University","keywords":"Scopus; Health care; Continuity of care; Quality (philosophy); Field (mathematics); MEDLINE; Political science","score_opus":0.35448846184358535,"score_gpt":0.5072129736584133,"score_spread":0.152724511814828,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4383681840","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98321146,0.00033805674,0.000085387765,0.015009229,0.000329056,0.0002574675,0.00021752022,0.000008046458,0.00054378906],"genre_scores_gemma":[0.9992746,0.00013361376,0.00024753387,0.00025012423,0.000029130286,0.00001169391,0.0000068662985,0.0000046818277,0.000041751715],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99640745,0.00017880372,0.0020494529,0.00029107413,0.0008372997,0.0002358997],"domain_scores_gemma":[0.994189,0.0015474268,0.0017214196,0.00033972456,0.002133207,0.00006917374],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.038137913,0.00008158962,0.0005959415,0.07065489,0.00015177093,0.000055922923,0.0013035084,0.00006355444,0.00008640845],"category_scores_gemma":[0.014828125,0.000070593334,0.00006829498,0.050156355,0.0005878461,0.0002936812,0.00021669803,0.00026001726,0.000032103362],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011702542,0.0004864829,0.9666356,0.00011913418,0.00089415704,0.000022980077,0.0021926658,0.0011239778,0.0014625046,0.021697814,0.0018749431,0.0033727097],"study_design_scores_gemma":[0.00471768,0.002001934,0.8948637,0.00060900126,0.0001580275,0.000024407133,0.042128883,0.03617355,0.0060008713,0.0049023023,0.007936973,0.0004826364],"about_ca_topic_score_codex":0.00017776368,"about_ca_topic_score_gemma":0.00015656633,"teacher_disagreement_score":0.07177187,"about_ca_system_score_codex":0.00025378316,"about_ca_system_score_gemma":0.0006232128,"threshold_uncertainty_score":0.9934704},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4384297535","doi":"10.30604/jika.v8i3.2096","title":"Cost-effectiveness of Breast Cancer Screening from 2013-2022: A Bibliometric Study","year":2023,"lang":"en","type":"article","venue":"Jurnal Aisyah Jurnal Ilmu Kesehatan","topic":"Public Health and Nutrition","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Breast cancer; Medicine; Scopus; Scope (computer science); Family medicine; Breast cancer screening; Cancer; Modalities; MEDLINE; Mammography; Internal medicine; Computer science","score_opus":0.09685671424301294,"score_gpt":0.4055742190182141,"score_spread":0.30871750477520116,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4384297535","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99223304,0.0017058614,0.00031487076,0.002256172,0.00062487193,0.0016101876,0.00089531904,0.00016610128,0.00019356076],"genre_scores_gemma":[0.99664503,0.001664516,0.00011303563,0.00012461684,0.0008855424,0.00023836597,0.00018181666,0.000061554034,0.000085527354],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9959681,0.00045145032,0.0008555214,0.00053364487,0.0014111437,0.0007801489],"domain_scores_gemma":[0.99680513,0.0007154352,0.00038318877,0.00044085903,0.0008081744,0.0008472168],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0023799986,0.00032171461,0.00087796216,0.018752407,0.00022975523,0.000082196726,0.00031996172,0.00018203599,0.00038598056],"category_scores_gemma":[0.00035076516,0.00027358055,0.00024349666,0.04620677,0.00009518092,0.0005336491,0.00014895524,0.00080565194,0.000072909286],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025209205,0.0010407617,0.8866398,0.00035137284,0.00032544677,0.00037102975,0.00014716046,0.000021731117,0.005856167,0.0000110306055,0.0024205062,0.10029412],"study_design_scores_gemma":[0.0063313874,0.00089080544,0.9870078,0.0010747092,0.00021201902,0.00021612429,0.00071929046,0.00064846635,0.0007014648,0.000061880804,0.0018949934,0.00024103415],"about_ca_topic_score_codex":0.0053920825,"about_ca_topic_score_gemma":0.00014176786,"teacher_disagreement_score":0.10036807,"about_ca_system_score_codex":0.00029366394,"about_ca_system_score_gemma":0.0005450229,"threshold_uncertainty_score":0.9999716},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4384523817","doi":"10.1111/jan.15793","title":"Surveillance and patient safety in nursing research: A bibliometric analysis from 1993 to 2023","year":2023,"lang":"en","type":"review","venue":"Journal of Advanced Nursing","topic":"Patient Safety and Medication Errors","field":"Health Professions","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Trois-Rivières","funders":"","keywords":"Thematic analysis; Nursing; Patient safety; Medicine; Scopus; MEDLINE; Safety culture; Health care; Qualitative research","score_opus":0.26104582030097156,"score_gpt":0.5773274960499174,"score_spread":0.31628167574894583,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4384523817","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018940705,0.9914215,0.0011824812,0.0012302696,0.002222913,0.0011831647,0.0001660967,0.000024108149,0.00067539746],"genre_scores_gemma":[0.0011030447,0.9963205,0.0018671053,0.00006682848,0.0003463685,0.000039728096,0.00007257183,0.00005976956,0.00012412303],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99211663,0.002558121,0.0027412085,0.00048259884,0.0012702581,0.0008312024],"domain_scores_gemma":[0.9906724,0.0054234206,0.0023056848,0.0004196998,0.000614059,0.0005647103],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0038864939,0.00031656452,0.0024168114,0.1072126,0.0004787282,0.000007756639,0.000385453,0.00038723895,0.000044905602],"category_scores_gemma":[0.004454481,0.0002645426,0.0003793477,0.18048963,0.00013345937,0.00023882493,0.00008193205,0.0026078376,0.00007250469],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015109275,0.000106397645,0.00038408677,0.00097305607,0.00018006923,0.000029252335,0.0034110555,0.00039343478,0.0000010792621,0.000015316944,0.0015077688,0.9928474],"study_design_scores_gemma":[0.0008276196,0.00031222915,0.026917724,0.19944021,0.00066693214,0.000008389557,0.0073805,0.00003098362,1.9741222e-7,0.00083924324,0.76320755,0.00036843345],"about_ca_topic_score_codex":0.00005363418,"about_ca_topic_score_gemma":0.000079581885,"teacher_disagreement_score":0.99247897,"about_ca_system_score_codex":0.002192258,"about_ca_system_score_gemma":0.0008061603,"threshold_uncertainty_score":0.9999807},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4384698425","doi":"10.1080/08941920.2023.2230453","title":"<i>Seeing the Forest for the Trees</i> Sequel I: An Extension of the 1985–2017 Bibliometric Analysis of Environmental and Resource Sociology","year":2023,"lang":"en","type":"article","venue":"Society & Natural Resources","topic":"Forest Management and Policy","field":"Environmental Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Sociology; Resource (disambiguation); Natural resource; Environmental sociology; Social science; Extension (predicate logic); Bibliometrics; Environmental studies; Natural resource management; Regional science; Library science; Political science; Computer science; Law","score_opus":0.018917833290644198,"score_gpt":0.2648230821774819,"score_spread":0.2459052488868377,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4384698425","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99617666,0.0014491688,0.000003504503,0.0015665581,0.00005424341,0.00035672128,0.00004943371,0.000020578003,0.00032311384],"genre_scores_gemma":[0.9968929,0.0005235835,0.000041686566,0.0005824334,0.00005371819,0.000016929922,0.000023526674,0.000012976359,0.0018521874],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9985951,0.00011739766,0.00025072516,0.00029043073,0.00043084318,0.00031547254],"domain_scores_gemma":[0.99828887,0.00089450576,0.00025498954,0.0005157834,0.000006286733,0.00003955178],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009595468,0.0001517371,0.00022354495,0.00079851534,0.0006646746,0.0000307416,0.00071772427,0.000096015654,0.00007571114],"category_scores_gemma":[0.00009488099,0.00007186827,0.00046841364,0.013378193,0.0016559282,0.00010958882,0.0006417148,0.00019812136,0.000008978526],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014079719,0.000121062156,0.77247137,0.00006601005,0.0016657003,0.0000012630519,0.03435086,0.020692447,0.040409196,0.000554538,0.08625503,0.0432717],"study_design_scores_gemma":[0.0002038134,0.000057297326,0.9244575,0.000004845556,0.00044414276,7.2042354e-7,0.0021303007,0.029935518,0.00009862663,0.000211317,0.042362325,0.0000935834],"about_ca_topic_score_codex":0.00039823347,"about_ca_topic_score_gemma":0.00022187257,"teacher_disagreement_score":0.15198612,"about_ca_system_score_codex":0.00004138461,"about_ca_system_score_gemma":0.000003815632,"threshold_uncertainty_score":0.6427778},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4385211235","doi":"10.25110/arqsaude.v27i7.2023-028","title":"TRANSTORNOS DE ANSIEDADE EM IDOSOS: UM ESTUDO BIBLIOMÉTRICO ANXIETY DISORDERS IN THE ELDERLY PEOPLE: A BIBLIOMETRIC STUDY","year":2023,"lang":"pt","type":"article","venue":"Arquivos de Ciências da Saúde da UNIPAR","topic":"Healthcare during COVID-19 Pandemic","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Trois-Rivières","funders":"","keywords":"Anxiety; Psychology; Scopus; Humanities; Psychiatry; Political science; MEDLINE; Philosophy","score_opus":0.04541313456108754,"score_gpt":0.33071608740795905,"score_spread":0.2853029528468715,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385211235","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9748086,0.0023236114,0.008397795,0.0075271185,0.0012895459,0.0042339433,0.000121119556,0.0010430382,0.00025523332],"genre_scores_gemma":[0.99023163,0.004916009,0.00043684998,0.0029507622,0.00038686156,0.0006856184,0.000045098182,0.0001971063,0.00015006089],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9829452,0.004558559,0.0021781626,0.0027503092,0.0031364106,0.004431369],"domain_scores_gemma":[0.99013776,0.0038656318,0.00064323837,0.003856164,0.00036621213,0.001130994],"candidate_categories":["metaepi_narrow","bibliometrics","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaepi_narrow","bibliometrics"],"category_scores_codex":[0.012100407,0.0013596795,0.0014762757,0.10610882,0.0012144045,0.0022106466,0.0076360027,0.00085283804,0.00005262395],"category_scores_gemma":[0.0025660412,0.0012943222,0.00056128437,0.47629073,0.0002813016,0.0014937164,0.0010870028,0.002893084,0.0005094182],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000107995234,0.0020800584,0.7948273,0.0005004826,0.00021676865,0.0005756371,0.0960091,0.0018072492,0.00013637684,0.000318903,0.0076977815,0.09572239],"study_design_scores_gemma":[0.0040776427,0.002539286,0.93362164,0.00026862082,0.00021230738,0.00017329642,0.017100392,0.02539832,0.00003099655,0.00062116596,0.014424732,0.0015316041],"about_ca_topic_score_codex":0.006934451,"about_ca_topic_score_gemma":0.019694982,"teacher_disagreement_score":0.37018192,"about_ca_system_score_codex":0.001086576,"about_ca_system_score_gemma":0.0032726617,"threshold_uncertainty_score":0.9999154},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4385251539","doi":"10.3138/jsp-2022-0070","title":"Combining Bibliometric and Social Network Analysis to Understand the Scholarly Publications on Artificial Intelligence","year":2023,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Social network analysis; Rank correlation; Rank (graph theory); Thematic analysis; Spearman's rank correlation coefficient; Bibliometrics; Network analysis; Test (biology); Data science; Computer science; Social science; Sociology; Library science; Engineering; Qualitative research; Mathematics; Social capital; Ecology; Machine learning","score_opus":0.6468525460429406,"score_gpt":0.5326740116500904,"score_spread":0.1141785343928502,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385251539","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7972415,0.0007744983,0.0506745,0.14675088,0.0011665904,0.00026462405,0.00003107337,0.000051669424,0.0030446372],"genre_scores_gemma":[0.99621034,0.00017201006,0.0011445346,0.0013875064,0.00064732233,0.0000033966553,0.0000039613874,0.000016203569,0.00041472263],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.97983557,0.00081457064,0.001793589,0.000640048,0.016027264,0.0008889344],"domain_scores_gemma":[0.9672282,0.013587741,0.0012383839,0.0007119904,0.01640943,0.0008243053],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.13056369,0.00019431885,0.0005402768,0.5533087,0.0020369315,0.34038255,0.0055726864,0.0002424097,0.0001273391],"category_scores_gemma":[0.22188647,0.000121579586,0.0003838136,0.8940364,0.0001743363,0.061048556,0.0013625985,0.002693906,0.00012324267],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010288674,0.0001715949,0.16278838,0.0000042466427,0.0006247878,0.000031313575,0.0025789388,0.005865434,0.00018448534,0.08837257,0.16376549,0.57550985],"study_design_scores_gemma":[0.00021032858,0.00037674603,0.8159425,0.000018356935,0.00014142266,0.000020551504,0.0136411395,0.004943859,0.0000391777,0.13478005,0.029610388,0.00027546176],"about_ca_topic_score_codex":0.000042305506,"about_ca_topic_score_gemma":0.000029566114,"teacher_disagreement_score":0.65315413,"about_ca_system_score_codex":0.00026053376,"about_ca_system_score_gemma":0.00040743672,"threshold_uncertainty_score":0.99980766},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4385283054","doi":"10.11591/ijere.v12i3.22922","title":"Bibliometric analysis of leadership and physical education based on Scopus data","year":2023,"lang":"en","type":"article","venue":"International Journal of Evaluation and Research in Education (IJERE)","topic":"Sports and Physical Education Research","field":"Health Professions","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Universiti Teknologi MARA; Universitas Islam Riau; Universiti Utara Malaysia","keywords":"Scopus; Transformational leadership; Curriculum; Physical education; Publication; Educational leadership; Psychology; Medical education; Political science; Pedagogy; MEDLINE; Medicine; Social psychology","score_opus":0.6129480723669827,"score_gpt":0.6894746453085682,"score_spread":0.07652657294158549,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385283054","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97643614,0.00044865336,0.000024382145,0.019056333,0.0013062849,0.0004982538,0.00002898191,0.0000067153624,0.0021942772],"genre_scores_gemma":[0.99725944,0.00051275623,0.00011480969,0.0004814217,0.00067895517,0.000061206585,0.00031562347,0.000011626271,0.000564141],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9946893,0.0013070154,0.0007141862,0.00029916497,0.0027141625,0.00027618057],"domain_scores_gemma":[0.99109566,0.0028873014,0.00042815603,0.00036561678,0.005005475,0.00021777186],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.011382529,0.00009935195,0.0002774913,0.0682666,0.00016796315,0.000063715466,0.00047109157,0.00008974922,0.00055827125],"category_scores_gemma":[0.004622165,0.000082337254,0.00006150462,0.046922784,0.0001704862,0.00036134495,0.00013342965,0.00073552923,0.000030560248],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033948882,0.0023547306,0.26209825,0.00018307113,0.0003198953,0.0000020930777,0.005138623,0.0010941637,0.00039208963,0.0029086026,0.031601556,0.6935674],"study_design_scores_gemma":[0.00070220465,0.0001271716,0.9097538,0.0003959824,0.00009202216,0.0000012011568,0.0145980315,0.066487625,0.000027745264,0.0022448723,0.005489382,0.000079944795],"about_ca_topic_score_codex":0.00039901436,"about_ca_topic_score_gemma":0.0000600293,"teacher_disagreement_score":0.69348747,"about_ca_system_score_codex":0.0002984972,"about_ca_system_score_gemma":0.0061601163,"threshold_uncertainty_score":0.99947405},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4385318831","doi":"10.18280/ijdne.180308","title":"Trend in Publications Related to Biomethane Using a Bibliometric Approach","year":2023,"lang":"en","type":"article","venue":"International Journal of Design & Nature and Ecodynamics","topic":"Atmospheric and Environmental Gas Dynamics","field":"Environmental Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"European Regional Development Fund; Universidad de Extremadura; Junta de Extremadura; European Commission","keywords":"Engineering; Environmental science","score_opus":0.020441581795556767,"score_gpt":0.2759646970016098,"score_spread":0.25552311520605303,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385318831","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9463943,0.00010247454,0.051749423,0.0006652777,0.00030204697,0.00011383768,0.000007026401,0.00001785991,0.0006477356],"genre_scores_gemma":[0.9173608,0.00033844195,0.08174842,0.00017110608,0.000030801377,0.0000028387685,0.000014110675,0.000016987464,0.00031653186],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987122,0.00004492765,0.00038743662,0.00018450331,0.00047352377,0.00019741997],"domain_scores_gemma":[0.9994835,0.000089153255,0.00018800884,0.00009271393,0.000014668678,0.0001319306],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00062187517,0.00011819465,0.00015579973,0.0031640392,0.00004030903,0.000060180304,0.00038874312,0.00016162785,0.00006829409],"category_scores_gemma":[0.00009642246,0.000106142485,0.000057614736,0.0154862935,0.00006443812,0.000364871,0.00017678016,0.00035307888,0.000016123953],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000930682,0.0002447647,0.08483087,0.0000043155246,0.00011381808,0.000084402396,0.00054559956,0.8734125,0.004810063,0.00038662087,0.0009808266,0.03449313],"study_design_scores_gemma":[0.0009883867,0.00015664325,0.32458034,0.000032879856,0.00003193182,0.0005816908,0.00043921024,0.66725636,0.00003462224,0.0027534515,0.002843297,0.00030121894],"about_ca_topic_score_codex":0.00002685805,"about_ca_topic_score_gemma":0.000014968067,"teacher_disagreement_score":0.23974945,"about_ca_system_score_codex":0.00032861577,"about_ca_system_score_gemma":0.000015748332,"threshold_uncertainty_score":0.74406505},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4385390253","doi":"10.37908/mkutbd.1234256","title":"A bibliometric analysis of global publications on flax (Linum usitatissimum L.) disease during 2001-2021","year":2023,"lang":"en","type":"article","venue":"Mustafa Kemal Üniversitesi Tarım Bilimleri Dergisi","topic":"Plant Pathogens and Resistance","field":"Agricultural and Biological Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Linum; Scopus; Bibliometrics; Web of science; Library science; MEDLINE; Computer science; Political science; Biology; Botany","score_opus":0.02446473312684098,"score_gpt":0.2386314290471773,"score_spread":0.21416669592033633,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385390253","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9933927,0.00019398057,0.000008590984,0.0013776142,0.000090269976,0.00020653024,0.0030506342,0.0001373489,0.001542342],"genre_scores_gemma":[0.99622226,0.00027141848,0.00007618539,0.000057099915,0.00007944456,0.000008385756,0.0007242289,0.0000018417487,0.002559142],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99783534,0.00010478471,0.0003260338,0.00060600106,0.0005633097,0.0005645614],"domain_scores_gemma":[0.998595,0.00037056435,0.00020332883,0.00020744532,0.00022233176,0.00040130955],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00023232213,0.00023089613,0.00038244211,0.005968833,0.00038729372,0.000086937354,0.00046858116,0.00011080179,0.001619903],"category_scores_gemma":[0.0002497767,0.00011686934,0.00038781206,0.19367008,0.00009570074,0.00024900818,0.00017862611,0.000101303376,0.00013096076],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00055181334,0.00073110266,0.76113605,0.00008924374,0.0010601936,0.0004619636,0.000080073194,0.0068340222,0.20475784,0.00096684584,0.014370023,0.0089608105],"study_design_scores_gemma":[0.00018261574,0.00008956423,0.9915811,0.000022017039,0.00032398527,0.0000012671899,0.000342886,0.001748415,0.0001552146,0.000038898506,0.0052649495,0.00024907704],"about_ca_topic_score_codex":0.00017948101,"about_ca_topic_score_gemma":0.000852672,"teacher_disagreement_score":0.23044504,"about_ca_system_score_codex":0.00014731406,"about_ca_system_score_gemma":0.000047373018,"threshold_uncertainty_score":0.99929273},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4385812956","doi":"10.3390/jrfm16080373","title":"Integration of Digital Technologies in Corporate Social Responsibility (CSR) Activities: A Systematic Literature Review and Bibliometric Analysis","year":2023,"lang":"en","type":"article","venue":"Journal of risk and financial management","topic":"Innovation, Sustainability, Human-Machine Systems","field":"Social Sciences","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Corporate social responsibility; Digital transformation; Context (archaeology); Knowledge management; Scientific literature; Data science; Digitization; Computer science; Business; Political science; World Wide Web; Public relations","score_opus":0.021571752571746382,"score_gpt":0.31379831068861014,"score_spread":0.29222655811686377,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385812956","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98078966,0.015203922,0.002006308,0.00059563085,0.00012073962,0.0009738345,0.000031874188,0.000039248676,0.00023878374],"genre_scores_gemma":[0.9774829,0.02227561,0.00008336631,0.0000075339562,0.000035815956,0.000015645415,0.0000037218008,0.0000039313354,0.00009147875],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99799365,0.00040084956,0.00085178146,0.00015880243,0.00043203172,0.00016287455],"domain_scores_gemma":[0.9979528,0.00025398954,0.0012009016,0.00012629796,0.00044414177,0.00002182419],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0060806065,0.00011028316,0.00065015437,0.025062675,0.00020846563,0.00016493158,0.00016031603,0.00009403873,0.0000013521852],"category_scores_gemma":[0.0038968548,0.00008684018,0.000118302014,0.0897167,0.0001942094,0.00054290716,0.00008688151,0.0002161993,2.638266e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00039836083,0.00048313374,0.50979805,0.14054,0.0007451847,0.00023679508,0.07172002,0.000054131866,0.000020291922,0.12031846,0.0013749204,0.15431066],"study_design_scores_gemma":[0.00071498996,0.00024357876,0.85855585,0.010300488,0.0015536471,0.000005210968,0.06348175,0.00009205084,0.000004708738,0.06393273,0.0007973102,0.00031767882],"about_ca_topic_score_codex":0.00005876528,"about_ca_topic_score_gemma":0.0003009744,"teacher_disagreement_score":0.34875783,"about_ca_system_score_codex":0.00014718407,"about_ca_system_score_gemma":0.00006788363,"threshold_uncertainty_score":0.9859874},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4385851222","doi":"10.2196/46620","title":"Altmetric Analysis of Dermatology Manuscript Dissemination During the COVID-19 Era: Cross-Sectional Study","year":2023,"lang":"en","type":"article","venue":"JMIR Dermatology","topic":"Social Media in Health Education","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Altmetrics; Impact factor; Coronavirus disease 2019 (COVID-19); Citation; MEDLINE; Logistic regression; Social media; Medicine; Library science; Computer science; Pathology; Political science; World Wide Web; Internal medicine","score_opus":0.10609429027645394,"score_gpt":0.47842853237698724,"score_spread":0.3723342421005333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385851222","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9855166,0.00008291918,0.000030735966,0.009912591,0.002429378,0.0015361604,0.000013989873,0.00017828669,0.00029930795],"genre_scores_gemma":[0.99764085,0.000057708643,0.000014947499,0.00017407011,0.00017561352,0.001733721,0.000046395788,0.000011447903,0.00014525784],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99675876,0.0012052503,0.00065357046,0.00033811518,0.00055697054,0.00048736198],"domain_scores_gemma":[0.9935186,0.0053867307,0.0003723048,0.00028955578,0.00024170276,0.00019113369],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00080238143,0.00012204479,0.00038775618,0.0015633181,0.0012945578,0.000050485516,0.00042949454,0.00025229322,0.00025501652],"category_scores_gemma":[0.0059118564,0.00011202162,0.00013761868,0.007988392,0.000568024,0.00015401504,0.00007591849,0.00024252661,0.00007962859],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000018222088,0.000107691514,0.96313107,0.000057897516,0.00017146718,0.000028954373,0.03325548,0.000044242915,0.000018580324,0.0020468836,0.0009888492,0.00013067194],"study_design_scores_gemma":[0.00023623684,0.000011338076,0.9853221,0.0000013252627,0.0001015132,0.000033959204,0.009625894,0.00015991443,0.000027160084,0.00032365415,0.004063661,0.00009322452],"about_ca_topic_score_codex":0.0040523624,"about_ca_topic_score_gemma":0.006166928,"teacher_disagreement_score":0.023629585,"about_ca_system_score_codex":0.0004884691,"about_ca_system_score_gemma":0.0005629665,"threshold_uncertainty_score":0.9956823},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4385994600","doi":"10.3389/fsuep.2023.1207675","title":"Brown-out of policy ideas? A bibliometric review and computational text analysis of research on energy access","year":2023,"lang":"en","type":"review","venue":"Frontiers in Sustainable Energy Policy","topic":"Energy and Environment Impacts","field":"Environmental Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Public policy; Policy analysis; Multidisciplinary approach; Policy studies; Energy policy; Field (mathematics); Political science; Energy (signal processing); Process (computing); Public relations; Computer science; Management science; Data science; Sociology; Public administration; Social science; Economics; Engineering","score_opus":0.06981091892582034,"score_gpt":0.41055087509800514,"score_spread":0.3407399561721848,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385994600","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000030467914,0.9870101,0.0006848864,0.00022492342,0.00006712614,0.00031721903,0.000094982,0.000028532808,0.011541752],"genre_scores_gemma":[0.0011581789,0.9799236,0.00017701401,0.00032078841,0.00010175668,0.00017170809,0.00021971508,0.00008508176,0.01784216],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9943948,0.0008430322,0.0013183831,0.0008882176,0.001336234,0.0012193249],"domain_scores_gemma":[0.9971539,0.00081452506,0.0007684016,0.0008485828,0.000067235014,0.00034735736],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0018330731,0.0005670106,0.0028806443,0.07757657,0.00015982085,0.00006619149,0.0012783491,0.00043683735,0.00020476594],"category_scores_gemma":[0.0013745074,0.00050763984,0.00052656996,0.1770218,0.0008887038,0.00039370198,0.0014847004,0.0004173116,0.0000106532825],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003995717,0.00047799255,0.0009188669,0.031385683,0.0024438987,0.00010420598,0.00013611348,0.13738577,1.8869538e-7,0.057673838,0.054226752,0.71520674],"study_design_scores_gemma":[0.00026605162,0.00015151287,0.0020199742,0.004296081,0.0014416041,0.000003453647,0.00013029056,0.00027107025,0.0000019442114,0.0092853205,0.98158467,0.0005480187],"about_ca_topic_score_codex":0.029039053,"about_ca_topic_score_gemma":0.00020396383,"teacher_disagreement_score":0.9273579,"about_ca_system_score_codex":0.0020607284,"about_ca_system_score_gemma":0.0006256903,"threshold_uncertainty_score":0.9997375},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4386019850","doi":"10.2196/44766","title":"Virtual Reality Technology in Nursing Professional Skills Training: Bibliometric Analysis","year":2023,"lang":"en","type":"article","venue":"JMIR Serious Games","topic":"Virtual Reality Applications and Impacts","field":"Computer Science","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"","funders":"Natural Science Foundation of Zhejiang Province","keywords":"Virtual reality; Professional development; Medical education; Descriptive statistics; Nurse education; Psychology; Nursing; Medicine; Computer science","score_opus":0.029437779891871328,"score_gpt":0.3675708841509119,"score_spread":0.33813310425904053,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386019850","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9613624,0.00007171345,0.025726318,0.01027756,0.00019766901,0.00040889706,0.0000152460325,0.00073002005,0.001210172],"genre_scores_gemma":[0.9979741,0.00006554696,0.001029539,0.000118700584,0.000030992876,0.00016452713,0.000022824672,0.000009986521,0.0005837945],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99807376,0.00008824011,0.0003731308,0.0005248556,0.0003994099,0.000540579],"domain_scores_gemma":[0.9987331,0.00016343055,0.00013026912,0.0007551367,0.00007930517,0.00013874595],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00060722354,0.00015939889,0.00034083409,0.06742227,0.00013365642,0.00013518435,0.0008669892,0.00017838708,0.000012755507],"category_scores_gemma":[0.00020604109,0.00014615538,0.00010602966,0.3294886,0.000121430545,0.00036831203,0.00020799386,0.000242366,0.000089650595],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000050284566,0.00029393221,0.0036759388,0.0000055709847,0.000049672406,0.000015345135,0.0028129346,0.00055260013,0.0006602814,0.037591524,0.0023846508,0.9519525],"study_design_scores_gemma":[0.00037274018,0.00015258662,0.9423679,0.00006322926,0.00002070994,0.000011701475,0.0008835516,0.030823743,0.00034781097,0.02072093,0.003953382,0.00028170747],"about_ca_topic_score_codex":0.000035040764,"about_ca_topic_score_gemma":0.00003545717,"teacher_disagreement_score":0.9516708,"about_ca_system_score_codex":0.000113733666,"about_ca_system_score_gemma":0.00016693924,"threshold_uncertainty_score":0.9431477},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4386084175","doi":"10.1007/s11356-023-29243-9","title":"Thirty years of research on photoacoustic imaging in the field of cancer: A scientometric analysis of hotspots, bursts, and research trends","year":2023,"lang":"en","type":"article","venue":"Environmental Science and Pollution Research","topic":"Photoacoustic and Ultrasonic Imaging","field":"Engineering","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Impact","funders":"","keywords":"Photoacoustic imaging in biomedicine; Citation; Field (mathematics); China; Science Citation Index; Web of science; Cancer; Library science; Medical physics; Political science; Medicine; Computer science; MEDLINE; Physics; Internal medicine","score_opus":0.045996874955005776,"score_gpt":0.4128751438158682,"score_spread":0.36687826886086244,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386084175","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9975873,0.00083648483,0.000009837354,0.00030563504,0.000031508956,0.00015446029,0.00006187946,0.0000077857185,0.0010050576],"genre_scores_gemma":[0.9978552,0.0020014995,0.000012592722,0.000013051251,0.000010946789,0.000022140055,0.0000021163212,0.000005665925,0.000076772056],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.99584895,0.00024043831,0.00022433136,0.00029052963,0.002763726,0.0006320419],"domain_scores_gemma":[0.9981898,0.0013596029,0.000025060564,0.00027529773,0.00006301314,0.00008724456],"candidate_categories":["bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.014711187,0.00006779203,0.00016823506,0.00988833,0.00026094457,0.000042547825,0.00044356272,0.00003720027,0.000066744826],"category_scores_gemma":[0.0005965306,0.00005575919,0.0000312802,0.030574154,0.0027327733,0.00013654615,0.00020019477,0.0005001205,0.0000023725245],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000692495,0.00018481271,0.12194958,0.0001450899,0.0000640716,0.00002251352,0.0079215625,0.013146519,0.6836348,0.0003773122,0.0016853343,0.17079912],"study_design_scores_gemma":[0.00018584328,0.00010048203,0.88773566,0.000065183835,0.000015326672,0.0000010929284,0.006635308,0.09141787,0.0135763325,0.000067202745,0.00014011525,0.00005957109],"about_ca_topic_score_codex":0.0016211288,"about_ca_topic_score_gemma":0.000043688757,"teacher_disagreement_score":0.7657861,"about_ca_system_score_codex":0.00021832818,"about_ca_system_score_gemma":0.0000853622,"threshold_uncertainty_score":0.9999812},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4386211900","doi":"10.1101/2023.08.26.23294661","title":"Neuropsychiatric Disturbances in Mild Cognitive Impairment: A Scientometric Analysis","year":2023,"lang":"en","type":"preprint","venue":"medRxiv","topic":"Dementia and Cognitive Impairment Research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital; University of Ottawa","funders":"","keywords":"Dementia; Cognitive impairment; Psychology; Cognition; Citation; Delirium; Psychological intervention; Clinical psychology; Medicine; Psychiatry; Disease; Computer science; Library science; Pathology","score_opus":0.05000880192050656,"score_gpt":0.36860456159023275,"score_spread":0.3185957596697262,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386211900","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.991021,0.0016470148,0.0002776567,0.00096173433,0.00066385453,0.0017719175,0.00012786327,0.00018838627,0.003340568],"genre_scores_gemma":[0.99298584,0.0010089513,0.00008847888,0.00023003013,0.00021175518,0.00047550746,0.00040889878,0.000061947336,0.004528604],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99494535,0.0003198538,0.0007268512,0.001412972,0.0016831947,0.0009117759],"domain_scores_gemma":[0.9980298,0.00033421852,0.00024976052,0.0005979362,0.0004042918,0.00038398997],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0016588665,0.00046722838,0.0010252657,0.01349735,0.000108442655,0.00015934871,0.00043164266,0.00022240813,0.00042248884],"category_scores_gemma":[0.0006538457,0.00040436143,0.0007285517,0.025185687,0.00021409366,0.000076547934,0.00089494325,0.0012192996,0.0004138809],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027351466,0.0008497332,0.99480844,0.00046232878,0.0012150296,0.00047304414,0.00019028047,0.000023145536,0.000023207444,0.000006160849,0.0002760251,0.0013990706],"study_design_scores_gemma":[0.0020407403,0.00040881333,0.9926087,0.00038910532,0.0023525031,0.000004940033,0.0001827703,0.0012344974,0.00012539263,0.00021969096,0.000096113065,0.0003366961],"about_ca_topic_score_codex":0.00031793653,"about_ca_topic_score_gemma":0.00013775866,"teacher_disagreement_score":0.011688338,"about_ca_system_score_codex":0.00018724863,"about_ca_system_score_gemma":0.0003845609,"threshold_uncertainty_score":0.9998408},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4386402131","doi":"10.1007/s11192-023-04818-y","title":"The Matthew Effect in China’s social sciences and humanities research: a comparative analysis of CSSCI and SSCI","year":2023,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":7,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Planning Office of Philosophy and Social Science","keywords":"Elite; China; Humanities; Social science; Citation analysis; Political science; Sociology; Citation; Philosophy; Law; Politics","score_opus":0.8933759979167228,"score_gpt":0.6993538977340414,"score_spread":0.1940221001826814,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386402131","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98807365,0.0024008544,0.000027933145,0.0010264065,0.00023022701,0.00035622797,0.000050694627,0.000020692538,0.0078133205],"genre_scores_gemma":[0.9977904,0.0008165282,0.00006689679,0.000028725173,0.000025472054,0.000019708847,0.000002392713,0.0000048460092,0.0012450548],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.97931486,0.0014578373,0.0008258275,0.0010195652,0.016147384,0.0012345301],"domain_scores_gemma":[0.9631076,0.033782538,0.00029161482,0.0004913129,0.0020990123,0.00022795447],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.16835909,0.00017386013,0.0006612338,0.34124807,0.002451258,0.0050068367,0.0025928724,0.00009899321,0.000043391454],"category_scores_gemma":[0.054379962,0.000099471035,0.00014606814,0.80526614,0.005054522,0.00059656997,0.0021120864,0.00042851074,0.000044365595],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005090946,0.00010226754,0.87923145,0.000024762016,0.0001277269,0.000012168233,0.008029198,0.0001813181,0.00031745507,0.023858776,0.010844247,0.077219695],"study_design_scores_gemma":[0.00029308698,0.00031634653,0.95793754,0.0000064820692,0.000022880628,7.830575e-7,0.0045360043,0.025352975,0.00013026592,0.007529927,0.0037489533,0.00012477401],"about_ca_topic_score_codex":0.00046130814,"about_ca_topic_score_gemma":0.00046325274,"teacher_disagreement_score":0.46401805,"about_ca_system_score_codex":0.0001129867,"about_ca_system_score_gemma":0.00020266067,"threshold_uncertainty_score":0.9988474},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4386422741","doi":"10.1108/ijcma-02-2023-0025","title":"Mapping the knowledge base and theoretical evolution of workplace conflict outcomes: a bibliometric and qualitative review, 1972–2022","year":2023,"lang":"en","type":"article","venue":"International Journal of Conflict Management","topic":"Conflict Management and Negotiation","field":"Social Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Originality; Scopus; Citation; Negotiation; Industrial and organizational psychology; Qualitative research; Empirical research; Sociology; Bibliometrics; Psychology; Knowledge management; Management science; Social science; Social psychology; Computer science; Political science; Epistemology; Library science; Engineering","score_opus":0.07089405749771596,"score_gpt":0.40496547903535085,"score_spread":0.3340714215376349,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386422741","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5829583,0.075673535,0.021033736,0.16300598,0.00493292,0.003915687,0.000052627864,0.00020226892,0.14822498],"genre_scores_gemma":[0.94497955,0.051186435,0.00011692835,0.0005560223,0.00015114176,0.000017312295,0.0000042099,0.000009756607,0.0029786301],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975621,0.0004774479,0.00061386166,0.00015863993,0.0009847742,0.00020313746],"domain_scores_gemma":[0.99759585,0.0011050544,0.0005320914,0.00012293115,0.0005543116,0.00008975639],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.006101504,0.00012470792,0.00026456316,0.006364061,0.00016101863,0.00011583904,0.00046785857,0.000039638653,0.00018673613],"category_scores_gemma":[0.0008708708,0.000092716335,0.000113105234,0.00695078,0.00037794464,0.00027859487,0.00025792146,0.00014906347,0.000015573405],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000089886846,0.00010095293,0.005506318,0.00029954422,0.0008026846,0.000033799985,0.04872542,0.000008651763,0.000049108166,0.87810224,0.010619985,0.055661418],"study_design_scores_gemma":[0.0018463451,0.00013334081,0.14614576,0.0014888202,0.0002675655,0.000009223079,0.060581118,0.0006767466,0.000012891915,0.0072627594,0.78129673,0.00027867884],"about_ca_topic_score_codex":0.00006782509,"about_ca_topic_score_gemma":0.000023267019,"teacher_disagreement_score":0.8708395,"about_ca_system_score_codex":0.00013100158,"about_ca_system_score_gemma":0.000052308653,"threshold_uncertainty_score":0.5678585},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4386446501","doi":"10.20944/preprints202309.0239.v1","title":"Circular Economy and Solid Waste Management: Connections from a Bibliometric Analysis","year":2023,"lang":"en","type":"preprint","venue":"Preprints.org","topic":"Sustainable Supply Chain Management","field":"Business, Management and Accounting","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Fundação de Apoio à Pesquisa do Distrito Federal","keywords":"Circular economy; Reuse; Scopus; Consumption (sociology); China; Environmental economics; Business; Municipal solid waste; Point (geometry); Solid waste management; Cleaner production; Regional science; Waste management; Engineering; Economics; Geography; Political science; Sociology; Mathematics; Social science","score_opus":0.08414809923357626,"score_gpt":0.30564548428159344,"score_spread":0.2214973850480172,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386446501","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93810916,0.00042128746,0.004721406,0.0018171803,0.0012439872,0.0022037365,0.000045146797,0.0012372559,0.050200842],"genre_scores_gemma":[0.9913187,0.0004353066,0.00015553502,0.00086208846,0.0011656086,0.0008744427,0.0007052897,0.00015684536,0.0043261903],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99508345,0.00006445199,0.0009799285,0.0024228992,0.0005801407,0.00086911255],"domain_scores_gemma":[0.9961426,0.00017280114,0.0008287174,0.0024006653,0.0003803331,0.00007487563],"candidate_categories":["metaepi_narrow","bibliometrics","open_science","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0015516913,0.00076220767,0.0010621465,0.10369541,0.00041721622,0.00090662134,0.0013127703,0.00034794412,0.0021252693],"category_scores_gemma":[0.00032314006,0.00090593804,0.00062184746,0.073615626,0.00013544087,0.00085455074,0.011462685,0.00073096104,0.004068382],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003633114,0.00020846275,0.94365484,0.003076778,0.011355053,0.00031118534,0.00015754862,0.031171873,0.00002754253,0.00706748,0.0015839525,0.0013489367],"study_design_scores_gemma":[0.00096218276,0.0000036281106,0.80905175,0.00016080127,0.007069271,6.954088e-7,0.003120578,0.029415336,0.00006440132,0.071023025,0.07776207,0.0013662659],"about_ca_topic_score_codex":0.0063315774,"about_ca_topic_score_gemma":0.00015286605,"teacher_disagreement_score":0.13460311,"about_ca_system_score_codex":0.00029167402,"about_ca_system_score_gemma":0.00003759209,"threshold_uncertainty_score":0.9993391},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4386563176","doi":"10.1186/s12961-023-01046-x","title":"Characterizing Canadian funded partnered health research projects between 2011 and 2019: a retrospective analysis","year":2023,"lang":"en","type":"article","venue":"Health Research Policy and Systems","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Ottawa; University of British Columbia; International Collaboration On Repair Discoveries; Ottawa Hospital; University of Manitoba; University of British Columbia, Okanagan Campus; George & Fay Yee Centre for Healthcare Innovation","funders":"Canadian Institutes of Health Research; Canada Research Chairs","keywords":"Health services research; General partnership; Inclusion (mineral); Health care; Relevance (law); Descriptive statistics; Research design; Medicine; Health administration; Medical education; Public health; Political science; Nursing; Sociology; Social science","score_opus":0.9354305500257228,"score_gpt":0.7575997565129052,"score_spread":0.1778307935128176,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386563176","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.49455166,0.00084276375,0.000033430388,0.48309213,0.0004410454,0.011897996,0.0026672494,0.00031222543,0.0061614965],"genre_scores_gemma":[0.97713274,0.003119037,0.00003888246,0.007938101,0.001766157,0.0015212944,0.00017861681,0.000058845897,0.008246339],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9613483,0.025363108,0.0020328136,0.0012362885,0.002794204,0.0072253165],"domain_scores_gemma":[0.9818057,0.009091418,0.00055137224,0.000918728,0.001711709,0.005921077],"candidate_categories":["metaresearch","metaepi_narrow","sts","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.10706316,0.0002913873,0.0012306394,0.009033334,0.010594462,0.0002619894,0.0004952037,0.0003321855,0.00008899237],"category_scores_gemma":[0.01033441,0.00026886354,0.00006400247,0.011993256,0.00061932683,0.00039405611,0.00047364016,0.0027914343,0.0009136998],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000040729952,0.000018842173,0.55154204,0.003741105,0.00017937676,0.000015948579,0.101919144,0.0000012590916,0.00003079884,0.020275598,0.31799695,0.004238186],"study_design_scores_gemma":[0.00069757935,0.0004936885,0.7650903,0.0003841768,0.000006281965,0.0000030824124,0.021712027,0.0005620967,0.0000010643845,0.0005004107,0.21035403,0.00019526039],"about_ca_topic_score_codex":0.8925401,"about_ca_topic_score_gemma":0.44834232,"teacher_disagreement_score":0.48258108,"about_ca_system_score_codex":0.0054068333,"about_ca_system_score_gemma":0.031672936,"threshold_uncertainty_score":0.99997634},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"medium"},{"model":"gpt","categories":["metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W4386660338","doi":"10.1093/eurheartj/ehab724.3174","title":"Global representation of heart failure clinical trial leaders and collaborators: a systematic bibliometric review 2000–2020","year":2021,"lang":"en","type":"article","venue":"European Heart Journal","topic":"Health Systems, Economic Evaluations, Quality of Life","field":"Economics, Econometrics and Finance","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Population Health Research Institute; McMaster University","funders":"","keywords":"Medicine; CINAHL; Odds ratio; Clinical trial; MEDLINE; Odds; Logistic regression; Randomized controlled trial; Demography; Family medicine; Internal medicine; Psychological intervention; Nursing; Political science","score_opus":0.554892551983914,"score_gpt":0.5265177304932481,"score_spread":0.028374821490665836,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386660338","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5213897,0.3105408,0.0013378505,0.15645675,0.0039375094,0.0031673696,0.00033210136,0.000078063225,0.0027598704],"genre_scores_gemma":[0.841017,0.054865174,0.027343206,0.06845186,0.0068489425,0.000076763594,0.00007418418,0.0002046124,0.0011182592],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9850644,0.0063968264,0.0073786518,0.00054650975,0.00026221186,0.0003514202],"domain_scores_gemma":[0.99387676,0.0015898419,0.0031885523,0.000543036,0.000385081,0.0004167072],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.042424623,0.00019876924,0.0020386553,0.0018462434,0.00022210249,0.0002253545,0.0002176893,0.0000922655,0.00040058047],"category_scores_gemma":[0.0331594,0.00022397873,0.00034734,0.011449812,0.000119660675,0.0005149596,0.00008340426,0.0003809833,0.00073776365],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020241918,0.00023479,0.17566112,0.017867763,0.00025866693,0.000038541795,0.00037141246,0.000021358403,0.0000031542588,0.0009850704,0.80426437,0.000091319256],"study_design_scores_gemma":[0.051439773,0.002239629,0.44760633,0.044978242,0.0004169474,0.004202882,0.0091071315,0.0006115713,0.000011594216,0.0017132357,0.43588576,0.0017868984],"about_ca_topic_score_codex":0.000021522726,"about_ca_topic_score_gemma":0.0000133297135,"teacher_disagreement_score":0.3683786,"about_ca_system_score_codex":0.00024675584,"about_ca_system_score_gemma":0.0005551795,"threshold_uncertainty_score":0.98602533},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4386661486","doi":"10.1093/eurheartj/ehab724.3171","title":"Temporal trends in the inclusion of patient-reported outcomes in heart failure randomized trials published in high-impact medical journals: a systematic bibliometric review","year":2021,"lang":"en","type":"article","venue":"European Heart Journal","topic":"Health Systems, Economic Evaluations, Quality of Life","field":"Economics, Econometrics and Finance","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McMaster University; Impact","funders":"Canadian Institutes of Health Research","keywords":"Medicine; Randomized controlled trial; Odds ratio; Confidence interval; CINAHL; MEDLINE; Internal medicine; Clinical endpoint; Psychological intervention; Nursing","score_opus":0.37615181340314185,"score_gpt":0.49000686126012366,"score_spread":0.1138550478569818,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386661486","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.71253663,0.09420852,0.00011946385,0.18980248,0.00083920767,0.0018942838,0.00004671476,0.000017025497,0.0005356559],"genre_scores_gemma":[0.9833646,0.004170442,0.0008410671,0.011202791,0.00023986831,0.00005411572,0.00002800888,0.000047036654,0.000052111067],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9126401,0.05467475,0.029747505,0.00072461244,0.0013454062,0.00086764095],"domain_scores_gemma":[0.9783967,0.011007563,0.009177486,0.00077598885,0.00025905535,0.0003831865],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.37912217,0.00034405847,0.0074103437,0.02235838,0.00018699306,0.0003497719,0.0006529801,0.00015929766,0.0032577112],"category_scores_gemma":[0.22414815,0.00024583458,0.0011074479,0.020126184,0.000080105456,0.00093398005,0.000279562,0.0012545782,0.00010283684],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000809542,0.001294623,0.87089086,0.008998711,0.0004590917,0.00057394954,0.0046721003,0.0001913257,0.0000054209913,0.00056721224,0.111068204,0.0004689389],"study_design_scores_gemma":[0.100839384,0.00032737604,0.82849884,0.05325659,0.00013330056,0.0026721742,0.0018709094,0.0009098349,0.000003230287,0.0033670957,0.0072627743,0.00085848145],"about_ca_topic_score_codex":0.0005772989,"about_ca_topic_score_gemma":0.00028739325,"teacher_disagreement_score":0.27082792,"about_ca_system_score_codex":0.00054894964,"about_ca_system_score_gemma":0.0005826643,"threshold_uncertainty_score":0.9999994},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"systematic_review","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4386747798","doi":"10.56206/husbd.1353189","title":"45 Years of Authentic Leadership Research: A Bibliometric Analysis between 1978 and 2023","year":2023,"lang":"en","type":"article","venue":"Haliç Üniversitesi Sosyal Bilimler Dergisi","topic":"Organizational Leadership and Management Strategies","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Authentic leadership; Sociology; Transactional leadership; Educational leadership; Political science; Management; Library science; Pedagogy; Public relations","score_opus":0.17669288708987463,"score_gpt":0.3023773471243232,"score_spread":0.12568446003444855,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386747798","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9777868,0.0003327082,0.00022560733,0.003281188,0.00015445608,0.0003604359,0.000026647509,0.0002986627,0.017533539],"genre_scores_gemma":[0.9945751,0.00008047318,0.000058316313,0.00013674637,0.00029490402,0.0000035179846,0.00018110029,0.000024136623,0.004645735],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99769104,0.00005814327,0.00029370986,0.0005067361,0.00087933044,0.0005710619],"domain_scores_gemma":[0.998619,0.00050632795,0.00016873654,0.00032890163,0.0003318582,0.00004514121],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0011556839,0.00018797207,0.0003517571,0.073800266,0.0002578752,0.00032037418,0.0004284332,0.00012533813,0.0005286336],"category_scores_gemma":[0.0003731361,0.00020841195,0.00015954165,0.18693294,0.00029300942,0.0008664334,0.00045705945,0.00020499894,0.0009273175],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007240405,0.00015725895,0.9149703,0.0012915947,0.0021633406,0.00017200618,0.0009435519,0.0005678175,0.0006448102,0.039641995,0.03103071,0.008344215],"study_design_scores_gemma":[0.00061503664,0.000029204473,0.96731097,0.000047511156,0.001033748,4.7409634e-7,0.011741466,0.00080070185,0.000063019696,0.003019576,0.014991663,0.00034663288],"about_ca_topic_score_codex":0.00054363103,"about_ca_topic_score_gemma":0.000092201415,"teacher_disagreement_score":0.11313267,"about_ca_system_score_codex":0.000034739238,"about_ca_system_score_gemma":0.00003114776,"threshold_uncertainty_score":0.9998506},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4386818031","doi":"10.1097/js9.0000000000000720","title":"Global trends and prospects in health economics of robotic surgery: a bibliometric analysis","year":2023,"lang":"en","type":"article","venue":"International Journal of Surgery","topic":"Surgical Simulation and Training","field":"Medicine","cited_by":46,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"SKiN Health","funders":"Ministry of Industry and Information Technology of the People's Republic of China; Central South University","keywords":"Medicine; Bibliometrics; Health economics; General surgery; Medical physics; Library science; Public health; Pathology","score_opus":0.08465062413288361,"score_gpt":0.3681334662346186,"score_spread":0.283482842101735,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386818031","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9960752,0.00045211377,0.000075087366,0.0026771969,0.0004061824,0.000021138298,0.0000086324,0.0000092160535,0.00027522322],"genre_scores_gemma":[0.9983398,0.0012713672,0.00006225522,0.00015857331,0.00009857756,5.4017244e-7,0.000020091604,0.000005172641,0.000043628002],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9984906,0.000044006116,0.00088265113,0.00009980758,0.00034554634,0.00013739508],"domain_scores_gemma":[0.99800646,0.0010342667,0.0005434446,0.000054456304,0.00022481986,0.00013656696],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0014917096,0.00007124801,0.00059968355,0.05839427,0.0000108080285,0.000022674165,0.000053241554,0.00003611777,0.00010742542],"category_scores_gemma":[0.0005584911,0.00006221502,0.00031753493,0.047022276,0.000031282383,0.00012231953,0.000021349397,0.00007970782,0.0000017221129],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014874028,0.00007375485,0.88342696,0.000016656975,0.0005191428,0.00014598212,0.0000287924,0.0055990275,9.567543e-7,0.00012961285,0.00016324234,0.10974715],"study_design_scores_gemma":[0.00061044004,0.000022798195,0.9936929,0.0001147915,0.00004917175,0.00008241257,0.000033253353,0.0046356847,0.0000046951045,0.0001605849,0.0005492365,0.000044038316],"about_ca_topic_score_codex":0.000071763345,"about_ca_topic_score_gemma":0.000024661907,"teacher_disagreement_score":0.110265955,"about_ca_system_score_codex":0.0001392606,"about_ca_system_score_gemma":0.00022091078,"threshold_uncertainty_score":0.9732337},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4386846136","doi":"10.1080/0886022x.2023.2257804","title":"A bibliometric and knowledge-map analysis of antibody-mediated rejection in kidney transplantation","year":2023,"lang":"en","type":"article","venue":"Renal Failure","topic":"Renal Transplantation Outcomes and Treatments","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Natural Science Foundation of China","keywords":"Medicine; Transplantation; Kidney transplantation; Antibody; Kidney; Graft rejection; Intensive care medicine; Internal medicine; Immunology","score_opus":0.018660289224601206,"score_gpt":0.32727666424704194,"score_spread":0.3086163750224407,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386846136","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9981302,0.00024245807,0.00048614296,0.0003815279,0.000043899876,0.00021829373,0.00013391906,0.00005848181,0.0003050933],"genre_scores_gemma":[0.99534744,0.0021523964,0.0003170829,0.000031257212,0.000016113625,0.00001026936,0.001840447,0.000009946861,0.00027506234],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99910635,0.00004449586,0.00028945645,0.00020206354,0.0002078004,0.00014981536],"domain_scores_gemma":[0.99948186,0.00015300086,0.000068166344,0.000099724915,0.00007355888,0.00012366317],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00017065204,0.00011001934,0.0003485233,0.034118954,0.000027250508,0.000010449662,0.000028696932,0.00008748905,0.000051011077],"category_scores_gemma":[0.000049355254,0.00008878471,0.00011421864,0.07408796,0.000026087753,0.00008751781,0.000004821959,0.00008955718,0.000032005046],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038320362,0.0002208529,0.9260591,0.0015559442,0.0013520329,0.00024996648,0.0012598573,0.00017441432,0.06570221,0.000093988114,0.00015195452,0.0027964446],"study_design_scores_gemma":[0.0025183342,0.00013903757,0.9902108,0.0005929939,0.0014041641,0.000021049991,0.00007328397,0.0025050137,0.002174434,0.00001555905,0.0002645262,0.00008082061],"about_ca_topic_score_codex":0.0001477804,"about_ca_topic_score_gemma":0.000146976,"teacher_disagreement_score":0.06415165,"about_ca_system_score_codex":0.000025890704,"about_ca_system_score_gemma":0.00005847551,"threshold_uncertainty_score":0.97682846},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4386958572","doi":"10.47750/jett.2023.14.04.019","title":"Bibliometric Analysis of Studies on Well-Being","year":2023,"lang":"en","type":"article","venue":"Journal for Educators Teachers and Trainers","topic":"Psychological Well-being and Life Satisfaction","field":"Psychology","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Monash University; University College London; King's College London; Deakin University; University of Oxford; University of Toronto; World Health Organization","keywords":"Computer science","score_opus":0.074757083180157,"score_gpt":0.4347583891030321,"score_spread":0.3600013059228751,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386958572","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9871043,0.00065383245,0.000035648427,0.0009570363,0.0031357592,0.00010169301,0.0000047746257,0.00004794513,0.007959035],"genre_scores_gemma":[0.99586457,0.00045258732,0.000085675056,0.00022874873,0.00026927117,0.0000183954,0.000006600044,0.000017189317,0.0030569849],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99868345,0.000112336056,0.00042187376,0.0002641329,0.0002001552,0.00031806927],"domain_scores_gemma":[0.9986919,0.0006655159,0.0002556268,0.00014658312,0.00007188685,0.00016848979],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.001443599,0.0001549402,0.00041534155,0.044380862,0.00020112033,0.000034817862,0.00010794179,0.00010996641,0.00012012581],"category_scores_gemma":[0.0003389926,0.0001153425,0.0003231607,0.054655556,0.00010467524,0.000055530792,0.000010206737,0.00029577632,0.00001958787],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00042981785,0.00073236524,0.38300562,0.00009113793,0.02047416,0.00004288383,0.099824175,0.0008671419,0.00046399093,0.015557724,0.22385849,0.2546525],"study_design_scores_gemma":[0.0011362802,0.0007908641,0.9124991,0.000050876555,0.0014477747,0.00002525498,0.06263795,0.000078442215,0.000043210974,0.0050882157,0.015918639,0.00028336575],"about_ca_topic_score_codex":0.000024794937,"about_ca_topic_score_gemma":0.0000029468233,"teacher_disagreement_score":0.5294935,"about_ca_system_score_codex":0.00004681408,"about_ca_system_score_gemma":0.000020325493,"threshold_uncertainty_score":0.9664503},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4386964325","doi":"10.15244/pjoes/169016","title":"Bibliometric Overview of Research on Tasteand Odor in Drinking Waterduring the 1980-2022","year":2023,"lang":"en","type":"article","venue":"Polish Journal of Environmental Studies","topic":"Advanced Chemical Sensor Technologies","field":"Engineering","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Odor; Taste; Environmental science; Environmental chemistry; Chemistry; Food science; Organic chemistry","score_opus":0.14154273648788782,"score_gpt":0.3778527665140644,"score_spread":0.2363100300261766,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386964325","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9810513,0.018349623,0.000001440443,0.0001904616,0.00011753871,0.00007316961,0.0000069208536,0.000035958026,0.00017357142],"genre_scores_gemma":[0.9474649,0.052372836,0.000042380823,0.000008559763,0.00005937478,0.0000048888423,4.369249e-7,0.000018903942,0.000027750732],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99861753,0.000037777056,0.00038375528,0.00009384914,0.0005362811,0.00033081634],"domain_scores_gemma":[0.9992399,0.0004931836,0.000083080755,0.00014320937,0.000012377643,0.000028255738],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004475944,0.000114387076,0.00026972624,0.00627523,0.00007205019,0.000011112384,0.000317752,0.000042724554,0.000011843349],"category_scores_gemma":[0.00023308239,0.000076564575,0.000066949564,0.009561344,0.00024693407,0.00011257881,0.00030636764,0.00050901854,0.000013546818],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000111978494,0.0003101159,0.092275426,0.0007896762,0.00093724986,0.0004307446,0.0041098557,0.09469361,0.6972886,0.00045469997,0.0051523405,0.10344574],"study_design_scores_gemma":[0.0010638394,0.00032818067,0.3261079,0.00093809667,0.00003187652,0.000053373144,0.017200444,0.00018287089,0.6438357,0.0053057563,0.004628629,0.0003232825],"about_ca_topic_score_codex":0.0000036527404,"about_ca_topic_score_gemma":0.0000029898251,"teacher_disagreement_score":0.23383248,"about_ca_system_score_codex":0.00027574244,"about_ca_system_score_gemma":0.0000010514129,"threshold_uncertainty_score":0.55993223},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4387132604","doi":"10.1080/01431161.2023.2257862","title":"Visual analysis of coal fire detection research based on bibliometrics","year":2023,"lang":"en","type":"article","venue":"International Journal of Remote Sensing","topic":"Coal Properties and Utilization","field":"Engineering","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Natural Science Foundation of China","keywords":"Coal; China; Remote sensing; Bibliometrics; Data science; Geography; Environmental science; Environmental resource management; Computer science; Library science; Archaeology","score_opus":0.06608073034098433,"score_gpt":0.3609703599536663,"score_spread":0.29488962961268195,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387132604","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8645962,0.000055918415,0.13375166,0.00018871014,0.0008840237,0.000030626597,0.0000036866816,0.000036936104,0.00045224972],"genre_scores_gemma":[0.9990011,0.00015878961,0.00060400215,0.000019055062,0.00017626019,8.802185e-9,0.0000067490637,0.000013369336,0.000020704498],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984859,0.00005056615,0.00033338208,0.00006738528,0.00094423455,0.000118498254],"domain_scores_gemma":[0.9986601,0.00021356998,0.00009634679,0.00006729498,0.0009209222,0.00004175773],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00076195324,0.00006186964,0.00014920923,0.011455788,0.00003649699,0.000053134565,0.00011623806,0.000050045164,0.00001409404],"category_scores_gemma":[0.00038390327,0.000056491812,0.00013702021,0.009361891,0.000028135448,0.00009293655,0.000023311139,0.0002099039,0.0000060093444],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000689431,0.00000766597,0.0000827875,0.000008345201,0.00026103517,0.000033924924,0.0000509505,0.58439296,0.009577205,0.000001509622,0.00017568043,0.405339],"study_design_scores_gemma":[0.00018189446,0.00009284625,0.00453623,0.00008072498,0.00005146991,0.000008861797,0.00007311961,0.97530603,0.018311054,0.00003688733,0.0012732064,0.000047668742],"about_ca_topic_score_codex":0.00006501457,"about_ca_topic_score_gemma":0.000024673944,"teacher_disagreement_score":0.40529132,"about_ca_system_score_codex":0.00014389073,"about_ca_system_score_gemma":0.000024502255,"threshold_uncertainty_score":0.9997485},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4387271680","doi":"10.59876/a-epek-m1w1","title":"De la “brecha de género” al “techo de cristal”: un análisis bibliométrico de la presencia de mujeres en los consejos de administración de empresas a través de mapas de evolución","year":2023,"lang":"en","type":"article","venue":"Management international","topic":"Business, Innovation, and Economy","field":"Economics, Econometrics and Finance","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Humanities; Scientific field; Web of science; Relevance (law); Political science; Sociology; Art; Work (physics); Engineering; Law","score_opus":0.02123969862122938,"score_gpt":0.2816802421827961,"score_spread":0.2604405435615667,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387271680","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8682068,0.00022166866,0.09126883,0.0020065564,0.00024521005,0.0003291029,0.0004154472,0.0002057041,0.0371007],"genre_scores_gemma":[0.97424155,0.00108105,0.017792925,0.0025811964,0.0004147026,0.0003504791,0.00013870824,0.00008532093,0.0033140606],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9971581,0.0001674696,0.0007957247,0.000612192,0.00017481264,0.0010917059],"domain_scores_gemma":[0.99832475,0.0005193645,0.00039546198,0.0004042242,0.00011705981,0.00023911853],"candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.004016409,0.0003427397,0.00035828768,0.002939885,0.00023654422,0.00064700574,0.0010324914,0.00034461575,0.0011929654],"category_scores_gemma":[0.0007221005,0.000468473,0.0002023541,0.0017677525,0.00016476378,0.00044318705,0.00028438953,0.0004341746,0.00016105697],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012789891,0.00043519656,0.8280381,0.00019612662,0.000624278,0.00034560676,0.001462504,0.003994618,0.00029280598,0.074289285,0.086831845,0.0033617555],"study_design_scores_gemma":[0.0012935563,0.000041929437,0.74377084,0.00007341819,0.00006095009,0.00017757938,0.00069904106,0.053880073,0.00096422486,0.07393928,0.12457261,0.00052650995],"about_ca_topic_score_codex":0.0022257655,"about_ca_topic_score_gemma":0.00021947504,"teacher_disagreement_score":0.10603478,"about_ca_system_score_codex":0.0022442432,"about_ca_system_score_gemma":0.00049898087,"threshold_uncertainty_score":0.9997767},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4387276382","doi":"10.1016/j.respol.2023.104895","title":"Gender homophily: In-group citation preferences and the gender disadvantage","year":2023,"lang":"en","type":"article","venue":"Research Policy","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":39,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Fundamental Research Funds for the Central Universities; National Natural Science Foundation of China","keywords":"Homophily; Citation; Disadvantage; Gender gap; Set (abstract data type); Field (mathematics); Sociology; Psychology; Social psychology; Political science; Demographic economics; Computer science","score_opus":0.8735278668469351,"score_gpt":0.6882343069839548,"score_spread":0.18529355986298024,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387276382","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9542719,0.0010186238,0.00019386955,0.013346482,0.00011797197,0.0007445675,0.000029399516,0.000043095457,0.030234082],"genre_scores_gemma":[0.995448,0.002490666,0.000072302995,0.00018825273,0.0002084966,0.000092169146,0.000005240861,0.000012507266,0.0014823994],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9831686,0.002005097,0.0005509446,0.0007606454,0.012216867,0.0012978163],"domain_scores_gemma":[0.9761494,0.020934854,0.00008522092,0.0008161785,0.0015664946,0.0004478601],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.077597044,0.00012797992,0.00025847784,0.08438631,0.0005231457,0.0027653167,0.002111685,0.00010502212,0.00016649648],"category_scores_gemma":[0.093715794,0.00006848329,0.00008225002,0.27412868,0.0010935054,0.0005956709,0.0015272305,0.00064000615,0.0009851077],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022643528,0.0001842767,0.2673964,0.00004929052,0.000039519182,0.000065762404,0.009688186,0.00009517499,0.00086504413,0.34833595,0.043433987,0.32961997],"study_design_scores_gemma":[0.0008367837,0.000047657322,0.590648,0.000003912742,8.1768644e-7,0.0000029285038,0.00354194,0.004818919,0.00002330403,0.39222282,0.0077676293,0.00008529945],"about_ca_topic_score_codex":0.0011117919,"about_ca_topic_score_gemma":0.00013438713,"teacher_disagreement_score":0.32953468,"about_ca_system_score_codex":0.00015086264,"about_ca_system_score_gemma":0.0004418918,"threshold_uncertainty_score":0.99979275},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4387340059","doi":"10.54535/rep.1346215","title":"How Do We Perceive The Phenomenon of \"Co-Parenting\"? Do We Share Responsibilities Sufficiently?: A Bibliometric Study","year":2023,"lang":"en","type":"article","venue":"Research on Education and Psychology","topic":"Family Dynamics and Relationships","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Content analysis; Context (archaeology); Citation; Web of science; Scope (computer science); Psychology; Citation analysis; Sociology; Library science; Social science; Meta-analysis; Computer science; Medicine; Geography","score_opus":0.27644053816956454,"score_gpt":0.5357274157311833,"score_spread":0.2592868775616188,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387340059","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92542547,0.0025571676,0.0000019298432,0.048110284,0.0005715256,0.00069700094,0.000023406305,0.000029010382,0.02258422],"genre_scores_gemma":[0.9727945,0.00982055,0.000009069967,0.000059176255,0.00015194701,0.000098902194,0.000009780107,0.000009933811,0.01704615],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9962712,0.0020028364,0.00019573251,0.0003682054,0.0007590517,0.0004029766],"domain_scores_gemma":[0.99701744,0.001968374,0.00007757085,0.00037671768,0.00044168817,0.00011818201],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00585721,0.00008499301,0.0001330078,0.012501393,0.0010250931,0.0002718774,0.00038004984,0.000108864115,0.00020455274],"category_scores_gemma":[0.0012872645,0.00006160127,0.00004146305,0.029637557,0.000662409,0.00008744228,0.000053786625,0.00041108366,0.000085191015],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022765413,0.003610064,0.117289156,0.0000914882,0.00008528868,0.0000033496867,0.44426987,0.000012992721,0.00017413453,0.15275891,0.09751266,0.18396445],"study_design_scores_gemma":[0.00020807816,0.0003938183,0.39977443,0.000039108647,0.000004511973,6.457033e-7,0.47040302,0.000021155643,6.556942e-7,0.010223347,0.118857406,0.00007382538],"about_ca_topic_score_codex":0.00021345206,"about_ca_topic_score_gemma":0.00019345764,"teacher_disagreement_score":0.28248528,"about_ca_system_score_codex":0.00006804765,"about_ca_system_score_gemma":0.00038260495,"threshold_uncertainty_score":0.9986911},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4387568360","doi":"10.1057/s41599-023-02109-8","title":"Bibliometric analysis of willingness to communicate in the English as a second language (ESL) context","year":2023,"lang":"en","type":"article","venue":"Humanities and Social Sciences Communications","topic":"Second Language Learning and Teaching","field":"Arts and Humanities","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"China; Context (archaeology); Willingness to communicate; Citation; Beijing; Thematic analysis; Political science; Library science; Regional science; Social science; Geography; Psychology; Sociology; Qualitative research; Computer science; Social psychology","score_opus":0.1474462717913809,"score_gpt":0.34423037352521485,"score_spread":0.19678410173383395,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387568360","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.89640063,0.0011208695,8.3315734e-7,0.0009550752,0.000045562698,0.00010151665,0.000039888353,0.000050489765,0.10128516],"genre_scores_gemma":[0.9967338,0.00012122429,0.00001635482,0.001666242,0.000056549106,0.00003078053,0.000047521517,0.0000053301487,0.0013222111],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9988942,0.0003528486,0.0002363773,0.00011988051,0.00020518001,0.00019147617],"domain_scores_gemma":[0.9983481,0.0010302742,0.000102743106,0.0004192693,0.00008595477,0.000013646635],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.0014936766,0.00008140015,0.00021062893,0.010265153,0.0021348007,0.00045315272,0.0011804485,0.000025531359,0.0005676002],"category_scores_gemma":[0.0001494382,0.00006219734,0.000080874896,0.01752681,0.0010535991,0.00017653996,0.00028414658,0.00021557955,0.000011327879],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000012309146,0.00002553782,0.001159913,0.0000070923425,0.00004443393,4.81271e-7,0.7267401,0.0000071748277,0.0000031484408,0.2644336,0.0003342472,0.0072430503],"study_design_scores_gemma":[0.00008817288,0.000042757372,0.015361612,0.000012040482,0.00006602377,2.0287028e-7,0.80668974,0.00032660158,0.0000010380901,0.00026504917,0.17704959,0.00009719054],"about_ca_topic_score_codex":0.009188335,"about_ca_topic_score_gemma":0.08409041,"teacher_disagreement_score":0.26416853,"about_ca_system_score_codex":0.000012253138,"about_ca_system_score_gemma":0.000021952692,"threshold_uncertainty_score":0.9991643},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4387761133","doi":"10.14349/sumapsi.2023.v30.n1.8","title":"30 años de investigación en maltrato infantil: un análisis bibliométrico (1991-2020)","year":2023,"lang":"es","type":"article","venue":"Suma Psicológica","topic":"Child Abuse and Trauma","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Humanities; Political science; Psychology; Sexual abuse; Poison control; Medicine; Philosophy; Human factors and ergonomics","score_opus":0.022159230021386214,"score_gpt":0.3106780684847461,"score_spread":0.2885188384633599,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387761133","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9337953,0.0024236913,0.000009634395,0.008410591,0.0016243806,0.0009682544,0.00043814428,0.0010117363,0.051318288],"genre_scores_gemma":[0.984704,0.0047311876,0.00024137244,0.003674428,0.0018943461,0.00018703428,0.00025530963,0.00022868863,0.004083643],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99391186,0.00082420046,0.0010469985,0.0014905245,0.0007426039,0.0019838104],"domain_scores_gemma":[0.99605894,0.0010050964,0.0003412619,0.0015983634,0.00014287226,0.0008534598],"candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00094247935,0.0008464664,0.0009570784,0.0040912,0.00053222565,0.00056547154,0.0012218974,0.0009906059,0.018653706],"category_scores_gemma":[0.00029905376,0.0008641252,0.0005435244,0.01680269,0.0004093182,0.00031579557,0.00040364038,0.0012370799,0.033190012],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004187082,0.0012830707,0.19022258,0.0005023608,0.0015439525,0.003589684,0.010014412,0.000049872593,0.004192655,0.026215395,0.68813586,0.07383144],"study_design_scores_gemma":[0.002802736,0.00030777324,0.5469938,0.00023735811,0.00027684122,0.0001462877,0.0007639544,0.00014070234,0.0013901948,0.002608587,0.44342455,0.00090721616],"about_ca_topic_score_codex":0.0006890868,"about_ca_topic_score_gemma":0.000066627144,"teacher_disagreement_score":0.3567712,"about_ca_system_score_codex":0.00012727288,"about_ca_system_score_gemma":0.0002518526,"threshold_uncertainty_score":0.99938095},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4387812970","doi":"10.1590/1809-2950/e22021823en","title":"Alberta infant motor scale in Brazilian research: a bibliometric study","year":2023,"lang":"en","type":"article","venue":"Fisioterapia e Pesquisa","topic":"Infant Development and Preterm Care","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; SciELO; Gross motor skill; Portuguese; Web of science; Context (archaeology); Scale (ratio); Inclusion (mineral); Psychology; Geography; Motor skill; MEDLINE; Political science; Developmental psychology; Cartography; Social psychology","score_opus":0.0672997424082333,"score_gpt":0.3784307429083146,"score_spread":0.31113100050008125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387812970","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98291737,0.000118228665,5.566991e-7,0.0011585981,0.00037089057,0.0014930704,0.000033816716,0.00013827342,0.01376919],"genre_scores_gemma":[0.9919603,0.00010479184,0.00020319012,0.00030087403,0.00021069887,0.00017268899,0.00019510406,0.000046720215,0.0068056644],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99666196,0.00023237758,0.0005169669,0.0006002446,0.0011361822,0.0008522762],"domain_scores_gemma":[0.9982426,0.000533409,0.000052243617,0.0006770597,0.00023455975,0.00026013056],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0018100603,0.00023792178,0.00045506633,0.046476822,0.00016127189,0.000110247856,0.0003300453,0.00016932777,0.00093895465],"category_scores_gemma":[0.0006860907,0.00020607976,0.00011451372,0.0754625,0.000077680954,0.0001863933,0.000306447,0.000648456,0.00079637783],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046040348,0.0001541798,0.94345415,0.00011706426,0.000059289876,0.00046416582,0.015332309,0.0000012157703,0.0014401955,0.0000066347347,0.029890323,0.008620101],"study_design_scores_gemma":[0.0025949476,0.0008180972,0.9213397,0.00020612104,0.0000140639495,0.00002474422,0.00084293995,0.0002954661,0.00031241289,0.00007476476,0.073199406,0.0002773306],"about_ca_topic_score_codex":0.001601909,"about_ca_topic_score_gemma":0.0011513686,"teacher_disagreement_score":0.043309078,"about_ca_system_score_codex":0.00012090699,"about_ca_system_score_gemma":0.00017375524,"threshold_uncertainty_score":0.99998164},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4388037349","doi":"10.1016/j.rcot.2023.10.004","title":"Analyse bibliométrique des articles négatifs publiés dans les dix revues majeures d’orthopédie-traumatologie en 2009–2010 et 2019–2020 : comparaison à Orthopaedics &amp; Traumatology: Surgery &amp; Research avec la même analyse bibliométrique sur les articles soumis et acceptés","year":2023,"lang":"fr","type":"article","venue":"Revue de Chirurgie Orthopédique et Traumatologique","topic":"Venous Thromboembolism Diagnosis and Management","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"","keywords":"Humanities; Traumatology; Philosophy; Medicine; Orthopedic surgery; Surgery","score_opus":0.184297280005222,"score_gpt":0.4057570095264749,"score_spread":0.22145972952125292,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388037349","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.88595605,0.025148317,0.008585268,0.07022098,0.0010123187,0.0031553714,0.0007741008,0.0022545608,0.002893042],"genre_scores_gemma":[0.7577197,0.22713387,0.007084773,0.0013837965,0.000496738,0.0012189227,0.0015411141,0.0005352191,0.002885882],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9602908,0.023067014,0.0052967006,0.0036312465,0.002091591,0.005622671],"domain_scores_gemma":[0.9653631,0.024988525,0.0018693516,0.003522241,0.0020061948,0.0022505827],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","research_integrity","insufficient_payload"],"category_scores_codex":[0.037676893,0.002590298,0.0060147145,0.022757873,0.0015065027,0.0017761507,0.0022706948,0.0039210934,0.0016594502],"category_scores_gemma":[0.022326602,0.002586329,0.0027189057,0.037814744,0.003793427,0.0021833936,0.0015030279,0.007367985,0.00090902124],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006458922,0.0045964103,0.70923066,0.00649531,0.0027757592,0.006671226,0.019204462,0.005019351,0.001942263,0.021301921,0.18560009,0.036516648],"study_design_scores_gemma":[0.0029458026,0.0005739236,0.8287778,0.00710302,0.0016802717,0.0054243696,0.011732833,0.0031913777,0.003686888,0.006165709,0.12555836,0.0031596296],"about_ca_topic_score_codex":0.018382227,"about_ca_topic_score_gemma":0.17390116,"teacher_disagreement_score":0.20198555,"about_ca_system_score_codex":0.0009850633,"about_ca_system_score_gemma":0.0033920212,"threshold_uncertainty_score":0.99986887},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4388420614","doi":"10.1080/02791072.2023.2276230","title":"Three Decades of Research on the Development of Ibogaine Treatment of Substance Use Disorders: A Scientometric Analysis","year":2023,"lang":"en","type":"article","venue":"Journal of Psychoactive Drugs","topic":"Psychedelics and Drug Studies","field":"Psychology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Substance abuse; Medicine; Meta-analysis; Psychology; Psychiatry","score_opus":0.2743457244069428,"score_gpt":0.48804947606214655,"score_spread":0.21370375165520372,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388420614","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99353564,0.00087000226,0.0006929739,0.000805831,0.00031175758,0.00023780733,0.000032070195,0.000005299984,0.003508594],"genre_scores_gemma":[0.9983645,0.00066758465,0.00056338584,0.000009650557,0.00005099922,0.000020419957,0.0000017965905,0.000014411787,0.00030723654],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9970444,0.00023868789,0.00094322796,0.00026079695,0.0011600509,0.00035286025],"domain_scores_gemma":[0.9951319,0.002360354,0.00096777827,0.0004564239,0.0010023088,0.000081237275],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027133557,0.00017259501,0.0006386137,0.0047369483,0.0001438004,0.0000150248125,0.00042978176,0.000068437985,0.000100660334],"category_scores_gemma":[0.00016717835,0.00010485801,0.0004104229,0.013822374,0.00036763411,0.000086510874,0.000026906346,0.00020491082,0.00001323177],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0054757884,0.010141743,0.6152348,0.0000980512,0.025699537,0.000017699867,0.23286787,0.00050028757,0.0040148087,0.011453581,0.011260303,0.08323554],"study_design_scores_gemma":[0.0018058782,0.0011110994,0.9617434,0.000113552174,0.00027832683,0.000001069251,0.027024554,0.00003911457,0.0030284284,0.002273278,0.0024540466,0.00012722325],"about_ca_topic_score_codex":0.00010551685,"about_ca_topic_score_gemma":0.00046058474,"teacher_disagreement_score":0.34650862,"about_ca_system_score_codex":0.00014760897,"about_ca_system_score_gemma":0.000095600364,"threshold_uncertainty_score":0.66411924},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4388566240","doi":"10.18280/ijsse.130503","title":"Bibliometric Analysis of Electric Vehicle Adoption Research: Trends, Implications, and Future Directions","year":2023,"lang":"en","type":"article","venue":"International Journal of Safety and Security Engineering","topic":"Electric Vehicles and Infrastructure","field":"Engineering","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Human factors and ergonomics; Transport engineering; Poison control; Computer science; Regional science; Engineering; Geography; Environmental health; Medicine","score_opus":0.01413739329238981,"score_gpt":0.27682189447032074,"score_spread":0.2626845011779309,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388566240","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9913368,0.006276496,0.0010265101,0.00072459836,0.000306399,0.000031753723,0.00003886716,0.000060261187,0.00019828274],"genre_scores_gemma":[0.9631584,0.036383413,0.00014491893,0.0000051540665,0.00027552625,0.0000012362723,0.000014536179,0.000010191621,0.000006598895],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9990838,0.000017580847,0.00034435815,0.00008379291,0.0003181944,0.0001522858],"domain_scores_gemma":[0.99931645,0.00014935869,0.00006799057,0.00006703374,0.00032246654,0.0000766972],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00045408067,0.000082388025,0.00019328456,0.045063984,0.00004850064,0.00004055581,0.00013922674,0.00007410448,0.000017770113],"category_scores_gemma":[0.000052673477,0.00007994877,0.00008281201,0.05649408,0.000015028096,0.00021922502,0.000028736482,0.0003297567,4.5582047e-7],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008808011,0.00005118867,0.013607547,0.00009623891,0.0032357269,0.000020140398,0.0014231179,0.11020831,0.028920583,0.008370021,0.0009713112,0.83300775],"study_design_scores_gemma":[0.00024737685,0.000051314008,0.75794756,0.000020602656,0.00010804475,0.00004971128,0.00007598587,0.21817036,0.0004489509,0.0002899882,0.02249664,0.00009346411],"about_ca_topic_score_codex":0.0000086430355,"about_ca_topic_score_gemma":0.000003471532,"teacher_disagreement_score":0.8329143,"about_ca_system_score_codex":0.000066958884,"about_ca_system_score_gemma":0.000011807248,"threshold_uncertainty_score":0.9657594},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4388750665","doi":"10.1016/j.erss.2023.103340","title":"From resistance to resilience: A comprehensive bibliometric analysis of carbon pricing public acceptance","year":2023,"lang":"en","type":"article","venue":"Energy Research & Social Science","topic":"Energy, Environment, Economic Growth","field":"Economics, Econometrics and Finance","cited_by":13,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Trinity Western University; Western University; York University","funders":"","keywords":"Thematic analysis; Citation; Bibliometrics; Resilience (materials science); Business; Environmental economics; Political science; Public relations; Knowledge management; Sociology; Computer science; Qualitative research; Economics; Social science; Library science","score_opus":0.09182355085197103,"score_gpt":0.33383403514756566,"score_spread":0.24201048429559463,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388750665","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9466089,0.00072787964,0.00043788351,0.0016047171,0.00021965949,0.00013929236,0.00013693723,0.00006367812,0.050061077],"genre_scores_gemma":[0.99622667,0.0005123112,0.0003445308,0.00011882218,0.00016786794,0.0000796009,0.000018538612,0.000026495782,0.0025051846],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99565184,0.00013323223,0.0008430324,0.0013778944,0.00067500316,0.0013189819],"domain_scores_gemma":[0.99756724,0.00056026765,0.00041349002,0.0008070334,0.0002594595,0.00039249976],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.004199721,0.00019792492,0.00072090334,0.071274154,0.00072983553,0.00028551347,0.0019774102,0.00012851984,0.00020689105],"category_scores_gemma":[0.0016225937,0.00024245221,0.00018645327,0.33196747,0.0012840665,0.00057678414,0.00090061035,0.00028092504,0.00013198238],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006451865,0.00021346903,0.23764245,0.00002832573,0.0005252509,0.000021567304,0.0041230693,0.004984195,0.01750134,0.72408247,0.0035095015,0.0073038046],"study_design_scores_gemma":[0.00022769113,0.000048472506,0.9546524,0.000011367631,0.0000134962775,8.0131734e-8,0.0009211328,0.004497058,0.0017200278,0.014035608,0.023528539,0.00034411377],"about_ca_topic_score_codex":0.004674801,"about_ca_topic_score_gemma":0.0008286145,"teacher_disagreement_score":0.71700996,"about_ca_system_score_codex":0.00088382093,"about_ca_system_score_gemma":0.00017747645,"threshold_uncertainty_score":0.98869145},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4388837278","doi":"10.1016/j.arr.2023.102129","title":"Neuropsychiatric disturbances in mild cognitive impairment: A scientometric analysis","year":2023,"lang":"en","type":"review","venue":"Ageing Research Reviews","topic":"Dementia and Cognitive Impairment Research","field":"Medicine","cited_by":20,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital; University of Ottawa","funders":"Chulalongkorn University","keywords":"Dementia; Delirium; Cognition; Psychology; Citation; Cognitive impairment; Web of science; Clinical psychology; Medicine; Psychiatry; Disease; Meta-analysis; Computer science; Library science; Pathology","score_opus":0.3064288749104257,"score_gpt":0.5371566520940467,"score_spread":0.23072777718362103,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388837278","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000380347,0.9848474,0.000019656318,0.000082273415,0.00016108999,0.0099787675,0.00006459552,0.000098182725,0.0043676756],"genre_scores_gemma":[0.00024846036,0.98653316,0.00007174883,0.00005797561,0.00031315393,0.0032439653,0.0005014009,0.00013206148,0.008898091],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9857088,0.0045444346,0.0021665294,0.0019680916,0.0033804257,0.0022317234],"domain_scores_gemma":[0.99431324,0.0027044574,0.0005169444,0.0010599127,0.0006704773,0.00073499087],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.017399263,0.0007905486,0.00513465,0.031039212,0.00038535654,0.0003717185,0.0009314408,0.00031189428,0.00052483735],"category_scores_gemma":[0.006172035,0.00055759674,0.002557648,0.12152005,0.000463437,0.00020365049,0.0008168503,0.00304374,0.004346776],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000046056866,0.00065042713,0.007467511,0.056374937,0.0013170943,0.0016809237,0.00006691997,5.544983e-8,1.4199388e-7,0.000015986976,0.0028364786,0.92954344],"study_design_scores_gemma":[0.0010214658,0.0007876927,0.033933967,0.059615105,0.0063082622,0.000028172079,0.0000689289,0.000022902359,2.807539e-7,0.000044873137,0.89764994,0.00051841204],"about_ca_topic_score_codex":0.0002744688,"about_ca_topic_score_gemma":0.000118759264,"teacher_disagreement_score":0.92902505,"about_ca_system_score_codex":0.0006745023,"about_ca_system_score_gemma":0.0011093956,"threshold_uncertainty_score":0.99968755},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4388848410","doi":"10.1016/j.bej.2023.109155","title":"Bibliometric analysis of research trends in microbial fuel cells for wastewater treatment","year":2023,"lang":"en","type":"article","venue":"Biochemical Engineering Journal","topic":"Microbial Fuel Cells and Bioremediation","field":"Environmental Science","cited_by":15,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Microbial fuel cell; Sewage treatment; Wastewater; Environmental science; Waste management; Pulp and paper industry; Biochemical engineering; Chemistry; Environmental engineering; Engineering","score_opus":0.03134486947288326,"score_gpt":0.2917924260973652,"score_spread":0.26044755662448194,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388848410","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99949664,0.00006602288,0.00008012753,0.00007976456,0.00012664587,0.00006635808,0.000026315507,0.000013394821,0.00004471752],"genre_scores_gemma":[0.99819475,0.00020841225,0.0011397409,0.000002959799,0.000099862584,0.0000060779603,0.000043332468,0.000010385456,0.0002944599],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9989413,0.000017174312,0.000275746,0.00017038899,0.00024997586,0.00034540062],"domain_scores_gemma":[0.99964803,0.000091596085,0.00004366857,0.000099572644,0.00001833529,0.0000988223],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00061193143,0.000091789996,0.0001849325,0.03077142,0.000027766922,0.000028693763,0.00014843728,0.00008272519,0.0004035363],"category_scores_gemma":[0.000044185766,0.00007281635,0.00016817672,0.09309164,0.00003579609,0.000046989175,0.000056360313,0.000115405914,0.000026813985],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019877087,0.00006008978,0.0014644243,0.000013519933,0.00005116792,0.0000038756484,0.00008392699,0.015580443,0.9774461,7.0201366e-7,0.003329777,0.0019461028],"study_design_scores_gemma":[0.0007226556,0.00017056536,0.045702785,0.000016749924,0.00009037312,0.000003794422,0.000018420085,0.031506374,0.91547465,0.000007898165,0.006127286,0.00015846679],"about_ca_topic_score_codex":0.000051204162,"about_ca_topic_score_gemma":0.0000077016175,"teacher_disagreement_score":0.06232022,"about_ca_system_score_codex":0.0002492408,"about_ca_system_score_gemma":0.000006800738,"threshold_uncertainty_score":0.98021394},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4388851052","doi":"10.1002/asi.24852","title":"The decrease in uncited articles and its effect on the concentration of citations","year":2023,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"Canada Research Chairs","keywords":"Computer science","score_opus":0.20225363668285168,"score_gpt":0.48218410740754425,"score_spread":0.27993047072469257,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388851052","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9777452,0.000048354505,0.000057444046,0.02146923,0.00017142508,0.00024182875,0.00000642467,0.0000042728907,0.0002558378],"genre_scores_gemma":[0.999736,0.000093724535,0.0000138249425,0.00009037648,0.000006032454,0.0000071152926,1.8205232e-7,7.52073e-7,0.0000519967],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.996307,0.00009879429,0.00054134626,0.00006496532,0.002807683,0.00018021373],"domain_scores_gemma":[0.9859705,0.008291154,0.0009329929,0.00014536854,0.004609943,0.000050046656],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.04063336,0.000038103935,0.00009696258,0.008245004,0.0005600778,0.00057876314,0.0007876801,0.000052501902,0.0000013535935],"category_scores_gemma":[0.22826537,0.000015566971,0.000032505577,0.07203114,0.0002487827,0.0011028645,0.00013180339,0.00013568945,0.0000055936694],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000114363196,0.000047961566,0.42123926,0.000010193347,0.000032898475,4.421067e-7,0.0019473666,0.0005072628,0.011086654,0.29706132,0.011802135,0.25615016],"study_design_scores_gemma":[0.0016681366,0.0007969904,0.73380077,0.000038776598,0.000016137701,0.000011907273,0.005706922,0.10000037,0.04311464,0.10031005,0.014427559,0.00010775811],"about_ca_topic_score_codex":0.0000021866533,"about_ca_topic_score_gemma":0.000005248946,"teacher_disagreement_score":0.3125615,"about_ca_system_score_codex":0.00011806785,"about_ca_system_score_gemma":0.00023153894,"threshold_uncertainty_score":0.98786986},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4388946994","doi":"10.3390/plants12233947","title":"The Development of Floral Scent Research: A Comprehensive Bibliometric Analysis (1987–2022)","year":2023,"lang":"en","type":"article","venue":"Plants","topic":"Plant and animal studies","field":"Agricultural and Biological Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Forestry and Grassland Administration; China Scholarship Council; National Natural Science Foundation of China; Nanjing Forestry University","keywords":"Diversity (politics); Scale (ratio); Ecology; Data science; Sociology; Geography; Biology; Computer science","score_opus":0.3927051653201991,"score_gpt":0.3520854464145326,"score_spread":0.04061971890566651,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388946994","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9983243,0.00060877315,1.265909e-7,0.00031959277,0.000028531282,0.000077441975,0.00005979555,0.00003187676,0.0005495687],"genre_scores_gemma":[0.9986899,0.0008920669,0.000013415684,0.000013656609,0.00004078951,0.000009657367,0.000062928935,2.9959665e-7,0.0002772702],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99879766,0.000069276735,0.00017635159,0.00015414269,0.00047768737,0.00032490556],"domain_scores_gemma":[0.9988824,0.0008939883,0.000045717075,0.000021871112,0.00010594171,0.00005009636],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0005563473,0.0000668492,0.00015139146,0.0021506066,0.00062589534,0.000040608676,0.0002535404,0.000027687985,0.00004596158],"category_scores_gemma":[0.000076028955,0.000019284445,0.0000613194,0.07618027,0.00006408403,0.000024805722,0.00021057209,0.000081051556,0.00016977394],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001403327,0.0000991541,0.6015337,0.000014116889,0.0007424789,0.000026053807,0.00073043845,0.000006517088,0.11647433,0.00006493415,0.028746564,0.25142136],"study_design_scores_gemma":[0.000034796394,0.00003579845,0.9511103,0.0000064657947,0.000009981276,4.4117883e-7,0.0010802018,0.00006385673,0.0010563415,0.000043045857,0.04650516,0.00005362379],"about_ca_topic_score_codex":0.000080648046,"about_ca_topic_score_gemma":0.00380271,"teacher_disagreement_score":0.34957656,"about_ca_system_score_codex":0.000013205098,"about_ca_system_score_gemma":0.0000068768786,"threshold_uncertainty_score":0.94345593},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4388979087","doi":"10.23736/s2724-606x.23.05391-5","title":"Gynecologic oncology top-cited articles: an international analysis","year":2023,"lang":"en","type":"article","venue":"Minerva Obstetrics and Gynecology","topic":"Ovarian cancer diagnosis and treatment","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University; Jewish General Hospital","funders":"","keywords":"Subspecialty; Gynecologic oncology; Obstetrics and gynaecology; Medicine; Gynecology; Obstetrics; Family medicine; Library science; Oncology; Pregnancy","score_opus":0.04218200524399773,"score_gpt":0.3318545002556497,"score_spread":0.289672495011652,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388979087","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9889987,0.00039410667,0.0000348926,0.005678114,0.0010787437,0.00023650983,0.000030110605,0.00011963182,0.0034292028],"genre_scores_gemma":[0.9943734,0.001470454,0.0007446546,0.0015827978,0.000095681455,0.00009565501,0.00024183748,0.000014616285,0.0013809124],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99860114,0.000073951145,0.00032590958,0.00046900718,0.00017128357,0.00035870093],"domain_scores_gemma":[0.996581,0.002608279,0.000101274425,0.00030856332,0.0001598635,0.0002410295],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00025532657,0.00016094171,0.00048764292,0.0010869014,0.00008657275,0.000033757555,0.00011369466,0.00021636745,0.0008936866],"category_scores_gemma":[0.002306842,0.00013272524,0.00013414129,0.0026583963,0.00007057975,0.00008995895,0.00009781708,0.00013519719,0.00011599151],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000034247918,0.00076673477,0.78838915,0.000011545225,0.001220861,0.0009812616,0.00032911025,0.000120987,0.00020205393,0.0015798323,0.001309869,0.20505437],"study_design_scores_gemma":[0.0024193125,0.0014363409,0.8901048,0.0000037050138,0.0009646088,0.000029095141,0.00035959182,0.012125299,0.00007974935,0.00021750561,0.09211502,0.00014501068],"about_ca_topic_score_codex":0.000081365615,"about_ca_topic_score_gemma":0.00019272225,"teacher_disagreement_score":0.20490937,"about_ca_system_score_codex":0.00027877258,"about_ca_system_score_gemma":0.0000932337,"threshold_uncertainty_score":0.97852415},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4389085185","doi":"10.1142/s0218126624300046","title":"IoT Ecosystem Security via Distributed Ledger Technology (Blockchain versus IOTA): A Bibliometric Analysis Research","year":2023,"lang":"en","type":"article","venue":"Journal of Circuits Systems and Computers","topic":"Blockchain Technology Applications and Security","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Brandon University","funders":"","keywords":"Blockchain; Distributed ledger; Ledger; Computer science; Internet of Things; Data science; World Wide Web; Business; Computer security; Accounting","score_opus":0.035531737634544786,"score_gpt":0.30514661744127836,"score_spread":0.2696148798067336,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389085185","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6980976,0.0019851383,0.2968086,0.0014602292,0.0010522341,0.00029175953,0.000019516649,0.0002330061,0.00005194335],"genre_scores_gemma":[0.99925584,0.00012937702,0.000385254,0.000011530473,0.00016165191,0.000028253682,0.0000028221855,0.000012830336,0.000012454303],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99658936,0.00035580338,0.0009791655,0.00054294546,0.0008704716,0.000662246],"domain_scores_gemma":[0.9963886,0.0007255496,0.0006125179,0.0008925212,0.0011246068,0.00025618158],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.003971107,0.00023643214,0.0008074405,0.078020915,0.0004448097,0.00033245902,0.0018211176,0.00039483706,0.0000019027589],"category_scores_gemma":[0.00015136349,0.00021011401,0.00023467478,0.20701873,0.00017008423,0.00009688081,0.00058007945,0.0008476713,0.000023506353],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001316437,0.0015276164,0.039649133,0.0011515687,0.011168723,0.0020398078,0.0042524235,0.025679428,0.0033313518,0.45878166,0.027190259,0.4250964],"study_design_scores_gemma":[0.0028625024,0.0011360033,0.013310736,0.00022520068,0.00029128857,0.0010176931,0.0010997386,0.9492301,0.00038279602,0.01111744,0.018616168,0.0007103864],"about_ca_topic_score_codex":0.00004260479,"about_ca_topic_score_gemma":0.000020400657,"teacher_disagreement_score":0.9235506,"about_ca_system_score_codex":0.00017685964,"about_ca_system_score_gemma":0.00013219257,"threshold_uncertainty_score":0.93242896},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4389149394","doi":"10.3389/fendo.2023.1218968","title":"A bibliometric analysis of PCSK9 inhibitors from 2007 to 2022","year":2023,"lang":"en","type":"article","venue":"Frontiers in Endocrinology","topic":"Lipoproteins and Cardiovascular Health","field":"Medicine","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Key Research and Development Program of Hunan Province of China; Natural Science Foundation of Hunan Province; China Postdoctoral Science Foundation; National Natural Science Foundation of China; Health Commission of Hunan Province","keywords":"PCSK9; Medicine; Internal medicine; Cholesterol; LDL receptor","score_opus":0.016185149421587688,"score_gpt":0.28662311544951335,"score_spread":0.2704379660279257,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389149394","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9899701,0.0038602962,0.0025314484,0.00065919966,0.0015372139,0.0004063196,0.000084587584,0.000079735175,0.0008711159],"genre_scores_gemma":[0.9920072,0.0018610359,0.0044062273,0.00047717954,0.00029230132,0.00008972317,0.00017415656,0.00002705581,0.00066510105],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9981112,0.00010291317,0.0004284616,0.000397567,0.00042048324,0.00053938874],"domain_scores_gemma":[0.99900717,0.000053012525,0.00007199989,0.0006219858,0.00007949967,0.000166325],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00034021423,0.00013238798,0.00092387584,0.136174,0.000026091046,0.0000046025716,0.00014713121,0.00010751323,0.00021599256],"category_scores_gemma":[0.00029952693,0.00012582159,0.00036638396,0.2199627,0.000059362872,0.00003376409,0.0001185408,0.00022406693,0.00005133089],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012546749,0.00008250168,0.7553979,0.000040951316,0.002253308,0.0003120302,0.00018933286,0.00023975442,0.00035843777,0.000029274019,0.05130702,0.18966404],"study_design_scores_gemma":[0.001117284,0.00018451894,0.92416966,0.000018816803,0.0006613175,0.0000066927687,0.0005118925,0.0004968433,0.0013082211,0.00010331867,0.07130377,0.00011766161],"about_ca_topic_score_codex":0.0020926422,"about_ca_topic_score_gemma":0.00009529697,"teacher_disagreement_score":0.18954638,"about_ca_system_score_codex":0.00012534486,"about_ca_system_score_gemma":0.00008532606,"threshold_uncertainty_score":0.87361676},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4389196824","doi":"10.18280/ijsdp.181115","title":"The Village Fund Program and Indonesia’s 18th Sustainable Development Goal: A Bibliometric and Content Study","year":2023,"lang":"en","type":"article","venue":"International Journal of Sustainable Development and Planning","topic":"Local Governance and Development","field":"Social Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Sustainable development; Business; Environmental planning; Environmental economics; Agricultural economics; Economic growth; Environmental resource management; Regional science; Economics; Political science; Environmental science; Geography","score_opus":0.05407915065711168,"score_gpt":0.3393757396689597,"score_spread":0.285296589011848,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389196824","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.994701,0.0020412903,0.00007930224,0.0009909164,0.00030457662,0.00062137627,5.611582e-7,0.000045238525,0.0012157309],"genre_scores_gemma":[0.991144,0.0011329006,0.000660278,0.00009566914,0.00014527774,0.00005724059,0.0000038753933,0.000013798955,0.0067469836],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9971834,0.00010274152,0.0006236436,0.00023160719,0.0012026817,0.0006559197],"domain_scores_gemma":[0.99770665,0.00039682025,0.00037670464,0.000058568214,0.0012297354,0.00023152179],"candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0037278817,0.00018962934,0.0002466165,0.0038600252,0.0013756083,0.001049878,0.00035316063,0.00007166327,0.000005520435],"category_scores_gemma":[0.00054948893,0.00013908015,0.000027502108,0.004523317,0.00015853919,0.0006400398,0.00040528155,0.00021443018,0.0000027510255],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040885643,0.0002543944,0.5459375,0.00008756472,0.0006703013,0.0070096725,0.17664729,0.000021230573,0.000008860376,0.010704107,0.0028865295,0.25536373],"study_design_scores_gemma":[0.0007242866,0.000081283,0.40020224,0.00005540809,0.000008980426,0.0000244072,0.35629088,0.000006335212,0.000022194987,0.00016565587,0.24227858,0.00013976516],"about_ca_topic_score_codex":0.00026066028,"about_ca_topic_score_gemma":0.00007038934,"teacher_disagreement_score":0.25522396,"about_ca_system_score_codex":0.00037381944,"about_ca_system_score_gemma":0.0010826418,"threshold_uncertainty_score":0.9999871},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4389222933","doi":"10.1590/2318-0889202335e236917","title":"Ethnographic studies in Information Science: bibliometric analysis in the Web of Science database","year":2023,"lang":"en","type":"article","venue":"Transinformação","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Conselho Nacional de Desenvolvimento Científico e Tecnológico","keywords":"Ethnography; Information science; Web of science; Field (mathematics); Sociology; Resource (disambiguation); Library science; Computer science; MEDLINE; Political science; Anthropology; Mathematics","score_opus":0.16143876319376632,"score_gpt":0.4726963591968311,"score_spread":0.3112575960030648,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389222933","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9915624,0.00013671868,0.0005470664,0.00054275244,0.00011727853,0.00017604537,0.000009120602,0.000032695694,0.00687588],"genre_scores_gemma":[0.99841833,0.0010396146,0.00041003115,0.00008959792,0.000010365903,0.000009208891,0.0000115802795,0.0000011375254,0.000010105124],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.9969152,0.00012925931,0.0005166058,0.00013862368,0.0019047492,0.00039554574],"domain_scores_gemma":[0.99852765,0.000591377,0.0001274343,0.00020780816,0.0004874845,0.000058259728],"candidate_categories":["bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.019920524,0.00007983033,0.00022340164,0.17839904,0.000579029,0.00012339944,0.0008238026,0.00002916056,0.000016180207],"category_scores_gemma":[0.0012807568,0.000057308116,0.00010595474,0.72945786,0.0035278406,0.003740647,0.00005603967,0.00013736708,0.000013637461],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003100577,0.00011689643,0.16581447,0.00011891159,0.00015667666,0.000006843279,0.5547885,0.02398645,0.00034406825,0.052000318,0.00011817567,0.2025177],"study_design_scores_gemma":[0.0003824637,0.000025540197,0.8248469,0.000027918191,0.0001102163,3.3441862e-7,0.12834185,0.043212794,0.000083039384,0.0014045468,0.0014074061,0.00015694175],"about_ca_topic_score_codex":0.0011879588,"about_ca_topic_score_gemma":0.0035143667,"teacher_disagreement_score":0.65903246,"about_ca_system_score_codex":0.00007081925,"about_ca_system_score_gemma":0.00086252094,"threshold_uncertainty_score":0.999184},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4389448528","doi":"10.47203/ijch.2023.v35i03.024","title":"Sibling Authors Propose Bipolar Disorder Of Publishing Biomedical Research: With MEDLINE Footprint Imploding And Retraction Footprint Fudging, Can Pursuit For PubMed® Footprint Challenge Humane Human Imprinting By Physician Scientists?","year":2023,"lang":"en","type":"article","venue":"Indian Journal of Community Health","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"U.S. National Library of Medicine; National Institutes of Health","keywords":"Footprint; Publishing; Medicine; China; Sibling; Management; Geography; Law; Archaeology; Political science; Economics","score_opus":0.2953544542754976,"score_gpt":0.4867851006620565,"score_spread":0.19143064638655888,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389448528","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96077555,0.0006858702,0.001802411,0.035026956,0.00058027613,0.00086609775,0.000047832717,0.000060252634,0.00015476644],"genre_scores_gemma":[0.9980514,0.00026241894,0.0009380382,0.00014979324,0.00041337777,0.00003338865,0.000029404033,0.000057847697,0.00006434569],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.98766726,0.0031246655,0.0025985558,0.00053078186,0.0045595192,0.0015192059],"domain_scores_gemma":[0.9885654,0.0036736825,0.003218931,0.0010333884,0.002241689,0.0012669106],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","research_integrity"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.11808114,0.0003334408,0.0010247602,0.0038861388,0.0035800268,0.0036227198,0.004590689,0.00037472314,0.0000053820972],"category_scores_gemma":[0.016565297,0.00025987136,0.00019678049,0.0043454426,0.0009975488,0.0017360399,0.0016219169,0.0066752736,8.993788e-7],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014893508,0.00055163057,0.021932822,0.00046874123,0.00013328221,0.0000224967,0.030623617,0.000075905315,0.001965759,0.0020868664,0.0022409095,0.93974906],"study_design_scores_gemma":[0.011174845,0.010570714,0.33624887,0.005708997,0.00013561013,0.00075072784,0.34822905,0.001651326,0.007957051,0.13372318,0.14159961,0.0022499892],"about_ca_topic_score_codex":0.0047054417,"about_ca_topic_score_gemma":0.0038926648,"teacher_disagreement_score":0.93749905,"about_ca_system_score_codex":0.0006021175,"about_ca_system_score_gemma":0.0012082653,"threshold_uncertainty_score":0.99998534},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","research_integrity"],"domain":null,"study_design":"design_other","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4389606960","doi":"10.1111/pan.14811","title":"The pediatric anesthesiology publication activity and landscape over the past two decades: A longitudinal scientometric analysis","year":2023,"lang":"en","type":"article","venue":"Pediatric Anesthesia","topic":"Health and Medical Research Impacts","field":"Medicine","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Anesthesiology; Bibliometrics; Library science; Pathology","score_opus":0.0712793842709587,"score_gpt":0.3943527551865954,"score_spread":0.3230733709156367,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389606960","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96396416,0.002070039,0.00005626249,0.032972988,0.000042982767,0.0005851733,0.0000022470958,0.00012897735,0.00017714275],"genre_scores_gemma":[0.9898212,0.0067068147,0.000023083892,0.0014249182,0.0012832687,0.00010703688,0.000024713714,0.000021259139,0.00058768195],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9962386,0.0003460447,0.00037241232,0.00050819234,0.0012170023,0.0013177677],"domain_scores_gemma":[0.99366826,0.0030955393,0.00021088638,0.0006340241,0.0003073503,0.002083959],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0052920827,0.00021559103,0.00040470762,0.0025418818,0.0007367326,0.00016182664,0.0003446008,0.00015377287,0.00007444083],"category_scores_gemma":[0.0124347545,0.00011133156,0.00016520843,0.0224948,0.00023086458,0.00015282803,0.00008944295,0.00054610765,0.0001882974],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010441079,0.000065167485,0.9658634,0.000074928736,0.00005262198,0.00009781479,0.00008426466,0.000003680396,0.0000025095703,0.00007205232,0.022220029,0.011359096],"study_design_scores_gemma":[0.0007216398,0.00023709564,0.9861784,0.0000018522999,0.0008335451,0.00019730827,0.000055482567,0.0017513757,0.0000011731419,0.00003732688,0.009871103,0.000113681366],"about_ca_topic_score_codex":0.00013733716,"about_ca_topic_score_gemma":0.000016799104,"teacher_disagreement_score":0.031548068,"about_ca_system_score_codex":0.00008701933,"about_ca_system_score_gemma":0.0013985874,"threshold_uncertainty_score":0.99828255},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4389775406","doi":"10.1097/md.0000000000036478","title":"Bibliometrics and visual analysis of metformin and gut microbiota from 2012 to 2022: A systematic review","year":2023,"lang":"en","type":"review","venue":"Medicine","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"College & Association of Registered Nurses of Alberta","funders":"Chengdu Science and Technology Bureau","keywords":"Bibliometrics; Metformin; Medicine; Web of science; Gut flora; MEDLINE; Library science; Variety (cybernetics); China; Diabetes mellitus; Internal medicine; Political science; Meta-analysis; Computer science; Endocrinology","score_opus":0.051735975667300094,"score_gpt":0.4169435686275724,"score_spread":0.36520759296027233,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389775406","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00025003703,0.99772495,0.00006409506,0.000127852,0.00009644499,0.0013207963,0.00039345716,0.0000070559236,0.000015285155],"genre_scores_gemma":[0.000009396685,0.99703234,0.00008994028,0.0004929403,0.00011450839,0.000054801498,0.0014628544,0.000036621197,0.0007065865],"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977657,0.00018899837,0.0011004664,0.0005304191,0.00017398197,0.00024040067],"domain_scores_gemma":[0.99853015,0.00016161568,0.00053683313,0.00045465742,0.00010806181,0.00020866843],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011255345,0.00034532932,0.003784254,0.0047768797,0.000035293982,0.000009765513,0.00020075022,0.0002817367,0.000059708287],"category_scores_gemma":[0.000689303,0.0002309311,0.00024314725,0.011632538,0.00008276434,0.0000023766188,0.0002484892,0.00014328815,0.000013404421],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000036225504,0.000022128852,0.000007956337,0.952028,0.0026621695,0.0000062490276,0.00002535809,1.6846608e-8,0.0031341529,0.0000017826344,0.016036503,0.026072094],"study_design_scores_gemma":[0.00021872381,0.00033923922,0.00009635163,0.3509023,0.032701377,0.000033876466,0.000030122508,0.0000023894952,0.000011655713,9.3119263e-7,0.6153386,0.00032442925],"about_ca_topic_score_codex":0.00017300794,"about_ca_topic_score_gemma":0.00008075473,"teacher_disagreement_score":0.60112566,"about_ca_system_score_codex":0.000019521767,"about_ca_system_score_gemma":0.000094630974,"threshold_uncertainty_score":0.9417097},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4389786568","doi":"10.1186/s13643-023-02410-2","title":"Preliminary guideline for reporting bibliometric reviews of the biomedical literature (BIBLIO): a minimum requirements","year":2023,"lang":"en","type":"article","venue":"Systematic Reviews","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":296,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Checklist; Guideline; Systematic review; MEDLINE; Medicine; Scopus; Process (computing); Bibliometrics; Data science; Computer science; Management science; Library science; Psychology; Pathology; Political science","score_opus":0.7241108159152894,"score_gpt":0.6179719659443282,"score_spread":0.10613884997096112,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389786568","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.079014085,0.61244315,0.081806,0.020640364,0.022294277,0.1700873,0.0009310783,0.00048094481,0.012302809],"genre_scores_gemma":[0.4747724,0.25054976,0.10466994,0.006810553,0.005130322,0.037737053,0.0011071055,0.00068024494,0.11854261],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.95930165,0.0034981621,0.021621851,0.0013390244,0.013165121,0.0010741779],"domain_scores_gemma":[0.94598085,0.016439129,0.02419077,0.0053489073,0.007377742,0.00066262716],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.30402443,0.00039276018,0.0031086341,0.3693186,0.0003526997,0.0013466271,0.0047397395,0.00023282757,0.00015487136],"category_scores_gemma":[0.68068177,0.00016568825,0.0016642107,0.89886045,0.0001831617,0.00056143914,0.0012682425,0.00029456834,0.0006429006],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012068804,0.000114052025,0.0024249258,0.03509367,0.000033903045,0.0000116743395,0.00027098387,0.0000017677099,0.000920639,0.000100867604,0.9187132,0.04230223],"study_design_scores_gemma":[0.0006258288,0.00042187297,0.004296261,0.015031957,0.00011291822,0.00007117442,0.00030227817,0.006154674,0.00018412706,0.0034818898,0.96895295,0.0003640869],"about_ca_topic_score_codex":0.000012878002,"about_ca_topic_score_gemma":0.0000018533232,"teacher_disagreement_score":0.52954185,"about_ca_system_score_codex":0.000094898984,"about_ca_system_score_gemma":0.00021897434,"threshold_uncertainty_score":0.99969006},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"reporting","study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"reporting","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4390046827","doi":"10.1007/s00266-023-03752-x","title":"Top 50 Highly Cited Publications in Facelift Surgery: A 50-Year Bibliometric Analysis Review","year":2023,"lang":"en","type":"article","venue":"Aesthetic Plastic Surgery","topic":"Facial Rejuvenation and Surgery Techniques","field":"Medicine","cited_by":7,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Medicine; Bibliometrics; Rhytidectomy; Otorhinolaryngology; Surgery; Library science","score_opus":0.05055489199474596,"score_gpt":0.314992322557842,"score_spread":0.26443743056309604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390046827","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94923663,0.006021668,0.00074662594,0.038214903,0.0008626309,0.00084675127,0.00008804327,0.0018555322,0.0021272225],"genre_scores_gemma":[0.9619368,0.032605257,0.00012372679,0.0027670376,0.000023842744,0.00032824607,0.00094734493,0.000058648147,0.0012091384],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99647444,0.00024248392,0.0011686445,0.0005628637,0.00090492493,0.00064662646],"domain_scores_gemma":[0.9919605,0.006379594,0.00027020831,0.0006873056,0.0003467176,0.000355671],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0029399563,0.00030713144,0.0014052278,0.13773003,0.00009148716,0.00006925639,0.00015249441,0.00019090691,0.00086717925],"category_scores_gemma":[0.012085886,0.0002749428,0.00077313755,0.2915875,0.0000978121,0.00017599639,0.000063308646,0.00028878034,0.00067821826],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000040472107,0.00024603438,0.59842694,0.00022099291,0.00014925965,0.00012284987,0.000025496076,0.000011635933,0.00006364697,0.00027720566,0.36349115,0.03692434],"study_design_scores_gemma":[0.00015468788,0.000020753048,0.5718374,0.0009674355,0.00030000863,0.0000431534,0.000025532761,0.0003760043,0.00005432405,0.000058167876,0.42585188,0.00031061698],"about_ca_topic_score_codex":0.00015126386,"about_ca_topic_score_gemma":0.00002677614,"teacher_disagreement_score":0.15385745,"about_ca_system_score_codex":0.00012882616,"about_ca_system_score_gemma":0.0004770064,"threshold_uncertainty_score":0.99997026},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4390081751","doi":"10.53660/clm-2593-23u58","title":"Respiratory care in focus: a decadal bibliometric review of research dynamics","year":2023,"lang":"en","type":"article","venue":"Concilium","topic":"Delphi Technique in Research","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Multidisciplinary approach; Respiratory care; Scopus; Health care; MEDLINE; Medicine; Medical education; Library science; Political science; Sociology; Computer science; Social science; Intensive care medicine","score_opus":0.31574879321955684,"score_gpt":0.5656663666884038,"score_spread":0.2499175734688469,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390081751","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20195459,0.3580251,0.00008065236,0.02140182,0.0008693255,0.005880169,0.00031925185,0.0011094956,0.4103596],"genre_scores_gemma":[0.9402434,0.057688653,0.00018332514,0.00021877889,0.00018733967,0.00019625443,0.00002706807,0.000033423268,0.0012217262],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9948468,0.0013237974,0.00041889263,0.00038313918,0.0023417578,0.00068561826],"domain_scores_gemma":[0.99580044,0.0016762365,0.000067348024,0.0008306737,0.0014813549,0.00014395312],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.020334627,0.00008022474,0.00025591627,0.022969536,0.0001800358,0.000039786268,0.0016773641,0.00016845,0.00012476511],"category_scores_gemma":[0.017468486,0.00008339585,0.00005413708,0.16633628,0.00075172324,0.0001896147,0.00072919886,0.00052073924,0.00015480263],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000053753454,0.00010701382,0.0759511,0.010349788,0.0000225058,0.00036355146,0.007185498,0.0000019203512,0.00027196176,0.022547666,0.3033455,0.5797997],"study_design_scores_gemma":[0.0006736646,0.0003905904,0.028508937,0.009256676,0.000012865804,0.0000012466363,0.035795197,0.00029530874,0.0006844242,0.0047788145,0.9191135,0.00048876903],"about_ca_topic_score_codex":0.0040230737,"about_ca_topic_score_gemma":0.013759706,"teacher_disagreement_score":0.7382888,"about_ca_system_score_codex":0.0007109097,"about_ca_system_score_gemma":0.001897652,"threshold_uncertainty_score":0.9908078},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4390095083","doi":"10.1016/j.respol.2023.104938","title":"The interdependent influence of lobbying and intellectual capital on new drug development","year":2023,"lang":"en","type":"article","venue":"Research Policy","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"National University of Singapore; York University","keywords":"Interdependence; Business; Pharmaceutical industry; Criticism; Social capital; Industrial organization; Capital (architecture); Public economics; Sample (material); Economics; Biotechnology; Law","score_opus":0.731170739857711,"score_gpt":0.6596370523479049,"score_spread":0.07153368750980604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390095083","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9906028,0.00029607624,0.000048810216,0.004342723,0.00006497754,0.00023281688,0.0000047773665,0.000019875713,0.004387129],"genre_scores_gemma":[0.9881135,0.0007431098,0.000057875717,0.000049525763,0.00010437021,0.000009767534,6.6081753e-7,0.000009711581,0.010911461],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.98572725,0.0007331829,0.0005900731,0.0005164858,0.011399623,0.0010333603],"domain_scores_gemma":[0.9628907,0.032869842,0.00011034965,0.00078407663,0.0025962398,0.0007488288],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.05187302,0.000106582534,0.0001733825,0.04814566,0.0007379619,0.0017117886,0.002580343,0.00004602405,0.00006610871],"category_scores_gemma":[0.20017387,0.00006046615,0.00005600177,0.13163185,0.0005189168,0.00021292114,0.002909349,0.0005229724,0.0017853465],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000060586266,0.000051766132,0.0056198887,0.000010669945,0.000024218836,0.00001888205,0.0121386,0.00022694751,0.0011045067,0.0073430277,0.118251346,0.85514957],"study_design_scores_gemma":[0.0010497461,0.00065674685,0.5703473,0.00013525215,0.0000016922694,0.000016097272,0.017733408,0.0032478806,0.020172171,0.055237614,0.33099216,0.00040992367],"about_ca_topic_score_codex":0.001335142,"about_ca_topic_score_gemma":0.00011633337,"teacher_disagreement_score":0.85473967,"about_ca_system_score_codex":0.00023421075,"about_ca_system_score_gemma":0.0015304761,"threshold_uncertainty_score":0.9993245},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4390108069","doi":"10.47974/cjsim-2022-0006","title":"Russian Publications in Web of Science: A bibliometric study","year":2023,"lang":"en","type":"article","venue":"Collnet Journal of Scientometrics and Information Management","topic":"Scientific Research and Philosophical Inquiry","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Web of science; Library science; Citation; Citation analysis; Regional science; Citation impact; Visibility; Political science; Scopus; Geography; Computer science; MEDLINE; Law","score_opus":0.07717447785768403,"score_gpt":0.36760575724460465,"score_spread":0.2904312793869206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390108069","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9402288,0.00020508585,0.024809299,0.0033245513,0.0013794819,0.0009306196,0.000008163928,0.000046090336,0.029067893],"genre_scores_gemma":[0.9972625,0.00043222946,0.0021462701,0.000053586315,0.000012850634,0.000007120312,9.675791e-7,0.0000013366321,0.000083190025],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9961527,0.000042102376,0.00077905215,0.00015014196,0.0025639802,0.00031201378],"domain_scores_gemma":[0.9980725,0.0001267892,0.00041528948,0.00032798277,0.00079654896,0.00026087117],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.012142055,0.00007345153,0.00016048194,0.254007,0.00020564832,0.0011382467,0.0016837426,0.000021154154,0.000010440623],"category_scores_gemma":[0.00094763585,0.000056169978,0.000039175557,0.5089724,0.0003407869,0.0069096787,0.00076808303,0.00012686424,0.000019768671],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000023603225,0.00076686655,0.04339523,0.00015923659,0.00005552427,0.000022572463,0.0045766896,0.0012195855,0.000111309106,0.64220124,0.010478001,0.29699016],"study_design_scores_gemma":[0.0017045148,0.0006452956,0.88496524,0.00003953985,0.000007738616,0.00001018434,0.0020858014,0.07831099,0.000083841114,0.0067335893,0.025260048,0.00015322873],"about_ca_topic_score_codex":0.0000024591238,"about_ca_topic_score_gemma":4.9164584e-7,"teacher_disagreement_score":0.84157,"about_ca_system_score_codex":0.00008832925,"about_ca_system_score_gemma":0.00026424567,"threshold_uncertainty_score":0.9998987},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4390126519","doi":"10.18280/isi.280606","title":"A Bibliometric Analysis of the Evolution of IoT Applications in Smart Agriculture","year":2023,"lang":"fr","type":"article","venue":"Ingénierie des systèmes d information","topic":"Smart Agriculture and AI","field":"Agricultural and Biological Sciences","cited_by":14,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Internet of Things; Agriculture; Data science; Computer science; Geography; World Wide Web; Archaeology","score_opus":0.014140387557704694,"score_gpt":0.22387238785602703,"score_spread":0.20973200029832234,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390126519","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99412733,0.0019069269,0.00013667541,0.0005946363,0.00031322966,0.00056037557,0.0002703409,0.0000428016,0.0020476948],"genre_scores_gemma":[0.9985334,0.00025378884,0.000032895234,0.000030636278,0.00008348143,0.00006625824,0.00035006902,9.2627516e-7,0.00064852176],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9981472,0.00012404342,0.0008597203,0.00014328002,0.0004243482,0.00030139857],"domain_scores_gemma":[0.99849635,0.00023081864,0.00064003724,0.00013069983,0.00044631376,0.000055765882],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00054604415,0.0001725363,0.00038177337,0.008389128,0.00020054661,0.000078767305,0.000387559,0.0002544297,0.00010941221],"category_scores_gemma":[0.00022226431,0.000067179084,0.0003504842,0.33754367,0.00021911213,0.000833327,0.000120930534,0.0001443312,0.000047551588],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019158924,0.00021752666,0.8155663,0.00053436845,0.00047841464,4.117817e-7,0.0040742205,0.008333243,0.013421169,0.014865194,0.0048409114,0.1376491],"study_design_scores_gemma":[0.00009193631,0.00005693116,0.9783068,0.00013595787,0.00027186808,0.0000016373176,0.005705463,0.0013296941,0.0009657646,0.0008503535,0.01214999,0.00013363519],"about_ca_topic_score_codex":0.004164563,"about_ca_topic_score_gemma":0.0020348844,"teacher_disagreement_score":0.32915452,"about_ca_system_score_codex":0.0002651916,"about_ca_system_score_gemma":0.000039103226,"threshold_uncertainty_score":0.74855316},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4390202740","doi":"10.1177/21582440231219658","title":"Global Research Pattern of Cognitive Distortion: A Bibliometric Analysis","year":2023,"lang":"en","type":"article","venue":"SAGE Open","topic":"Mental Health Research Topics","field":"Psychology","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Bibliometrics; Cognition; Distortion (music); Mental health; Data science; Field (mathematics); Psychology; Applied psychology; Computer science; Political science; Library science; MEDLINE; Psychiatry","score_opus":0.37253572518129735,"score_gpt":0.6111373012961769,"score_spread":0.2386015761148796,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390202740","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.856505,0.0004659927,0.0007048376,0.00050680496,0.00022502168,0.0011410593,0.0004618279,0.000034941986,0.13995452],"genre_scores_gemma":[0.9860892,0.000084340805,0.000017005023,0.00007292068,0.000056855693,0.0002018939,0.00013907689,0.000009479328,0.013329204],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99727726,0.00078211544,0.00030446993,0.0003722043,0.0006942835,0.0005696736],"domain_scores_gemma":[0.9984394,0.0006115773,0.000075292635,0.00043533285,0.00023590717,0.00020251921],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0031880431,0.00007614045,0.00025592212,0.030059438,0.00014356228,0.00007436405,0.0007310519,0.0000792352,0.009212691],"category_scores_gemma":[0.00039249583,0.00007202983,0.00007987624,0.31096333,0.00014467299,0.00011015179,0.00080432475,0.000201364,0.002265276],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011127205,0.00017342418,0.4980534,0.000040651725,0.00032959707,0.00011486541,0.0006803185,5.941431e-7,0.0000021286075,0.00018782061,0.0101357335,0.49017018],"study_design_scores_gemma":[0.00060869247,0.00024133359,0.9948359,0.00004145194,0.000042245967,0.0000015342379,0.0017214319,0.00003289757,0.000016675333,0.0005054871,0.0018864452,0.00006590882],"about_ca_topic_score_codex":0.009647868,"about_ca_topic_score_gemma":0.0022571837,"teacher_disagreement_score":0.49678248,"about_ca_system_score_codex":0.0001516826,"about_ca_system_score_gemma":0.0000912706,"threshold_uncertainty_score":0.99851155},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4390236203","doi":"10.18280/ijdne.180630","title":"Global Trends and Collaborations in Dengue Virus Research: A Scientometric and Bibliometric Overview (1872–2019)","year":2023,"lang":"en","type":"article","venue":"International Journal of Design & Nature and Ecodynamics","topic":"Mosquito-borne diseases and control","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Qassim University","keywords":"Scopus; Dengue fever; Citation; Web of science; Bibliometrics; Data science; Autoregressive integrated moving average; Library science; Regional science; Geography; Political science; Operations research; Computer science; MEDLINE; Medicine; Engineering; Time series","score_opus":0.07209162776971244,"score_gpt":0.41043987869438303,"score_spread":0.3383482509246706,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390236203","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9458989,0.048533354,0.00046982578,0.0040417947,0.00046164016,0.00016766024,0.00015550708,0.000012860537,0.00025845756],"genre_scores_gemma":[0.9755621,0.023375895,0.00042504916,0.00026285232,0.00017721817,0.0000041209273,0.000020062824,0.000009027072,0.00016365472],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99828297,0.00009791626,0.00035740787,0.00019854915,0.000840906,0.00022226665],"domain_scores_gemma":[0.99831724,0.00035503274,0.00014037422,0.00008771588,0.00087054475,0.00022906999],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0014441931,0.00011061045,0.00025443645,0.03195583,0.00007178377,0.00021353354,0.00016523602,0.00014761653,0.00001938244],"category_scores_gemma":[0.0009605735,0.00009168849,0.00005449939,0.046120074,0.000081293656,0.00026242805,0.00009161673,0.0003888318,0.0000021800151],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022202632,0.00081808225,0.36335504,0.00018298687,0.00095163175,0.00276638,0.00040724262,0.00055429374,0.0020890473,0.01560716,0.028458843,0.58258903],"study_design_scores_gemma":[0.0029900272,0.00039656516,0.9661392,0.0001987141,0.00009177925,0.0005521613,0.00015119741,0.02285964,0.000008744813,0.0022973707,0.0041977973,0.00011680743],"about_ca_topic_score_codex":0.000032104264,"about_ca_topic_score_gemma":0.00013197168,"teacher_disagreement_score":0.60278416,"about_ca_system_score_codex":0.00015172912,"about_ca_system_score_gemma":0.00022358414,"threshold_uncertainty_score":0.9790161},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4390645995","doi":"10.1097/md.0000000000036670","title":"Bibliometric analysis of lupus nephritis in children from 1999 to 2022: A review","year":2024,"lang":"en","type":"review","venue":"Medicine","topic":"Systemic Lupus Erythematosus Research","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Lupus nephritis; Medicine; Web of science; Systemic lupus erythematosus; Internal medicine; Family medicine; Disease; Meta-analysis","score_opus":0.055871436353945994,"score_gpt":0.40987289773135654,"score_spread":0.35400146137741056,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390645995","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00033273041,0.9926055,0.0000128642805,0.0009284418,0.0005329548,0.004305062,0.0004671401,0.00006483243,0.0007504883],"genre_scores_gemma":[0.000028154736,0.9937289,0.00007772218,0.0004501952,0.0006676911,0.00046823075,0.001792561,0.00012985193,0.0026567148],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9927072,0.00044530912,0.0028277228,0.0011782683,0.0022167217,0.00062477705],"domain_scores_gemma":[0.9960918,0.0007961664,0.0004883851,0.0017933036,0.00026460554,0.0005657784],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0026045868,0.0007189856,0.010711797,0.27262807,0.000019061761,0.0000146605225,0.000718297,0.00042489622,0.0030931998],"category_scores_gemma":[0.0031713143,0.00045343753,0.0012533771,0.5211977,0.00014140396,0.000039082453,0.00035960326,0.001127938,0.00057355646],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010234558,0.000071633236,0.011929536,0.17293154,0.0060698264,0.00030108026,0.00011425905,4.737333e-8,0.0000012109591,0.0000047242165,0.11192943,0.6966365],"study_design_scores_gemma":[0.0005071107,0.00017071118,0.0007318185,0.38907933,0.032551017,0.0018777744,0.000018000854,0.0000041099174,1.269995e-7,0.0000037106274,0.5748491,0.00020717115],"about_ca_topic_score_codex":0.0032183307,"about_ca_topic_score_gemma":0.00013688492,"teacher_disagreement_score":0.6964293,"about_ca_system_score_codex":0.0005459969,"about_ca_system_score_gemma":0.0005460715,"threshold_uncertainty_score":0.99979174},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4390685818","doi":"10.20944/preprints202401.0545.v1","title":"Forestry Applications of Space-borne LiDAR Sensors: A Worldwide Bibliometric Analysis","year":2024,"lang":"en","type":"preprint","venue":"Preprints.org","topic":"Remote Sensing and LiDAR Applications","field":"Environmental Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Lidar; Remote sensing; Geography; Scale (ratio); Environmental science; Cartography","score_opus":0.047567937896465885,"score_gpt":0.3247469181409576,"score_spread":0.27717898024449167,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390685818","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9290499,0.0002498557,0.002925788,0.00089683145,0.00011030037,0.000949492,0.000100603385,0.00026224792,0.06545498],"genre_scores_gemma":[0.9906594,0.0002178943,0.0030611278,0.000032310498,0.0001200277,0.00014238668,0.00010054206,0.00005817131,0.0056081233],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9964517,0.000111627916,0.00077825284,0.0015033206,0.00073865673,0.00041647858],"domain_scores_gemma":[0.9962963,0.00017272084,0.0004775408,0.0027395,0.0000684436,0.00024548598],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0007588251,0.00041240596,0.00065268774,0.018255483,0.00012420071,0.000059121427,0.00092084473,0.00037002735,0.0021528476],"category_scores_gemma":[0.00014164663,0.00041949318,0.0006533863,0.089302234,0.00035176668,0.00005409432,0.0036905,0.0010319464,0.0073947995],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000025188448,0.00056765095,0.8646808,0.00045873498,0.002035413,0.00002006813,0.0009903762,0.09650284,0.015049814,0.0009339364,0.00086001836,0.017875114],"study_design_scores_gemma":[0.00010810598,0.000008429238,0.9446472,0.00008619121,0.0015747356,0.000008584899,0.00008501451,0.004146997,0.016690878,0.010146311,0.021962779,0.0005347906],"about_ca_topic_score_codex":0.0029397612,"about_ca_topic_score_gemma":0.00014862383,"teacher_disagreement_score":0.09235584,"about_ca_system_score_codex":0.00028192968,"about_ca_system_score_gemma":0.000076705706,"threshold_uncertainty_score":0.9998257},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4390699105","doi":"10.3389/fphar.2023.1322319","title":"A bibliometric and visual analysis of the use of ustekinumab in Crohn’s disease using CiteSpace","year":2024,"lang":"en","type":"review","venue":"Frontiers in Pharmacology","topic":"Inflammatory Bowel Disease","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Natural Science Foundation of China","keywords":"Ustekinumab; Bibliometrics; Medicine; Crohn's disease; Disease; Pathology; Library science; Infliximab; Computer science","score_opus":0.0447155069555816,"score_gpt":0.36825373761003904,"score_spread":0.3235382306544574,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390699105","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.17241381,0.8258866,0.000106034924,0.0000048724037,0.0007337243,0.000577073,0.00027094793,0.000002951098,0.0000039973847],"genre_scores_gemma":[0.011044564,0.98829275,0.00026761336,0.00005620022,0.00006345048,0.000061220235,0.00007785787,0.000043041153,0.00009330207],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978531,0.00055403524,0.00066997943,0.0004959284,0.00015382966,0.0002731067],"domain_scores_gemma":[0.99913454,0.000044132357,0.0003810595,0.00027455832,0.000060896236,0.00010478964],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00025764632,0.00030660583,0.0011083656,0.02375079,0.000018885024,0.000016386986,0.0002837757,0.0002949915,0.000015829934],"category_scores_gemma":[0.00015335075,0.00024230639,0.0005375574,0.029232033,0.00026986102,0.00000769651,0.0004095511,0.00025772306,3.6217563e-7],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002119477,0.0014226941,0.35338622,0.068965666,0.023254419,0.0031021652,0.0002316846,0.0062771537,0.0014053663,0.000023015797,0.043968216,0.49584392],"study_design_scores_gemma":[0.0022081798,0.00008954486,0.1567951,0.0035479413,0.06968738,0.00001698204,0.00005763861,0.02831574,0.00011333475,0.000022148188,0.7377806,0.0013654337],"about_ca_topic_score_codex":0.000026346868,"about_ca_topic_score_gemma":0.000020708449,"teacher_disagreement_score":0.6938124,"about_ca_system_score_codex":0.000099318255,"about_ca_system_score_gemma":0.0005389531,"threshold_uncertainty_score":0.9914021},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4390873698","doi":"10.32388/d0pndo","title":"Review of: \"Journal citation reports and the definition of a predatory journal: The case of the Multidisciplinary Digital Publishing Institute (MDPI)\"","year":2024,"lang":"en","type":"peer-review","venue":"","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Multidisciplinary approach; Citation; Publishing; Library science; Computer science; Political science; Sociology; Social science; Law","score_opus":0.54414708913449,"score_gpt":0.5464384669861437,"score_spread":0.0022913778516537864,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390873698","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026604014,0.87934035,0.00049702247,0.091806106,0.006000947,0.0013823779,0.0004649807,0.0000066992093,0.0178411],"genre_scores_gemma":[0.06539551,0.9065642,0.0008164852,0.0018944292,0.00057513185,0.00004584883,0.00007125362,0.000045735982,0.024591425],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9852963,0.0010782524,0.0036847158,0.0005196131,0.009106678,0.00031441165],"domain_scores_gemma":[0.9636744,0.010615121,0.006698289,0.0018782663,0.016853523,0.00028036017],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.10570685,0.00025060383,0.0009973665,0.015923908,0.0005127613,0.005477896,0.0026622363,0.00018511972,0.00025568428],"category_scores_gemma":[0.21508695,0.000085178275,0.0008960206,0.08535383,0.0011991805,0.002650707,0.0021968968,0.001732097,0.0000043101454],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011753959,0.000053282,0.00014215444,0.0054046456,0.000096458636,0.00028753327,0.00018606354,0.0000025516172,0.0000015609804,0.0011462572,0.8948493,0.097818464],"study_design_scores_gemma":[0.00052782387,0.0001487414,0.0005690987,0.051099833,0.000506516,0.03501027,0.0011275087,0.00031835106,0.000010277242,0.085216925,0.82520014,0.0002645349],"about_ca_topic_score_codex":0.0001084474,"about_ca_topic_score_gemma":0.000050462033,"teacher_disagreement_score":0.1093801,"about_ca_system_score_codex":0.00009136843,"about_ca_system_score_gemma":0.0015184697,"threshold_uncertainty_score":0.9955545},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","research_integrity","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4391012316","doi":"10.5267/j.ijdns.2024.1.010","title":"Bibliometric analysis of Indonesia's labor dynamics: Future works, digital transformations, and contemporary employment landscape shifts","year":2024,"lang":"en","type":"article","venue":"International Journal of Data and Network Science","topic":"COVID-19 Pandemic Impacts","field":"Economics, Econometrics and Finance","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Poverty; Government (linguistics); Informal sector; Scopus; Globalization; Digital transformation; Population; Human capital; Economic growth; Political science; Sociology; Economics","score_opus":0.035281742832049726,"score_gpt":0.30110041403185817,"score_spread":0.26581867119980845,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391012316","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.925748,0.0445424,0.022510739,0.0034471124,0.0011213323,0.000074590855,0.0020959766,0.000010389524,0.00044945203],"genre_scores_gemma":[0.99311846,0.0060563614,0.0003071267,0.00016109186,0.00025310466,5.187477e-7,0.00008842286,0.0000048760016,0.000010062313],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99874794,0.000007815995,0.000655876,0.00022854036,0.00021899957,0.0001408492],"domain_scores_gemma":[0.99903697,0.00020049459,0.00034553037,0.0001755333,0.00013163318,0.00010981775],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0014648285,0.000087560766,0.00028404794,0.02033581,0.00005322116,0.00067038496,0.0007507632,0.00004115958,0.000024207166],"category_scores_gemma":[0.00011887068,0.00007773625,0.000056394365,0.027714089,0.00013840181,0.0036148047,0.00018235209,0.00014235187,0.000001949458],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005854636,0.000049185634,0.8755283,0.000022877974,0.0008385621,0.000023463454,0.00050033577,0.0006150898,0.0000040397367,0.009958164,0.002227788,0.11017368],"study_design_scores_gemma":[0.00058110565,0.00009437375,0.77761775,0.00020133716,0.00009701309,0.000048322236,0.00024153569,0.19842477,0.0000026606476,0.0030911071,0.01941136,0.00018866149],"about_ca_topic_score_codex":0.000018311186,"about_ca_topic_score_gemma":0.000013352116,"teacher_disagreement_score":0.19780968,"about_ca_system_score_codex":0.000065402724,"about_ca_system_score_gemma":0.00015108816,"threshold_uncertainty_score":0.99295235},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4391175495","doi":"10.1007/s11192-024-04930-7","title":"Exploring relationships among eminent psychologists using co-occurrence analysis","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"Mental Health Research Topics","field":"Psychology","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Brock University","funders":"","keywords":"Psychology; Sociology; Data science; Computer science","score_opus":0.7976071504131107,"score_gpt":0.5730943597756223,"score_spread":0.2245127906374884,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391175495","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97463334,0.0015079525,0.009898913,0.00014086674,0.0033893678,0.0002958656,0.00013734693,0.00018116664,0.009815179],"genre_scores_gemma":[0.9968381,0.00009635101,0.001244376,0.000039226758,0.00013172405,0.0000604126,0.00007643272,0.000013340432,0.0015000475],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9966876,0.00029373335,0.000432894,0.00072139635,0.0011359481,0.000728425],"domain_scores_gemma":[0.99809766,0.0006726662,0.00007980044,0.00061340065,0.00012129664,0.0004151496],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.004063511,0.00015137512,0.00020993488,0.010053158,0.00032637076,0.0002589217,0.00043688747,0.000100307334,0.0010217414],"category_scores_gemma":[0.00092117564,0.00014300652,0.00014622704,0.04536474,0.00023003835,0.00043715828,0.00009798823,0.00066334475,0.0006957272],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000137270235,0.000188624,0.95384127,0.00013291002,0.00025234904,0.00018283584,0.0031880417,0.0002516458,0.000033292243,0.0048031057,0.003725286,0.033386923],"study_design_scores_gemma":[0.00018792531,0.000073138435,0.9709216,0.000054268243,0.00014526774,0.000013694206,0.0007894354,0.011449455,0.000057188357,0.00014722385,0.015899353,0.00026148843],"about_ca_topic_score_codex":0.0002012554,"about_ca_topic_score_gemma":0.000023863551,"teacher_disagreement_score":0.035311583,"about_ca_system_score_codex":0.00045954873,"about_ca_system_score_gemma":0.00008063863,"threshold_uncertainty_score":0.99989146},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4391238844","doi":"10.1007/s44268-024-00025-7","title":"Unveiling global research trends in construction productivity: a scientometric analysis of twenty-first century research","year":2024,"lang":"en","type":"article","venue":"Smart Construction and Sustainable Cities","topic":"Construction Project Management and Performance","field":"Decision Sciences","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Productivity; Regional science; Scientometrics; Economic geography; Geography; Sociology; Economics; Social science; Economic growth","score_opus":0.12880946874650467,"score_gpt":0.4424689900351253,"score_spread":0.3136595212886206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391238844","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95599717,0.0022123195,0.00047919928,0.0010975096,0.0013285599,0.00036656612,0.000030580974,0.00008267168,0.03840543],"genre_scores_gemma":[0.9914672,0.00047904538,0.00042538348,0.000006154284,0.00011056337,0.000040293737,0.000012624569,0.000009130784,0.0074496376],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.9937526,0.0008611085,0.0009211426,0.0010753238,0.0024129453,0.0009768566],"domain_scores_gemma":[0.9960342,0.0010215594,0.00013767043,0.00057308085,0.002094114,0.0001393567],"candidate_categories":["bibliometrics","sts","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.019072823,0.0002030367,0.00052237965,0.037986163,0.0010244722,0.001377644,0.00040426772,0.0001399699,0.0009783214],"category_scores_gemma":[0.0015896114,0.00018013868,0.00018002071,0.10128313,0.0029309508,0.001490074,0.00052827544,0.0005941647,0.000019041117],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008480827,0.000031172327,0.3466078,0.00016486517,0.00015882174,0.000022108823,0.0013037253,0.00017598206,0.000010294764,0.29907122,0.00093746994,0.35143176],"study_design_scores_gemma":[0.00097456056,0.00028501748,0.19265187,0.00017895918,0.00025079792,0.00017161472,0.3538111,0.021336276,0.0002772748,0.07739319,0.35204726,0.00062210334],"about_ca_topic_score_codex":0.0012382484,"about_ca_topic_score_gemma":0.0008105008,"teacher_disagreement_score":0.35250735,"about_ca_system_score_codex":0.00055624166,"about_ca_system_score_gemma":0.00043065072,"threshold_uncertainty_score":0.9999349},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4391239537","doi":"10.1101/2024.01.24.577031","title":"The Increasing Authorship trend in Neuroscience: A scientometric analysis across 11 countries","year":2024,"lang":"en","type":"preprint","venue":"bioRxiv (Cold Spring Harbor Laboratory)","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scientometrics; Bibliometrics; Neuroscience; Psychology; Data science; Library science; Social science; Sociology; Computer science","score_opus":0.24078796745115016,"score_gpt":0.47352454809833866,"score_spread":0.2327365806471885,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391239537","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9858526,0.0066298586,0.0005826178,0.0015532627,0.0037864754,0.0006638936,0.00069262396,0.00020075114,0.000037938302],"genre_scores_gemma":[0.9980666,0.0007543114,0.00045989206,0.00015798879,0.00025917654,0.00012563943,2.2997439e-7,0.000072054674,0.00010412995],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9737198,0.0015599574,0.0022421402,0.0038133408,0.015925042,0.002739756],"domain_scores_gemma":[0.9813561,0.008324931,0.0010728145,0.004163151,0.0038097778,0.0012732577],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","bibliometrics","open_science"],"category_scores_codex":[0.109637626,0.0007737341,0.0012635675,0.1553535,0.0015262954,0.034042485,0.0090827625,0.00075850147,0.000068725756],"category_scores_gemma":[0.087603614,0.00052636146,0.00075805763,0.622062,0.0014432286,0.00062786107,0.009024345,0.002436955,0.0002642817],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006797856,0.00027809775,0.97652084,0.00012525622,0.00021504174,0.00035698037,0.00015131505,0.00086671876,0.0153333545,0.0037411842,0.0022006372,0.00014261731],"study_design_scores_gemma":[0.00028584505,0.000050917017,0.9469934,0.00009273256,0.00015608997,6.5080854e-8,0.000047900885,0.020466443,0.002779227,0.00019339322,0.028132629,0.0008013268],"about_ca_topic_score_codex":0.00069197104,"about_ca_topic_score_gemma":0.00055444933,"teacher_disagreement_score":0.46670854,"about_ca_system_score_codex":0.0010754148,"about_ca_system_score_gemma":0.0019966816,"threshold_uncertainty_score":0.99986446},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4391292468","doi":"10.5152/addicta.2023.23148","title":"Bibliometric Analysis of Studies on Smoking Cessation in the Field of Nursing","year":2023,"lang":"en","type":"article","venue":"Addicta The Turkish Journal on Addictions","topic":"Smoking Behavior and Cessation","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Smoking cessation; Field (mathematics); Medicine; Nursing; Psychology; Mathematics","score_opus":0.09739566223375475,"score_gpt":0.4191120548892257,"score_spread":0.321716392655471,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391292468","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99331534,0.00019089037,0.00004195492,0.004504446,0.00045921575,0.00013859458,0.000021217804,0.000023634371,0.0013046951],"genre_scores_gemma":[0.99862367,0.0007100954,0.0000060939296,0.00044177,0.0001133371,0.000012686128,0.00002955521,0.000009294123,0.000053497904],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.998214,0.00018449739,0.00052707817,0.00012039396,0.00079759804,0.00015641877],"domain_scores_gemma":[0.99735314,0.0017080442,0.000385661,0.00029154168,0.0002263843,0.000035226458],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0014329564,0.00010395343,0.00029808233,0.037722275,0.00024735887,0.00002829343,0.0001764515,0.00007147334,0.00013163938],"category_scores_gemma":[0.0009548171,0.00005860227,0.00024408671,0.09451013,0.00008400547,0.00008624307,0.000012639261,0.00053665903,0.0000022889162],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00048649777,0.0016195615,0.77859575,0.000045045974,0.0051576677,0.00008305266,0.021976946,0.015209623,0.0011018912,0.0030245297,0.04590811,0.12679133],"study_design_scores_gemma":[0.0004800181,0.00045029627,0.98785895,0.00047129783,0.002413893,0.000029596189,0.0066229133,0.0004055824,0.0007107555,0.0003023132,0.00020179499,0.000052573465],"about_ca_topic_score_codex":0.00002086871,"about_ca_topic_score_gemma":0.000007791461,"teacher_disagreement_score":0.20926322,"about_ca_system_score_codex":0.00011061402,"about_ca_system_score_gemma":0.000042387597,"threshold_uncertainty_score":0.97318435},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4391311615","doi":"10.1038/s41598-024-52233-x","title":"Impact of medical technologies may be predicted using constructed graph bibliometrics","year":2024,"lang":"en","type":"article","venue":"Scientific Reports","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Bibliometrics; Computer science; Random forest; Citation; Data science; Pipeline (software); Machine learning; Data mining; Artificial intelligence; World Wide Web","score_opus":0.47367089176192456,"score_gpt":0.5817619050905887,"score_spread":0.10809101332866411,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391311615","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97705024,0.0034444998,0.005687648,0.0005526815,0.011001835,0.00036155994,0.00012999272,0.00036880263,0.0014027391],"genre_scores_gemma":[0.9978387,0.00019864769,0.0013297948,0.000005282988,0.000043670814,0.0000063222997,0.00002487353,0.000019743484,0.00053297303],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9639857,0.00020250262,0.0021056877,0.0019503529,0.03079354,0.0009622454],"domain_scores_gemma":[0.9875954,0.0032435649,0.0006678366,0.0024134442,0.0054384894,0.0006412715],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.056235865,0.00027310435,0.0006058931,0.4201764,0.00039925735,0.008424798,0.002976054,0.00047960167,0.0025639671],"category_scores_gemma":[0.11406645,0.00016952238,0.0006111692,0.7729387,0.002730061,0.001081087,0.0016412901,0.0006070742,0.00002629715],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017146582,0.0003417586,0.5224246,0.00007139412,0.00021866268,0.003301613,0.0001891188,0.00020330083,0.01763592,0.0011742245,0.2510716,0.20335068],"study_design_scores_gemma":[0.0013827758,0.00134495,0.1894015,0.00082794734,0.00020788549,0.010473404,0.0044822143,0.30683732,0.05690463,0.28932557,0.1363468,0.0024649843],"about_ca_topic_score_codex":0.00022615244,"about_ca_topic_score_gemma":0.000006747247,"teacher_disagreement_score":0.35276237,"about_ca_system_score_codex":0.0002117057,"about_ca_system_score_gemma":0.0030328706,"threshold_uncertainty_score":0.99998397},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4391581474","doi":"10.1007/978-3-031-47215-2_20","title":"Role of Nanoparticles in the Suppression of Diseases in Fruits and Vegetables to Improve Agricultural and Environmental Sustainability: A Bibliometric Analysis","year":2024,"lang":"en","type":"book-chapter","venue":"Green energy and technology","topic":"Insect Pest Control Strategies","field":"Agricultural and Biological Sciences","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Agriculture; Sustainability; Food security; Population; Biotechnology; Food safety; Food supply; Business; Toxicology; Agricultural science; Geography; Environmental health; Environmental science; Medicine; Food science; Biology","score_opus":0.0038265509936248002,"score_gpt":0.17858604045729956,"score_spread":0.17475948946367476,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391581474","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.985697,0.012813425,4.7645962e-8,0.00072924607,0.000005917622,0.00012687394,0.000105577485,0.00001082638,0.00051111606],"genre_scores_gemma":[0.99851376,0.0006984456,0.0000024176713,0.000009235282,0.00001186177,0.00002230513,0.000022536215,9.1957014e-7,0.0007185102],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.999222,0.000021025822,0.00021880833,0.00028533925,0.000103814025,0.00014900784],"domain_scores_gemma":[0.99968904,0.00012664768,0.00007562307,0.000061865336,0.000019784513,0.000027008977],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00008173487,0.0001448495,0.0003364924,0.0027681468,0.00003214789,0.000018582596,0.00012962137,0.00022244143,0.000008129971],"category_scores_gemma":[0.000020015632,0.000052152925,0.000040581657,0.0049537173,0.00025679727,0.00004914312,0.00019209784,0.00009389375,8.211438e-8],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010832057,0.00014103111,0.09654692,0.00013587323,0.00029066415,0.00002974621,0.00024013595,0.000013808021,0.35450622,0.16539909,0.000007818242,0.38258037],"study_design_scores_gemma":[0.00017498585,0.0005947378,0.857354,0.000059094255,0.00030006867,0.000009480196,0.002190631,0.00004471685,0.004174527,0.13370004,0.0011778295,0.00021989795],"about_ca_topic_score_codex":0.0015263043,"about_ca_topic_score_gemma":0.006324812,"teacher_disagreement_score":0.76080704,"about_ca_system_score_codex":0.0000138707055,"about_ca_system_score_gemma":0.0000046896353,"threshold_uncertainty_score":0.35293934},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4391744040","doi":"10.3390/healthcare12040463","title":"Adaptive Shooting Disciplines: A Scoping Review of the Literature with Bibliometric Analysis","year":2024,"lang":"en","type":"review","venue":"Healthcare","topic":"Sports injuries and prevention","field":"Medicine","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Scopus; Excellence; Computer science; Web of science; Applied psychology; Data science; Psychology; MEDLINE; Engineering ethics; Management science; Political science; Engineering","score_opus":0.08190370357073254,"score_gpt":0.4615035571650311,"score_spread":0.3795998535942986,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391744040","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000044236886,0.99630225,0.000014934074,0.0006168284,0.00020381024,0.002546833,0.00014347352,0.00003273938,0.00013469411],"genre_scores_gemma":[0.00006344634,0.998299,0.0002051989,0.00024869668,0.00024968974,0.0001351824,0.00035708334,0.000047315174,0.00039439838],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.99754065,0.00019091851,0.001027253,0.0004583576,0.0005491212,0.00023370938],"domain_scores_gemma":[0.997906,0.00009330283,0.0007765738,0.00072896195,0.00038290498,0.000112225454],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0006668549,0.00033793287,0.002611744,0.019720236,0.00008406507,0.000030312065,0.00016209962,0.00022216477,0.000070457965],"category_scores_gemma":[0.00012121378,0.00015249146,0.0013674861,0.23146634,0.000043249674,0.000044162436,0.00011626807,0.000768012,0.00000477862],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000034060852,0.0000086691325,0.0002125523,0.62414193,0.0005236468,0.000015485997,0.000055090895,4.9686385e-8,7.6209744e-10,0.00002667835,0.00015003138,0.37486243],"study_design_scores_gemma":[0.00004165601,0.00011353911,0.00022504418,0.92683464,0.01916127,0.00007613022,0.000038401067,0.0000015168256,4.200252e-8,0.0000065281065,0.05337315,0.00012806535],"about_ca_topic_score_codex":0.000049927676,"about_ca_topic_score_gemma":0.00009594609,"teacher_disagreement_score":0.37473437,"about_ca_system_score_codex":0.00011519098,"about_ca_system_score_gemma":0.0010314824,"threshold_uncertainty_score":0.9913904},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4391920904","doi":"10.1016/j.teler.2024.100124","title":"Understanding 21st century skills needed in response to industry 4.0: Exploring scholarly insights using bibliometric analysis","year":2024,"lang":"en","type":"article","venue":"Telematics and Informatics Reports","topic":"Entrepreneurship Studies and Influences","field":"Business, Management and Accounting","cited_by":26,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"Bibliometrics; Knowledge management; Data science; Engineering ethics; Sociology; Computer science; Engineering; Library science","score_opus":0.10779616540913634,"score_gpt":0.28845329906388906,"score_spread":0.18065713365475272,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391920904","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9939779,0.0005946831,0.00084561895,0.00017407679,0.00044658425,0.0002963498,0.000001633912,0.000111766676,0.0035513844],"genre_scores_gemma":[0.9979037,0.0004072157,0.0009095909,0.0005795842,0.00013305034,0.00001860496,0.00000695365,0.000021374444,0.000019932539],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99750537,0.000016276492,0.0012746109,0.00023345002,0.0005767912,0.0003935143],"domain_scores_gemma":[0.99891335,0.00019638521,0.00036731706,0.0003222786,0.00014475838,0.000055897897],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0015195305,0.00025717425,0.00044865298,0.048034213,0.00024907666,0.0028874795,0.00013635159,0.00014426235,0.000038687846],"category_scores_gemma":[0.0008374155,0.00021771813,0.00010623577,0.07692788,0.00004341811,0.005745097,0.00034593532,0.00040926764,0.00001441404],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011490182,0.00028753525,0.8945727,0.00521477,0.0017046888,0.0010447646,0.029808953,0.030823406,0.0004977958,0.020659253,0.0012384813,0.014032754],"study_design_scores_gemma":[0.0014111816,0.00017743623,0.41342416,0.007298252,0.0030537185,0.00024155459,0.25689328,0.23615626,0.0002627596,0.015525137,0.061612774,0.0039434694],"about_ca_topic_score_codex":0.00021252477,"about_ca_topic_score_gemma":0.000034670014,"teacher_disagreement_score":0.4811485,"about_ca_system_score_codex":0.000166159,"about_ca_system_score_gemma":0.000043117445,"threshold_uncertainty_score":0.9981476},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4392121664","doi":"10.3390/jrfm17030094","title":"Bibliometric Framing of Research Trends Regarding Public Sector Auditing to Fight Corruption and Prevent Fraud","year":2024,"lang":"en","type":"article","venue":"Journal of risk and financial management","topic":"Corruption and Economic Development","field":"Social Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Language change; Audit; Public sector; Framing (construction); Accounting; Analytics; Business; Public institution; Public relations; Political science; Law; Engineering; Computer science","score_opus":0.05778465782098792,"score_gpt":0.350770709879381,"score_spread":0.2929860520583931,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392121664","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97206885,0.002766295,0.010232121,0.0014964825,0.0014412872,0.00018045078,0.0000068332756,0.000019335714,0.011788347],"genre_scores_gemma":[0.9887657,0.008125078,0.0011284634,0.00003293218,0.00039238165,0.000004720224,4.8595643e-7,0.000006011111,0.0015442333],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99867636,0.00011592524,0.00038618376,0.00015744794,0.000424913,0.00023915718],"domain_scores_gemma":[0.9993238,0.00018265125,0.00014899613,0.000057045396,0.00013907782,0.0001484253],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0045068585,0.00006840684,0.00017211503,0.016688347,0.0003318279,0.000263577,0.000116111674,0.00005065949,0.00013070862],"category_scores_gemma":[0.00047627723,0.000062260486,0.000056398225,0.011884805,0.000058868147,0.00028815295,0.00012752559,0.00021481748,0.0000061739497],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016769234,0.000026043423,0.005754679,0.0000681221,0.000016774859,0.000017078555,0.004462338,0.000007466298,0.000008013,0.008302385,0.0019297572,0.97939056],"study_design_scores_gemma":[0.00022124156,0.00011458539,0.19843362,0.00034257927,0.000025080884,0.0000043539985,0.0034423969,0.00007902752,0.000007851361,0.0010152406,0.79621613,0.00009789462],"about_ca_topic_score_codex":0.000091010486,"about_ca_topic_score_gemma":0.00013561349,"teacher_disagreement_score":0.9792927,"about_ca_system_score_codex":0.00018148386,"about_ca_system_score_gemma":0.00008616808,"threshold_uncertainty_score":0.99445665},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4392352987","doi":"10.1007/s41347-024-00397-8","title":"Role of AI/ML in the Study of Autism Spectrum Disorders: A Bibliometric Analysis","year":2024,"lang":"en","type":"article","venue":"Journal of Technology in Behavioral Science","topic":"Autism Spectrum Disorder Research","field":"Neuroscience","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Montreal Neurological Institute and Hospital","funders":"","keywords":"Autism; Psychology; Spectrum (functional analysis); Clinical psychology; Developmental psychology; Physics","score_opus":0.03363858509691659,"score_gpt":0.3840502652647294,"score_spread":0.3504116801678128,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392352987","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98841023,0.000808454,0.000036324363,0.01023168,0.00008527888,0.00033692096,0.0000036273075,0.000020837555,0.00006663737],"genre_scores_gemma":[0.9996974,0.00019449224,0.00006898032,0.000010461004,0.0000039325428,0.000011721098,4.3520117e-8,0.0000076533615,0.0000053395947],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99644476,0.00016533092,0.0008624351,0.00041826122,0.0015958017,0.0005133989],"domain_scores_gemma":[0.99892133,0.00017259122,0.00032186692,0.0005112409,0.00002287177,0.000050067894],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.002474988,0.00014110624,0.00045318992,0.1507983,0.00008498127,0.00008929902,0.0028171958,0.00010827491,0.000042017815],"category_scores_gemma":[0.00036687782,0.00009664673,0.00013971287,0.39485598,0.0013487664,0.0006467402,0.00037459345,0.0011730988,0.000003214746],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000028781298,0.0032073893,0.83712566,0.000010948621,0.000013242672,0.00025292515,0.002219167,0.00022942855,0.11365101,0.02268,0.0000037208417,0.020577746],"study_design_scores_gemma":[0.00049336324,0.0023809797,0.88958365,0.000050331924,0.00010979001,0.00009827356,0.0030343698,0.0012292266,0.029791461,0.07307563,0.000022525135,0.00013041745],"about_ca_topic_score_codex":0.00070452417,"about_ca_topic_score_gemma":0.0011318189,"teacher_disagreement_score":0.24405767,"about_ca_system_score_codex":0.00016052692,"about_ca_system_score_gemma":0.00030976153,"threshold_uncertainty_score":0.8588267},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4392463913","doi":"10.2139/ssrn.4748213","title":"A Prisma-Driven Bibliometric Analysis of the Scientific Literature on Assurance Case Patterns","year":2024,"lang":"en","type":"preprint","venue":"SSRN Electronic Journal","topic":"Safety Systems Engineering in Autonomy","field":"Engineering","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Management science; Data science; Computer science; Engineering ethics; Engineering","score_opus":0.007424877971936863,"score_gpt":0.22754976659147602,"score_spread":0.22012488861953916,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392463913","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9460037,0.033416837,0.012681448,0.00013198255,0.006483105,0.0004227645,0.00029199137,0.0002640078,0.00030416576],"genre_scores_gemma":[0.99708784,0.0017158701,0.00006625973,0.0000055331225,0.00034427925,0.000031547497,0.000024341482,0.00010573858,0.0006185906],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99629444,0.00010836317,0.0007864093,0.00048970536,0.0007101058,0.0016110003],"domain_scores_gemma":[0.99822825,0.00014090416,0.0002717319,0.0010646576,0.00019875044,0.00009571893],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0018007866,0.0004778825,0.0006943032,0.053285737,0.00013089414,0.0005489682,0.0009637019,0.00037290252,0.000012461817],"category_scores_gemma":[0.000092147544,0.00036143968,0.0009620859,0.085912384,0.000045624463,0.00007214502,0.00032186593,0.007317304,0.00001676044],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004922706,0.000029312925,0.0015971913,0.0008670602,0.007124201,0.00020658679,0.0006195006,0.9820977,0.000253753,0.0020631454,0.00023821538,0.0048983977],"study_design_scores_gemma":[0.00101736,0.00024665383,0.035059884,0.00978802,0.009589534,0.012898292,0.00061384536,0.9050293,0.0014492559,0.016027553,0.0054692836,0.0028110358],"about_ca_topic_score_codex":0.000033307017,"about_ca_topic_score_gemma":0.00035990987,"teacher_disagreement_score":0.07706843,"about_ca_system_score_codex":0.0019813727,"about_ca_system_score_gemma":0.0011176995,"threshold_uncertainty_score":0.9998838},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4392557674","doi":"10.1002/leap.1600","title":"Using automated analysis of the bibliography to detect potential research integrity issues","year":2024,"lang":"en","type":"article","venue":"Learned Publishing","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Honeywell (Canada)","funders":"","keywords":"Workflow; Computer science; Data science; Key (lock); Research integrity; Bibliography; Data integrity; Information retrieval; Database; Library science; Computer security; Engineering ethics","score_opus":0.14494348183679898,"score_gpt":0.4536236169249125,"score_spread":0.30868013508811354,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392557674","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.86804676,0.0022704876,0.004938408,0.082554884,0.0022496982,0.00059625244,0.000082897444,0.0011150121,0.038145598],"genre_scores_gemma":[0.9969181,0.00020800489,0.0016467983,0.00024321898,0.00040416876,0.0000046518962,0.000006250503,0.000012461396,0.0005563453],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9951464,0.0019978464,0.0003101928,0.000379508,0.0016075657,0.00055847806],"domain_scores_gemma":[0.9981339,0.0007403244,0.00006501012,0.00029772095,0.0006130285,0.00015003142],"candidate_categories":["bibliometrics","scholarly_communication","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.0150416,0.00010576225,0.00023876714,0.008951196,0.00094990514,0.0032683264,0.001227143,0.0007767689,0.00035207288],"category_scores_gemma":[0.0055050286,0.00007887266,0.00034948956,0.06600222,0.0003710226,0.0018649213,0.00030252527,0.0042314734,0.000015983353],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011886735,0.00009758708,0.032715235,0.00020812821,0.0039657312,0.000056176224,0.15388416,0.011396741,0.01795819,0.14527695,0.5891402,0.045182064],"study_design_scores_gemma":[0.00024986223,0.00009172768,0.033138733,0.0010134247,0.0017475284,0.0000056732906,0.064571515,0.31245348,0.0040831426,0.056474946,0.52524275,0.00092723884],"about_ca_topic_score_codex":0.10474266,"about_ca_topic_score_gemma":0.005424292,"teacher_disagreement_score":0.3010567,"about_ca_system_score_codex":0.0000986333,"about_ca_system_score_gemma":0.00032207675,"threshold_uncertainty_score":0.9980658},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity","bibliometrics"],"domain":"evaluation","study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","research_integrity"],"domain":"methods","study_design":"observational","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4392657766","doi":"10.3390/f15030513","title":"Unearthing Current Knowledge Gaps in Our Understanding of Tree Stability: Review and Bibliometric Analysis","year":2024,"lang":"en","type":"article","venue":"Forests","topic":"Tree Root and Stability Studies","field":"Engineering","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Teagasc","keywords":"Windthrow; Climate change; Environmental resource management; Diversification (marketing strategy); Forest management; Geography; Ecological stability; Tree (set theory); Agroforestry; Ecology; Forestry; Ecosystem; Environmental science; Business; Biology","score_opus":0.07064291569321643,"score_gpt":0.32701156180183694,"score_spread":0.25636864610862053,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392657766","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.42304045,0.5744339,0.001539841,0.00010229972,0.00017281789,0.00015015504,0.000017173646,0.00010131918,0.0004420281],"genre_scores_gemma":[0.9865723,0.013349756,0.00002759824,8.3525526e-7,0.000017350729,0.0000103394,0.000004415836,0.000011011363,0.0000063795073],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99918985,0.000030383137,0.00030090782,0.00018377387,0.0001160262,0.00017903734],"domain_scores_gemma":[0.99961936,0.00014109403,0.000015004421,0.00015699671,0.000023394408,0.000044164877],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00044510324,0.00012095965,0.00039232065,0.015066666,0.000023239092,0.000029946987,0.00006118306,0.000030955955,0.000009905472],"category_scores_gemma":[0.00010872134,0.00010581446,0.00010893263,0.061255626,0.000023230303,0.0001176607,0.000048055415,0.00014219234,0.0000039689608],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000020064604,0.00003407146,0.9456422,0.012975782,0.0003415836,0.0000035697522,0.001107447,0.00026613212,0.0000120514915,0.0004992059,0.0006969893,0.03841898],"study_design_scores_gemma":[0.00012882538,0.000025522313,0.9820872,0.0012849913,0.0006012528,8.062374e-7,0.00025980812,0.012741969,0.00003912266,0.0010164417,0.001631632,0.0001824496],"about_ca_topic_score_codex":0.000016178485,"about_ca_topic_score_gemma":0.04761617,"teacher_disagreement_score":0.5635319,"about_ca_system_score_codex":0.00012665284,"about_ca_system_score_gemma":0.000013977287,"threshold_uncertainty_score":0.99609673},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4392807255","doi":"10.5005/jp-journals-10024-3641","title":"Overview, Trends, and Collaboration on Immunization, Vaccination, and Immunomodulation Therapies for Periodontitis: A Scientometric Study","year":2024,"lang":"en","type":"article","venue":"The Journal of Contemporary Dental Practice","topic":"Oral microbiology and periodontitis research","field":"Dentistry","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Immunization; Vaccination; Periodontitis; Medicine; Family medicine; MEDLINE; Political science; Immunology; Dentistry","score_opus":0.04107307378348924,"score_gpt":0.3859554589337916,"score_spread":0.34488238515030234,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392807255","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9225155,0.07318894,0.00033262002,0.0023533152,0.000731216,0.0004352753,0.000035688077,0.000014214669,0.00039322517],"genre_scores_gemma":[0.9971669,0.0014576409,0.00008221646,0.00010935367,0.000114560855,0.000007414426,0.000016576403,0.00001505361,0.0010303191],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99839437,0.0007080603,0.00040725528,0.00013416157,0.0002568136,0.00009931098],"domain_scores_gemma":[0.9980289,0.0010051558,0.00029114462,0.00012974402,0.0005078969,0.000037157366],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0027376548,0.00011222514,0.00018176837,0.00069162174,0.00036769913,0.00060580194,0.00015830866,0.000055595865,0.0000632465],"category_scores_gemma":[0.0006287469,0.0000775172,0.000039891886,0.0011660389,0.00006696015,0.0020397697,0.000052818043,0.00024221806,0.0000073270053],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.02220039,0.003854767,0.30804163,0.00084866537,0.005514174,0.000736297,0.06728416,0.00012213916,0.065821216,0.011925266,0.077438444,0.43621284],"study_design_scores_gemma":[0.0036575662,0.0037014736,0.8901335,0.00017026078,0.00026401662,0.002126438,0.032881867,0.0005048817,0.0005460848,0.00032596992,0.06544429,0.00024362406],"about_ca_topic_score_codex":0.00006462698,"about_ca_topic_score_gemma":0.00006658369,"teacher_disagreement_score":0.58209187,"about_ca_system_score_codex":0.00005475719,"about_ca_system_score_gemma":0.00011343479,"threshold_uncertainty_score":0.58417624},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4392914644","doi":"10.1061/9780784485279.057","title":"A Systematic and Bibliometric Review of Greenhouse Gas Emissions in the Construction Phase of Residential Buildings","year":2024,"lang":"en","type":"article","venue":"","topic":"Environmental Impact and Sustainability","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Greenhouse gas; Environmental science; Architectural engineering; Engineering; Geology","score_opus":0.014126245633626675,"score_gpt":0.30748962947981834,"score_spread":0.29336338384619165,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392914644","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9874961,0.010172938,0.00019840426,0.00029708704,0.000023131819,0.000553681,0.000003768401,0.00000958107,0.0012453279],"genre_scores_gemma":[0.9918591,0.0076449467,0.00034769817,0.000052213276,0.0000033261322,0.000012431614,7.380354e-7,0.0000039556526,0.00007562505],"study_design_codex":"systematic_review","study_design_gemma":"observational","domain_scores_codex":[0.998985,0.00012368266,0.00037366585,0.0001326699,0.00028767574,0.00009731368],"domain_scores_gemma":[0.9995476,0.00016217477,0.00006767751,0.00018159821,0.000002915588,0.000038040245],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010512859,0.000068364796,0.00019077687,0.0010857733,0.000024585652,0.000015745789,0.00011229718,0.000025624997,0.00076144206],"category_scores_gemma":[0.0002740833,0.000039541006,0.000050436218,0.009342454,0.00025400895,0.00018926235,0.00008121058,0.000066552355,0.0000055724313],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008818483,0.0018936328,0.16416632,0.6801551,0.00012306184,0.000108135835,0.005832003,0.00003064199,0.06585881,0.0034708213,0.010605627,0.06766764],"study_design_scores_gemma":[0.007795136,0.0051578335,0.50153524,0.33820915,0.0027375568,0.0024591642,0.037898596,0.00849087,0.06172942,0.026254026,0.0051726787,0.0025603329],"about_ca_topic_score_codex":0.000439013,"about_ca_topic_score_gemma":0.000015196635,"teacher_disagreement_score":0.34194595,"about_ca_system_score_codex":0.00006112802,"about_ca_system_score_gemma":0.0000072928074,"threshold_uncertainty_score":0.83372563},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4393240927","doi":"10.5751/es-14694-290133","title":"The evolution of social-ecological systems (SES) research: a co-authorship and co-citation network analysis","year":2024,"lang":"en","type":"article","venue":"Ecology and Society","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Ernest Oppenheimer Memorial Trust; National Research Foundation","keywords":"Social network analysis; Co-citation; Citation; Sociology; Ecology; Environmental resource management; Data science; Geography; Computer science; Social science; Biology; Environmental science; Library science; Social capital","score_opus":0.5661443252570436,"score_gpt":0.5981302566166787,"score_spread":0.03198593135963512,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393240927","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98654616,0.006571216,0.0017280033,0.0018364873,0.00038505276,0.00026110368,0.00002620647,0.000018977282,0.0026267744],"genre_scores_gemma":[0.9969796,0.0008802373,0.00008048448,0.0000250625,0.00012530679,0.000025261576,0.0000047447384,0.00000339405,0.0018758953],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99455154,0.0016601976,0.00047447297,0.0004922523,0.002301639,0.00051988417],"domain_scores_gemma":[0.97498244,0.02351868,0.00010936086,0.00018487529,0.0010663343,0.00013828684],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.070551194,0.00007958334,0.0002915687,0.00524122,0.0016091956,0.0013132321,0.0004645838,0.00032962146,0.00007494586],"category_scores_gemma":[0.008471711,0.000045967816,0.00019453127,0.080743395,0.0012318565,0.000164928,0.00022651693,0.00048897887,0.000029187471],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003899176,0.00009143432,0.5705145,0.000043176587,0.0006309984,0.000008413709,0.0052784598,0.00027410203,0.000113891096,0.19198689,0.22355385,0.007465304],"study_design_scores_gemma":[0.00010522931,0.00015179907,0.89457184,0.0000023883167,0.000039391285,0.0000023183768,0.0054503796,0.054848652,0.0000023752134,0.032160036,0.012606196,0.000059366896],"about_ca_topic_score_codex":0.00005400637,"about_ca_topic_score_gemma":0.00007026542,"teacher_disagreement_score":0.32405737,"about_ca_system_score_codex":0.00017525615,"about_ca_system_score_gemma":0.00019805481,"threshold_uncertainty_score":0.9998804},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4393315117","doi":"10.1177/21925682241243074","title":"Global Research Trends on Gait Rehabilitation in Individuals With Spinal Cord Injury- A Bibliometric Analysis","year":2024,"lang":"en","type":"review","venue":"Global Spine Journal","topic":"Spinal Cord Injury Research","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Rehabilitation; Spinal cord injury; Medicine; Thematic analysis; Physical medicine and rehabilitation; Gait; Bibliometrics; Medical education; Physical therapy; Qualitative research; Library science; Spinal cord; Social science; Computer science; Sociology","score_opus":0.15059498277543534,"score_gpt":0.5496904022283852,"score_spread":0.3990954194529498,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393315117","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006247469,0.9837342,0.00007214155,0.001223188,0.00037046013,0.0012856849,0.0005480671,0.00008259063,0.006436192],"genre_scores_gemma":[0.0025562495,0.9912514,0.0036610353,0.00007442047,0.0011395791,0.00018298814,0.0002098441,0.00008915368,0.00083533325],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9876785,0.0014761442,0.0022201808,0.001452689,0.0053678146,0.0018046497],"domain_scores_gemma":[0.99561423,0.00014119496,0.0006389435,0.0010992842,0.0013577688,0.001148601],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0074992557,0.0010014849,0.0039511668,0.20613967,0.0002515589,0.00079921004,0.0010821795,0.00075667654,0.00069639506],"category_scores_gemma":[0.0023701359,0.0006665361,0.0017909157,0.5998193,0.00041721246,0.00027930678,0.00044879882,0.004664641,0.0005560539],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025181368,0.00058137823,0.012549696,0.00492276,0.002771846,0.0012249275,0.0000035813125,0.000002472063,9.2050875e-8,0.00034004016,0.019251399,0.9558337],"study_design_scores_gemma":[0.001346342,0.07858179,0.18307352,0.030071752,0.009314295,0.0030871937,0.000066181965,0.000016114747,6.840435e-8,0.00048228927,0.69328594,0.00067449675],"about_ca_topic_score_codex":0.00031322325,"about_ca_topic_score_gemma":0.00020968955,"teacher_disagreement_score":0.9551592,"about_ca_system_score_codex":0.008088842,"about_ca_system_score_gemma":0.0031178605,"threshold_uncertainty_score":0.9995786},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4393372233","doi":"10.3386/w32285","title":"Explorations of Cumulative Advantage Using Data on French Physicists","year":2024,"lang":"en","type":"report","venue":"National Bureau of Economic Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Nautical Research Society","funders":"","keywords":"Computer science","score_opus":0.967097118024302,"score_gpt":0.7719073767370955,"score_spread":0.19518974128720645,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393372233","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.027016105,0.0031670618,0.00030665664,0.0009149749,0.0024727674,0.0012151145,0.0081297215,0.000019514813,0.9567581],"genre_scores_gemma":[0.9822636,0.001079427,0.000823025,0.000009225552,0.00074778503,0.000027355925,0.0010343322,0.000049879352,0.013965364],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.966808,0.0005368025,0.0020381408,0.0016945617,0.028290924,0.0006315274],"domain_scores_gemma":[0.9563561,0.01914313,0.00096281106,0.0021930009,0.02108203,0.00026290983],"candidate_categories":["metaresearch","bibliometrics","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.07457629,0.00026616082,0.0008755133,0.08593581,0.00018575316,0.00092256104,0.0058460943,0.0003569501,0.0010544836],"category_scores_gemma":[0.07477647,0.00021328343,0.00028416072,0.050070673,0.00066159497,0.0009582598,0.0032202676,0.001228563,0.0007546177],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000073940195,0.0007471492,0.0022609879,0.00030644718,0.00054671,0.000033486485,0.00023766179,0.0076179495,0.0017333284,0.23866692,0.71250415,0.03527127],"study_design_scores_gemma":[0.00041152915,0.0003638279,0.0012807474,0.00034187696,0.000028686409,0.0000108652885,0.00037717275,0.08759011,0.001629536,0.8513924,0.05615374,0.0004194931],"about_ca_topic_score_codex":0.0015254465,"about_ca_topic_score_gemma":0.000117554555,"teacher_disagreement_score":0.9552475,"about_ca_system_score_codex":0.0019419811,"about_ca_system_score_gemma":0.011363537,"threshold_uncertainty_score":0.9998587},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4393376287","doi":"10.29173/jchla29696","title":"Bibliometric Analysis of Librarian Involvement in Systematic Reviews at the University of Alberta","year":2024,"lang":"en","type":"article","venue":"Journal of the Canadian Health Libraries Association / Journal de l Association de bilbiothèques de la santé du Canada","topic":"Health Sciences Research and Education","field":"Health Professions","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"Alberta Library; University of Alberta","funders":"Canadian Institutes of Health Research; University of Alberta","keywords":"Systematic review; Acknowledgement; Library science; Web of science; Bibliometrics; MEDLINE; Medicine; Meta-analysis; Political science; Computer science","score_opus":0.03094961789980141,"score_gpt":0.34349786434803914,"score_spread":0.31254824644823775,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393376287","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7361216,0.16144893,0.00018415657,0.09461308,0.0033296787,0.002010066,0.0003222933,0.000012603065,0.0019575732],"genre_scores_gemma":[0.93395305,0.056182753,0.00029277065,0.005891662,0.0005314082,0.000008129645,0.000012601648,0.00002051434,0.0031071329],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9852189,0.010464195,0.0019886494,0.0001337141,0.0012423197,0.00095222093],"domain_scores_gemma":[0.97950953,0.01395114,0.0047129416,0.00022924296,0.000702875,0.00089424616],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.029667204,0.00014366144,0.00083896506,0.00951653,0.0012479991,0.00022377027,0.0007681553,0.00028735172,0.0005943251],"category_scores_gemma":[0.018459456,0.00009801497,0.0003318116,0.032356624,0.000057816396,0.0007414709,0.00007120574,0.0013506855,0.0000029119087],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000014776886,0.000018204271,0.5732476,0.0018813105,0.00038834134,0.000010147831,0.010645087,0.00024813207,0.000003873577,0.00031199554,0.41301355,0.00021698959],"study_design_scores_gemma":[0.00035011134,0.00006978324,0.72965753,0.0029061148,0.00030253502,0.000019933183,0.005338927,0.0019862514,0.0000067394217,0.0005138867,0.25873238,0.00011577094],"about_ca_topic_score_codex":0.7287453,"about_ca_topic_score_gemma":0.95596844,"teacher_disagreement_score":0.22722314,"about_ca_system_score_codex":0.041141514,"about_ca_system_score_gemma":0.07652961,"threshold_uncertainty_score":0.9991618},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W4393558576","doi":"10.5281/zenodo.3826037","title":"Open Access of COVID-19 related publications in the first quarter of 2020: a preliminary study based in PubMed","year":2020,"lang":"en","type":"dataset","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Coronavirus disease 2019 (COVID-19); Quarter (Canadian coin); Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); 2019-20 coronavirus outbreak; Virology; History; Medicine; Infectious disease (medical specialty)","score_opus":0.2023763604307022,"score_gpt":0.4203691591748092,"score_spread":0.217992798744107,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393558576","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0024302867,0.00005751791,0.000092630005,0.027244078,0.00014827472,0.0053145294,0.95389473,0.0000970665,0.0107208565],"genre_scores_gemma":[0.2110494,0.00003866496,0.000013772535,0.0013560194,0.000046576,0.0000034812883,0.78692126,0.00035342344,0.00021739809],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9914386,0.0036906337,0.0015791695,0.0009844801,0.0019246954,0.0003824071],"domain_scores_gemma":[0.9945707,0.0013107491,0.0012138119,0.0018333996,0.00081760035,0.0002537913],"candidate_categories":["metaresearch","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.01791987,0.0002334817,0.0005866844,0.0014203305,0.0007572228,0.007188517,0.035260946,0.00023257989,0.009860677],"category_scores_gemma":[0.06043715,0.00017458505,0.00007604608,0.008641908,0.00028222965,0.002299774,0.011966516,0.0012089644,0.0009564131],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002512527,0.00071040064,0.000663527,0.00006225192,0.000024474444,0.000018944516,0.0021609177,0.00019498495,3.153666e-7,0.00004368704,0.99318236,0.0026868687],"study_design_scores_gemma":[0.0011000473,0.00026334278,0.04434029,0.000037054964,0.00001938506,0.000010859174,0.0025619138,0.0002480281,8.7749964e-7,0.0002867326,0.950966,0.00016548405],"about_ca_topic_score_codex":0.0013462107,"about_ca_topic_score_gemma":0.0001749114,"teacher_disagreement_score":0.2086191,"about_ca_system_score_codex":0.00020513653,"about_ca_system_score_gemma":0.000118426484,"threshold_uncertainty_score":0.9998215},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"reporting","study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"design_other","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4393928628","doi":"10.1007/s41693-024-00117-x","title":"Integrating lean and robotics in the construction sector: a scientometric analysis","year":2024,"lang":"en","type":"article","venue":"Construction Robotics","topic":"BIM and Construction Integration","field":"Engineering","cited_by":8,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Robotics; Artificial intelligence; Manufacturing engineering; Engineering; Computer science; Robot","score_opus":0.014155584190876334,"score_gpt":0.23507196840438824,"score_spread":0.2209163842135119,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393928628","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13579182,0.0028673895,0.8483014,0.0007112168,0.0054330234,0.00036850877,0.000026657557,0.0008388808,0.0056611123],"genre_scores_gemma":[0.9606119,0.00023230174,0.038875964,0.000028098691,0.00016782882,0.000012882772,0.000021697782,0.000019592851,0.000029744486],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99877703,0.000050716873,0.00040132293,0.00027809234,0.0002619701,0.00023086458],"domain_scores_gemma":[0.9994637,0.0001532336,0.000041159576,0.00020710402,0.00007691273,0.000057906993],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003270891,0.00019937179,0.00021859452,0.0021095031,0.00014494645,0.00040882066,0.00012165753,0.0001387792,0.000044615146],"category_scores_gemma":[0.00005322394,0.0001601658,0.00010666816,0.007117232,0.00031674939,0.00030237588,0.0000179086,0.00045920766,0.000013090667],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000034788268,0.000010759663,0.027764209,0.000117560296,0.00029267056,0.000009849273,0.0006948746,0.37697387,0.0004937571,0.4473914,0.00023492375,0.14601265],"study_design_scores_gemma":[0.00020887816,0.00003068886,0.00366321,0.000083536856,0.00049135915,0.0006807769,0.004024928,0.9825049,0.00017974092,0.0065386235,0.0012470548,0.00034628366],"about_ca_topic_score_codex":0.000020611371,"about_ca_topic_score_gemma":0.00013784812,"teacher_disagreement_score":0.82482004,"about_ca_system_score_codex":0.00013361823,"about_ca_system_score_gemma":0.000042632797,"threshold_uncertainty_score":0.6531372},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4393948406","doi":"10.5539/jel.v13n3p62","title":"An Immersive Virtual Learning Environment for the Development of Hard Skills: A Scientometric Analysis and Systematic Review","year":2024,"lang":"en","type":"article","venue":"Journal of Education and Learning","topic":"Online Learning and Analytics","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Rajabhat Maha Sarakham University","keywords":"Virtual learning environment; Computer science; Psychology; Human–computer interaction; Multimedia","score_opus":0.012883523082057412,"score_gpt":0.31985779018403127,"score_spread":0.30697426710197384,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393948406","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13052082,0.07323793,0.7934453,0.0022738508,0.000259233,0.00023509255,2.7923082e-7,0.0000121982885,0.000015279551],"genre_scores_gemma":[0.96262145,0.0033962275,0.032929253,0.00008681507,0.000059509235,0.0000061907795,0.0000018302057,0.000005907629,0.0008928222],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99883395,0.00020080441,0.00046319858,0.00014151714,0.00026758623,0.00009291745],"domain_scores_gemma":[0.99868375,0.0005837235,0.0004270841,0.000099165474,0.00012794553,0.00007830403],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0028585808,0.000076719916,0.00028575884,0.00081471587,0.00023093083,0.00018213005,0.00019855742,0.00002036682,0.0000068705763],"category_scores_gemma":[0.00069133606,0.00004884487,0.000097185235,0.0012903172,0.000028594211,0.000225568,0.00003360571,0.00027138708,0.0000013389954],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000140376715,0.00080609997,0.01842625,0.07456758,0.0041220426,0.0000049394953,0.09546062,0.037557926,0.0013822885,0.0035627426,0.00025084766,0.7638446],"study_design_scores_gemma":[0.0006746601,0.0020453085,0.05497714,0.06642501,0.008810965,0.00025056448,0.044746216,0.7951652,0.00027102465,0.00016185765,0.025691098,0.0007809537],"about_ca_topic_score_codex":0.0000015071498,"about_ca_topic_score_gemma":3.214936e-7,"teacher_disagreement_score":0.83210063,"about_ca_system_score_codex":0.000039750572,"about_ca_system_score_gemma":0.00022333936,"threshold_uncertainty_score":0.1991836},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4394291074","doi":"10.6084/m9.figshare.21505976","title":"Trends in Graves’ orbitopathy research in the past two decades: a bibliometric analysis","year":2022,"lang":"en","type":"dataset","venue":"Figshare","topic":"Radioactive Decay and Measurement Techniques","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Geography; Regional science; Data science; Computer science","score_opus":0.15203210324421768,"score_gpt":0.4144794560251318,"score_spread":0.26244735278091413,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394291074","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00008801309,0.0003294056,7.729222e-7,0.000038992384,0.000028088461,0.0003571804,0.996764,0.00001633896,0.002377183],"genre_scores_gemma":[0.019113969,0.000013080465,0.0000048883317,0.000037522186,0.00028658425,0.0023518943,0.9781019,0.00001809674,0.00007203914],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9962659,0.0009250033,0.0004047597,0.0006116635,0.0012160586,0.00057665305],"domain_scores_gemma":[0.99818635,0.0005044901,0.00018582382,0.0009307921,0.00012305292,0.00006948296],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0013201012,0.00030271447,0.0004888783,0.14737901,0.00018182566,0.00024426475,0.001488479,0.000112309004,0.51837015],"category_scores_gemma":[0.000088710185,0.00024408005,0.00035038692,0.27462476,0.000015782596,0.00013764165,0.00029025823,0.0020232648,0.00014640887],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009453786,0.00025290137,0.0016326535,0.0000398615,0.00013417758,0.00005873493,0.00005992428,0.000016214062,5.610553e-7,0.000009575564,0.9906247,0.007161227],"study_design_scores_gemma":[0.00024846947,0.0000606344,0.014788797,0.00018191457,0.00007541922,8.4279367e-7,0.00024738905,0.000009336484,0.000012312817,0.00012324573,0.98399353,0.00025808945],"about_ca_topic_score_codex":0.0017436509,"about_ca_topic_score_gemma":0.00044471247,"teacher_disagreement_score":0.51822376,"about_ca_system_score_codex":0.0001749744,"about_ca_system_score_gemma":0.00010839392,"threshold_uncertainty_score":0.99532956},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4394768444","doi":"10.1007/s11192-024-04981-w","title":"Is interdisciplinarity more likely to produce novel or disruptive research?","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":17,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Social Science Fund of China","keywords":"Novelty; Scopus; Computer science; Data science; Quality (philosophy); Citation; Creativity; Psychology; Epistemology; Political science; Social psychology; MEDLINE; World Wide Web","score_opus":0.89847556445197,"score_gpt":0.7359858661900647,"score_spread":0.1624896982619053,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394768444","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7864584,0.0041509597,0.113623306,0.053777814,0.011330716,0.0030846884,0.0011825678,0.00049649767,0.025895108],"genre_scores_gemma":[0.92848015,0.000105886276,0.0052921106,0.00060701737,0.0004593851,0.00007905862,0.0000062897993,0.000049973307,0.06492012],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9515532,0.00034565243,0.001255796,0.0034354443,0.04102551,0.0023843846],"domain_scores_gemma":[0.9627688,0.016287642,0.00016072259,0.0028902027,0.01527341,0.0026192549],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.11287813,0.00038807414,0.00057937205,0.38840893,0.0011103536,0.018075868,0.007795368,0.00025483654,0.0016488236],"category_scores_gemma":[0.22622605,0.00023825964,0.0003503373,0.8426282,0.0008968755,0.0018212951,0.0076874127,0.0014015146,0.00772296],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015043531,0.00080072996,0.015969815,0.00007373959,0.00005597473,0.00020734915,0.011031082,0.00007333726,0.0037540898,0.0033578305,0.56636477,0.39816087],"study_design_scores_gemma":[0.0005463494,0.0014245929,0.10843466,0.00016273528,0.000017131479,0.00007058738,0.005119682,0.024558984,0.009207766,0.0081401635,0.8414085,0.0009088741],"about_ca_topic_score_codex":0.000119358214,"about_ca_topic_score_gemma":0.000041417497,"teacher_disagreement_score":0.45421925,"about_ca_system_score_codex":0.0010590259,"about_ca_system_score_gemma":0.0014067339,"threshold_uncertainty_score":0.9992638},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4394809073","doi":"10.19111/bulletinofmre.1450727","title":"Gold deposits and mineralization studies: A 2018-2022 Scopus-based bibliometric analysis","year":2024,"lang":"en","type":"article","venue":"Bulletin Of The Mineral Research and Exploration","topic":"Geochemistry and Geologic Mapping","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Mineralization (soil science); Web of science; Information retrieval; Geography; Geology; Computer science; Political science; MEDLINE; Soil science","score_opus":0.1025863561408532,"score_gpt":0.3488110058303381,"score_spread":0.24622464968948488,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394809073","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7725941,0.020887867,0.073446654,0.13030043,0.0002504198,0.00052986096,0.00000635962,0.00011678633,0.001867541],"genre_scores_gemma":[0.99231315,0.00074048626,0.0019221293,0.00008326893,0.00004601041,0.000034851382,0.000006748358,0.0000020575687,0.004851294],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9988352,0.0001794664,0.00015946077,0.00026397104,0.00036633923,0.00019553],"domain_scores_gemma":[0.9990687,0.00032439784,0.00003582141,0.0002290407,0.0002849327,0.000057147092],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0013681971,0.00007326369,0.0001188622,0.0054222676,0.00013414779,0.00021241499,0.00023587962,0.000042439176,0.000009112268],"category_scores_gemma":[0.00083791156,0.000050065722,0.000041170846,0.032224506,0.000114976014,0.00011092163,0.00024015592,0.000118611184,0.0000035525504],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020749413,0.0003939984,0.015941728,0.0031458407,0.0012350717,0.000075145574,0.0065159574,0.02019386,0.16912983,0.03770672,0.63644016,0.10901419],"study_design_scores_gemma":[0.00105994,0.0011125157,0.030997863,0.00079513906,0.00022275002,0.00002917484,0.0010363581,0.772555,0.044461943,0.027607318,0.11952097,0.00060103997],"about_ca_topic_score_codex":0.00016373897,"about_ca_topic_score_gemma":0.000054157677,"teacher_disagreement_score":0.7523611,"about_ca_system_score_codex":0.000026794654,"about_ca_system_score_gemma":0.00004435555,"threshold_uncertainty_score":0.98834604},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4394822545","doi":"10.2196/54490","title":"Application of AI in Sepsis: Citation Network Analysis and Evidence Synthesis","year":2024,"lang":"en","type":"article","venue":"Interactive Journal of Medical Research","topic":"Sepsis Diagnosis and Treatment","field":"Medicine","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Context (archaeology); Leverage (statistics); Citation; Library science; Web of science; Science Citation Index; Bibliometrics; Medicine; Data science; MEDLINE; Political science; Computer science; History; Artificial intelligence","score_opus":0.1466048943755881,"score_gpt":0.533394667631858,"score_spread":0.38678977325626995,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394822545","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94370824,0.015171795,0.0057812594,0.03483385,0.000095763775,0.00022600852,0.0000012762442,0.0000044364674,0.00017736622],"genre_scores_gemma":[0.9952934,0.0041094134,0.00025325525,0.00012641476,0.0001658702,0.000032123324,7.54832e-7,0.0000063759344,0.000012431947],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9970822,0.00037146223,0.0004820099,0.00015600484,0.001749305,0.00015902972],"domain_scores_gemma":[0.9899697,0.009047672,0.00009881804,0.00010952695,0.0005714514,0.00020285009],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0048393975,0.00006448584,0.0003607269,0.0010812967,0.000019966228,0.000033207405,0.00010996039,0.00008544694,0.0002399912],"category_scores_gemma":[0.008036232,0.000043411448,0.00014111534,0.0016457844,0.00011317315,0.00018399929,0.000045495213,0.00077864295,0.000007232269],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010736508,0.00076810207,0.4150377,0.00042593296,0.0031367182,0.0007568231,0.0012179544,0.00007074835,0.0014498262,0.00059896545,0.0032022558,0.57226133],"study_design_scores_gemma":[0.000556973,0.0008916929,0.94862777,0.011109692,0.0010953926,0.00011017576,0.00089765806,0.02878881,0.0049339724,0.0015293802,0.0013908676,0.000067629146],"about_ca_topic_score_codex":0.00033440153,"about_ca_topic_score_gemma":0.000105026986,"teacher_disagreement_score":0.57219374,"about_ca_system_score_codex":0.00021779328,"about_ca_system_score_gemma":0.0003216416,"threshold_uncertainty_score":0.96207005},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4394824014","doi":"10.1016/j.amjsurg.2024.04.005","title":"To infinity and beyond: A historical bibliometric analysis of medullary thyroid carcinoma","year":2024,"lang":"en","type":"article","venue":"The American Journal of Surgery","topic":"Thyroid Cancer Diagnosis and Treatment","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Paul's Hospital","funders":"","keywords":"Infinity; Thyroid carcinoma; Medullary cavity; Medicine; Thyroid; Pathology; Mathematics; Internal medicine; Mathematical analysis","score_opus":0.03281662873282666,"score_gpt":0.2948563026230918,"score_spread":0.26203967389026517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394824014","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.961619,0.03189274,0.000062636005,0.005971746,0.0002209441,0.00005426472,0.000011748917,0.000008643483,0.0001582745],"genre_scores_gemma":[0.99443936,0.004513508,0.00017074595,0.0006658608,0.00014386227,0.0000044467342,0.0000014868086,0.000013544843,0.000047199454],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99861115,0.00009445874,0.00052913494,0.00013487558,0.0004628281,0.0001675246],"domain_scores_gemma":[0.9975936,0.0014587742,0.0003167482,0.00022007136,0.00016322872,0.0002476041],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00085168105,0.00012107667,0.00096466276,0.032362156,0.000031284417,0.00002176241,0.00008188214,0.00001752146,0.000037685306],"category_scores_gemma":[0.00019938883,0.00007049062,0.00048652463,0.069860235,0.00011646391,0.000059488357,0.00003826853,0.00017897085,0.0000024665947],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00045581904,0.00022289179,0.8055571,0.000044910037,0.003498474,0.00046153,0.0005478067,0.000024760793,0.000813229,0.000036648995,0.015321741,0.17301512],"study_design_scores_gemma":[0.00013906803,0.00073830277,0.9908532,0.00006866273,0.0033041826,0.00022918741,0.00020874116,0.00003168586,0.00027795593,0.000025760402,0.004049955,0.00007332056],"about_ca_topic_score_codex":0.00039818612,"about_ca_topic_score_gemma":0.000003853116,"teacher_disagreement_score":0.18529612,"about_ca_system_score_codex":0.00029305235,"about_ca_system_score_gemma":0.00024329413,"threshold_uncertainty_score":0.9786052},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4394891681","doi":"10.55905/oelv22n4-118","title":"Levantamento bibliométrico sobre a distribuição de nematoides em solo brasileiro","year":2024,"lang":"pt","type":"article","venue":"OBSERVATÓRIO DE LA ECONOMÍA LATINOAMERICANA","topic":"Nematode management and characterization studies","field":"Agricultural and Biological Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Inversa Systems (Canada)","funders":"","keywords":"Political science","score_opus":0.025849633823607665,"score_gpt":0.2759352733414583,"score_spread":0.2500856395178506,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394891681","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9797089,0.0037115028,0.000363928,0.0053649824,0.00055587856,0.00048960635,0.00058882765,0.0004690236,0.008747335],"genre_scores_gemma":[0.98002505,0.0042330637,0.00019110955,0.0021625906,0.00068740593,0.000115863084,0.00025404705,0.000021655123,0.0123092355],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99650913,0.0003627229,0.0007898945,0.00094648835,0.00030677015,0.0010849652],"domain_scores_gemma":[0.9980347,0.00095402345,0.00033009172,0.00024347904,0.0000732765,0.0003644373],"candidate_categories":["metaepi_narrow","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00077164173,0.000617433,0.00077772187,0.000291335,0.00059373426,0.0019599274,0.0006860847,0.00026150775,0.0023618792],"category_scores_gemma":[0.000115033065,0.00036481288,0.00042350026,0.0038240391,0.0003112682,0.000626307,0.00046678554,0.00042369333,0.00050721515],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009192503,0.00045407846,0.65291977,0.00084872125,0.0010963684,0.0004370753,0.005173925,0.00009237528,0.014190349,0.0064058686,0.1245917,0.19369784],"study_design_scores_gemma":[0.00032385637,0.00026734936,0.7172317,0.00041027463,0.0002720096,0.0000539098,0.011343473,0.0073059164,0.000354132,0.0012153715,0.2601118,0.0011102061],"about_ca_topic_score_codex":0.000998136,"about_ca_topic_score_gemma":0.00064114324,"teacher_disagreement_score":0.19258763,"about_ca_system_score_codex":0.0002913169,"about_ca_system_score_gemma":0.00012403248,"threshold_uncertainty_score":0.9998804},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4394932987","doi":"10.1186/s12991-024-00498-x","title":"A bibliometric and visual analysis of cognitive function in bipolar disorder from 2012 to 2022","year":2024,"lang":"en","type":"article","venue":"Annals of General Psychiatry","topic":"Bipolar Disorder and Treatment","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Shanxi Scholarship Council of China; Health Commission of Shanxi Province","keywords":"Bipolar disorder; Neurocognitive; Cognition; Clinical psychology; Psychology; Mania; Hypomania; Psychiatry; Geriatric psychiatry; Young Mania Rating Scale; Medicine","score_opus":0.027114342851407944,"score_gpt":0.35300099057180206,"score_spread":0.3258866477203941,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394932987","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7385123,0.25932708,0.000329878,0.0011618667,0.00019000655,0.00014860691,0.00019290535,0.000011611258,0.00012579547],"genre_scores_gemma":[0.9954925,0.0033731777,0.00018422563,0.00041911157,0.00008976936,0.000014073202,0.00020279136,0.000015129145,0.00020919964],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9989936,0.00004001087,0.0002961215,0.00028600206,0.00022575857,0.00015855393],"domain_scores_gemma":[0.9995953,0.00006022163,0.000047086814,0.00010709845,0.00008625185,0.000104038416],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0001454938,0.00012905712,0.00037439738,0.021833908,0.000018008954,0.000011823053,0.000029528052,0.000078119716,0.00033500892],"category_scores_gemma":[0.00002324325,0.00010867809,0.00018151723,0.038755383,0.000030552947,0.00009388603,0.000030520194,0.00010273275,0.000010435446],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034054657,0.0006888064,0.87740153,0.00007858025,0.0029780965,0.0000014806939,0.00016011082,0.000013387277,0.0006133934,0.00008631064,0.0013794424,0.116258346],"study_design_scores_gemma":[0.000631733,0.0008388575,0.99207634,0.00011213345,0.0018836277,9.116419e-7,0.00015807214,0.0008587576,0.00025422213,0.00026005792,0.0028281678,0.000097134434],"about_ca_topic_score_codex":0.0021798622,"about_ca_topic_score_gemma":0.0018478631,"teacher_disagreement_score":0.25698027,"about_ca_system_score_codex":0.0000050857634,"about_ca_system_score_gemma":0.00005953144,"threshold_uncertainty_score":0.98925275},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4395003885","doi":"10.23977/aetp.2024.080308","title":"Bibliometric Analysis of Potential Themes and Trend Development of ChatGPT in the Field of Education","year":2024,"lang":"en","type":"article","venue":"Advances in Educational Technology and Psychology","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Field (mathematics); Engineering ethics; Data science; Regional science; Management science; Sociology; Political science; Psychology; Computer science; Engineering; Mathematics","score_opus":0.05087246105872769,"score_gpt":0.4924656654287985,"score_spread":0.4415932043700708,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4395003885","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9510109,0.03584531,0.00025539938,0.011994624,0.0002960194,0.00010096268,0.0000017997019,0.000003286012,0.00049170427],"genre_scores_gemma":[0.9937033,0.0044973297,0.0014988015,0.00020113231,0.000022615806,0.00003790144,0.000015587284,0.0000020244634,0.000021277538],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99931616,0.000024336196,0.00037304073,0.00014608922,0.00006706289,0.00007331507],"domain_scores_gemma":[0.9993234,0.00041489422,0.00007980116,0.00011512132,0.000055414814,0.00001135392],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.000244338,0.000048611415,0.00017073138,0.028628502,0.00001616722,0.0000015548998,0.000065039414,0.00010966171,0.00004503975],"category_scores_gemma":[0.0001776468,0.000037166516,0.000019324458,0.03163998,0.00018403248,0.00005626865,0.000008935591,0.00011869213,3.338741e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000037464128,0.0003377656,0.28616422,0.00014096906,0.00005032442,3.0149457e-7,0.002751307,0.000002792555,0.00041945203,0.025807014,0.00006267091,0.68422574],"study_design_scores_gemma":[0.00005079099,0.00022074036,0.89331514,0.00014005511,0.000108834385,0.000025504298,0.005785274,0.00005394768,0.0030296263,0.093370974,0.003842235,0.000056863537],"about_ca_topic_score_codex":0.00003161637,"about_ca_topic_score_gemma":0.00015769506,"teacher_disagreement_score":0.6841689,"about_ca_system_score_codex":0.000010124832,"about_ca_system_score_gemma":0.00018391586,"threshold_uncertainty_score":0.988943},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4395030071","doi":"10.4187/respcare.11616","title":"The Top 100 Cited Articles Focusing on Acute Lung Injury and ARDS: Bibliometric and Visualization Analyses","year":2024,"lang":"en","type":"article","venue":"Respiratory Care","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Health Commission of Jiangxi Province","keywords":"Medicine; ARDS; Visualization; Intensive care medicine; Lung; Internal medicine; Data mining","score_opus":0.4764789478124501,"score_gpt":0.614326852314032,"score_spread":0.13784790450158185,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4395030071","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9404131,0.057267822,0.0008216602,0.0002700712,0.00033485933,0.00020818313,0.00005155417,0.00005969713,0.00057303364],"genre_scores_gemma":[0.9981762,0.0012158141,0.00004033793,0.00025457467,0.00012005503,0.000009297995,0.000002405615,0.000017039265,0.00016428703],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9938258,0.00032074342,0.00048679445,0.0007118533,0.0042334264,0.00042141657],"domain_scores_gemma":[0.9934075,0.0042379894,0.000100316814,0.0005118402,0.0014519755,0.00029036647],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00639893,0.00016541315,0.00022009976,0.2934871,0.00067216187,0.008534826,0.0005745547,0.000108563334,0.00005537278],"category_scores_gemma":[0.009999512,0.00009422813,0.00008805052,0.6092474,0.00020832919,0.000634852,0.00042631608,0.00020970895,0.000039266022],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000037209185,0.000017712362,0.061631102,0.000024541414,0.00006470767,0.000031694006,0.0003145836,0.00001354184,0.008782749,0.0014362845,0.008402296,0.9192436],"study_design_scores_gemma":[0.0007399039,0.0010066166,0.64411455,0.00015582771,0.00014594209,0.000010578461,0.0037506858,0.013826423,0.03740397,0.0030539145,0.29505354,0.00073806965],"about_ca_topic_score_codex":0.000018116934,"about_ca_topic_score_gemma":0.000006560788,"teacher_disagreement_score":0.9185055,"about_ca_system_score_codex":0.00012103228,"about_ca_system_score_gemma":0.00015505949,"threshold_uncertainty_score":0.9983397},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4395042902","doi":"10.52783/jes.2137","title":"Research on Nurses’ Psychological Empowerment: A Bibliometric Analysis","year":2024,"lang":"en","type":"article","venue":"Journal of Electrical Systems","topic":"Resilience and Mental Health","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Psychology; Empowerment; Psychological research; Applied psychology; Social psychology; Political science","score_opus":0.11433940798048033,"score_gpt":0.5506317539601414,"score_spread":0.4362923459796611,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4395042902","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9359872,0.018052783,0.00041506032,0.00040593458,0.003544238,0.0002848695,0.0000039254114,0.00003386237,0.04127214],"genre_scores_gemma":[0.9934591,0.0004543441,0.000011714323,0.00009758085,0.0008310415,0.000018621186,0.0000012295546,0.00001544012,0.0051109456],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9955612,0.00096292014,0.00089451345,0.0003785677,0.0014765653,0.00072626973],"domain_scores_gemma":[0.99778146,0.0010860473,0.00018269014,0.00029839211,0.0002424702,0.0004089302],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0036976365,0.00014609018,0.0005189589,0.09067879,0.00011366701,0.00021846459,0.000456014,0.00021860914,0.0005203602],"category_scores_gemma":[0.0001231339,0.00009308289,0.0003995452,0.22175555,0.00006991048,0.000116544754,0.00001743093,0.0012739198,0.0009063731],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0022546065,0.0058067786,0.02742038,0.00018538465,0.0032548911,0.0063746385,0.0014679331,0.00028797734,0.0013277731,0.05915276,0.48229223,0.41017467],"study_design_scores_gemma":[0.0026718338,0.042310733,0.6360506,0.00076779694,0.0010212049,0.00563293,0.0027969184,0.0029335008,0.00016555461,0.0015999707,0.3032473,0.0008016726],"about_ca_topic_score_codex":0.00009993963,"about_ca_topic_score_gemma":0.0000012022365,"teacher_disagreement_score":0.6086302,"about_ca_system_score_codex":0.00041437594,"about_ca_system_score_gemma":0.00008903665,"threshold_uncertainty_score":0.99987155},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4395045665","doi":"10.1101/2024.04.22.24305806","title":"Cracking the Code: A Scoping Review to Unite Disciplines in Tackling Legal Issues in Health Artificial Intelligence","year":2024,"lang":"en","type":"review","venue":"medRxiv","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University; University of Toronto; Library and Archives Canada; Royal College of Physicians and Surgeons of Canada; Queen's University; Hospital for Sick Children; University of Ottawa","funders":"Canadian Institutes of Health Research; Hospital for Sick Children","keywords":"Code (set theory); Engineering; Engineering ethics; Computer science; Programming language","score_opus":0.3442469389447934,"score_gpt":0.5690300085512622,"score_spread":0.22478306960646877,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4395045665","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00023667693,0.964213,0.00018642368,0.029693225,0.0012758386,0.0042897495,0.000007315842,0.00005718283,0.000040591607],"genre_scores_gemma":[0.00032809595,0.99574596,0.00019420657,0.00168739,0.0011217676,0.00056875165,0.00004799172,0.00007437797,0.00023149038],"study_design_codex":"design_other","study_design_gemma":"systematic_review","domain_scores_codex":[0.99481875,0.00054328557,0.0028598472,0.00073329057,0.00040262225,0.0006421851],"domain_scores_gemma":[0.99806046,0.00060546224,0.00037123097,0.0006523713,0.00012820291,0.0001822728],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0041730274,0.00043992975,0.0020200578,0.00069481577,0.00014309544,0.00010937164,0.00037626404,0.00019019691,0.00006436445],"category_scores_gemma":[0.0018729508,0.00028505604,0.00026495973,0.0030367314,0.000079975995,0.00010617006,0.00015494796,0.0013126935,0.00045760238],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000058142687,0.000046278717,0.00009345593,0.20321876,0.000010721544,0.000043017666,0.0013116869,0.000012063314,1.5478894e-7,0.0005167482,0.00017361813,0.7945677],"study_design_scores_gemma":[0.0000030224355,0.00007148055,0.0000052379114,0.73763275,0.00012177003,0.000037858674,0.0004710183,0.000112271155,0.0000056572235,0.000464368,0.2608667,0.00020785315],"about_ca_topic_score_codex":0.0016029724,"about_ca_topic_score_gemma":0.008566907,"teacher_disagreement_score":0.7943598,"about_ca_system_score_codex":0.00047741938,"about_ca_system_score_gemma":0.0021638332,"threshold_uncertainty_score":0.9999602},"labels":[{"model":"gpt","categories":[],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"grok","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"opus","categories":["metaepi_broad"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4395956288","doi":"10.18280/ijsdp.190419","title":"Bibliometric Analysis of Tourism and Community Participation Research: A Comparison of Scopus and Web of Science Databases","year":2024,"lang":"en","type":"article","venue":"International Journal of Sustainable Development and Planning","topic":"Tourism, Volunteerism, and Development","field":"Social Sciences","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Scopus; Web of science; Tourism; World Wide Web; Bibliometrics; Database; Library science; Political science; Data science; Geography; Computer science; MEDLINE; Archaeology","score_opus":0.14630380073321456,"score_gpt":0.4506494813728912,"score_spread":0.3043456806396766,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4395956288","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99313414,0.0056099044,0.00044637683,0.00011032988,0.00014071778,0.00006971804,0.0000066709617,0.000004216634,0.0004779451],"genre_scores_gemma":[0.99747753,0.00065387285,0.0017233508,0.000004969682,0.000033230783,0.0000016311368,0.000005099541,0.000003107297,0.000097203156],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9977379,0.00020869033,0.0005904475,0.0001042096,0.0011504765,0.00020830845],"domain_scores_gemma":[0.99684286,0.00093289523,0.00030831,0.000054368837,0.0017553543,0.00010618411],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.007170504,0.00006936533,0.00027369292,0.03282536,0.00038545477,0.0001319236,0.0002528087,0.00003323237,0.000013938735],"category_scores_gemma":[0.0013616493,0.00006199686,0.000032242537,0.020718047,0.0007116715,0.0006147147,0.00021193712,0.0002502148,7.481218e-8],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019955674,0.00025059757,0.5604269,0.0003596864,0.0011709001,0.00018257325,0.39979216,0.0002487959,0.00079091097,0.0054910416,0.0007740533,0.030312864],"study_design_scores_gemma":[0.00027530317,0.000107282634,0.64496934,0.0005113758,0.000120711185,0.0000050088374,0.34371823,0.0011068326,0.002647394,0.00037509427,0.0060622566,0.00010118733],"about_ca_topic_score_codex":0.0013602705,"about_ca_topic_score_gemma":0.0000877519,"teacher_disagreement_score":0.08454245,"about_ca_system_score_codex":0.00012293403,"about_ca_system_score_gemma":0.0012549134,"threshold_uncertainty_score":0.99543345},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4396237870","doi":"10.1007/s11192-024-05017-z","title":"CFMf topic-model: comparison with LDA and Top2Vec","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":10,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Social Sciences and Humanities Research Council of Canada; Fonds de Recherche du Québec-Société et Culture; Canada Research Chairs","keywords":"Computer science; Topic model; Data science; Information retrieval","score_opus":0.08423497244561039,"score_gpt":0.45627159766623665,"score_spread":0.37203662522062625,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396237870","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.47743636,0.00519275,0.35261887,0.004740103,0.0010165056,0.00023299077,0.0000078155035,0.00029760587,0.15845698],"genre_scores_gemma":[0.97278535,0.000053758984,0.019266767,0.00008880592,0.00011581141,0.0000026590858,0.0000020427965,0.000004370612,0.0076804063],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99863535,0.00006568899,0.00011341393,0.00024109121,0.00074927876,0.00019514769],"domain_scores_gemma":[0.99927115,0.00037041778,0.000023685385,0.000076871605,0.00012318254,0.0001346823],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014654126,0.00006142944,0.00011136589,0.0010454407,0.00029196448,0.00042980697,0.0001420467,0.000034869132,0.000069614245],"category_scores_gemma":[0.0002837087,0.000046751673,0.000035239995,0.008901986,0.0001891716,0.00017888051,0.000045767392,0.0000870036,0.000018975235],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000585526,0.00006416789,0.039355103,0.000036970647,0.000048760427,0.0000112644375,0.015625361,0.0030580387,0.000023241513,0.41864195,0.0028870916,0.5202422],"study_design_scores_gemma":[0.00022055978,0.000095594325,0.022076758,0.00004342265,0.00009523424,0.000002982934,0.0022024696,0.4350793,0.00006114396,0.03086521,0.5088897,0.00036765245],"about_ca_topic_score_codex":0.00022054222,"about_ca_topic_score_gemma":0.0003273469,"teacher_disagreement_score":0.5198746,"about_ca_system_score_codex":0.00006472406,"about_ca_system_score_gemma":0.00015748545,"threshold_uncertainty_score":0.42771092},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4396568181","doi":"10.3126/njmr.v7i1.65142","title":"Decolonization in Focus: A Bibliometric Analysis of Scientific Articles from 2010 to 2023","year":2024,"lang":"en","type":"article","venue":"Nepal Journal of Multidisciplinary Research","topic":"Regional Socio-Economic Development Trends","field":"Social Sciences","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Focus (optics); Decolonization; Engineering ethics; Political science; Engineering; Politics; Law; Physics","score_opus":0.15031683868090379,"score_gpt":0.46216137017921133,"score_spread":0.31184453149830754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396568181","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99126065,0.0013104558,0.00017066814,0.0055462443,0.0006267863,0.00016244929,0.000025673677,0.000009867928,0.00088721234],"genre_scores_gemma":[0.9961631,0.0003210268,0.0016387,0.0000037340244,0.00017286927,0.0000079378615,0.0000082357965,0.000011010047,0.0016734236],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9962335,0.0005828626,0.000719461,0.0003151427,0.0016842316,0.00046476081],"domain_scores_gemma":[0.9967711,0.0018466183,0.00013684668,0.00017455808,0.0007782555,0.00029259606],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.009347955,0.00009368005,0.00034495367,0.09485819,0.00038319512,0.00040105,0.00066367764,0.00011632408,0.00065582973],"category_scores_gemma":[0.0011528244,0.00008428112,0.00020362108,0.19205171,0.00047083155,0.00060319685,0.00025061003,0.0004181304,0.00009986586],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026406744,0.00047546282,0.80437106,0.000024932515,0.00086105085,0.0002569019,0.04788942,0.0022866766,0.0045865136,0.0010196049,0.012023222,0.12594108],"study_design_scores_gemma":[0.00029449855,0.00012909627,0.9795931,0.0001403154,0.00007864003,0.0000011895615,0.008807675,0.004546161,0.0003069484,0.0034885227,0.0024945915,0.000119280296],"about_ca_topic_score_codex":0.0022792509,"about_ca_topic_score_gemma":0.009565155,"teacher_disagreement_score":0.17522201,"about_ca_system_score_codex":0.00053406786,"about_ca_system_score_gemma":0.001113585,"threshold_uncertainty_score":0.9154008},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4396609803","doi":"10.7759/cureus.59589","title":"Areas of Interest in Dental Education: A Bibliometric Analysis of the Last Decade","year":2024,"lang":"en","type":"article","venue":"Cureus","topic":"Dental Education, Practice, Research","field":"Health Professions","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Geography","score_opus":0.18195632590683905,"score_gpt":0.5397342070247001,"score_spread":0.357777881117861,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396609803","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9834648,0.004468818,0.000010288167,0.0007054083,0.001338125,0.00032875768,0.00002708203,0.000012623202,0.009644127],"genre_scores_gemma":[0.9975335,0.00027193432,0.000023628758,0.00009575192,0.00009457029,0.000079581965,0.000020818577,0.000012217739,0.001867994],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99854475,0.0003579265,0.00045814912,0.0001832895,0.00025609252,0.00019978326],"domain_scores_gemma":[0.9984836,0.00068185915,0.00015270797,0.00040309937,0.00020767818,0.00007105842],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0009113118,0.00007328433,0.00018507197,0.025394946,0.00008075358,0.00001687734,0.00038141143,0.000090501075,0.0017981469],"category_scores_gemma":[0.0014808557,0.00005374354,0.000111471054,0.13843961,0.00008112627,0.0002180565,0.00019680081,0.00045321634,0.00013028874],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010334032,0.00032854246,0.9847595,0.00022334485,0.00018536144,7.104548e-7,0.0024109872,0.0000028061804,0.0002821347,0.0024625969,0.002837556,0.0064961296],"study_design_scores_gemma":[0.00008072218,0.000015065837,0.9821879,0.00019849898,0.00016516008,0.0000016524665,0.010032956,0.00012308903,0.00018176778,0.00024900626,0.0067201178,0.000044058666],"about_ca_topic_score_codex":0.002477688,"about_ca_topic_score_gemma":0.0054475497,"teacher_disagreement_score":0.11304466,"about_ca_system_score_codex":0.0003399223,"about_ca_system_score_gemma":0.0014740023,"threshold_uncertainty_score":0.99911433},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4396630129","doi":"10.2196/52020","title":"Chinese Health Insurance in the Digital Era: Bibliometric Study","year":2024,"lang":"en","type":"article","venue":"Interactive Journal of Medical Research","topic":"Telemedicine and Telehealth Implementation","field":"Medicine","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"China; Health care; Digital health; Business; Telemedicine; The Internet; Beijing; Political science; Economic growth; Computer science; World Wide Web; Economics","score_opus":0.1229051645113694,"score_gpt":0.5851382341315122,"score_spread":0.4622330696201428,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396630129","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.91756827,0.003092299,0.00014377921,0.077477165,0.00043553524,0.0005775827,0.0000059645954,0.000007955448,0.00069144106],"genre_scores_gemma":[0.99654275,0.0011402509,0.000022046514,0.0012553496,0.00095844315,0.00001751653,0.000003930574,0.000013542655,0.000046157984],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9921953,0.00075730303,0.00095260656,0.00017831309,0.005487092,0.00042937542],"domain_scores_gemma":[0.99369496,0.0050458894,0.00013963692,0.00016761533,0.0005112205,0.0004406473],"candidate_categories":["metaresearch","bibliometrics","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.014779909,0.00010605006,0.00038148914,0.019297734,0.00007277242,0.00016098452,0.00039841965,0.000053070395,0.00043435968],"category_scores_gemma":[0.011556557,0.000052519343,0.00008985586,0.031170053,0.000108752974,0.00039921567,0.00007743611,0.0033795473,0.000038466587],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004434145,0.0011455249,0.26570627,0.000166251,0.000089895715,0.0031210368,0.006599541,1.5793663e-7,0.000010303086,0.000025409163,0.02229671,0.70039546],"study_design_scores_gemma":[0.0023671524,0.0065057175,0.95523685,0.00094227423,0.000005595373,0.0015251646,0.012122621,0.00012858686,0.0000041989506,0.00018834736,0.020932963,0.000040537336],"about_ca_topic_score_codex":0.00034580182,"about_ca_topic_score_gemma":0.00010046738,"teacher_disagreement_score":0.70035493,"about_ca_system_score_codex":0.0003258403,"about_ca_system_score_gemma":0.0018279114,"threshold_uncertainty_score":0.9989197},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4396671170","doi":"10.29333/ejmste/14577","title":"Thematic bibliometric analysis of 37 specialized journals in mathematical education research indexed in &lt;i&gt;Scopus&lt;/i&gt; or &lt;i&gt;Web of Science&lt;/i&gt;","year":2024,"lang":"en","type":"article","venue":"Eurasia Journal of Mathematics Science and Technology Education","topic":"Online Learning and Analytics","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Thematic map; Web of science; Mathematics education; Bibliometrics; Thematic analysis; Citation analysis; Computer science; Library science; Citation; Mathematics; Chemistry; Qualitative research; Sociology; Social science; MEDLINE; Geography","score_opus":0.03943951582574493,"score_gpt":0.3862124787954752,"score_spread":0.34677296296973026,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396671170","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9819538,0.004331219,0.0048024156,0.006310295,0.0005364347,0.00040072433,0.0000027142846,0.00004627681,0.0016161263],"genre_scores_gemma":[0.9495111,0.0013173139,0.04874437,0.000020675247,0.000089503206,0.00001651377,9.833393e-7,0.000020323405,0.0002792305],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9925704,0.00026386656,0.002427751,0.000682094,0.0032579612,0.0007979152],"domain_scores_gemma":[0.9925182,0.0013400833,0.0014352234,0.0009937766,0.003426416,0.0002863069],"candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.021152247,0.00031906457,0.0011624041,0.14648943,0.00033553818,0.0006334422,0.0031601654,0.00026511124,0.000071825125],"category_scores_gemma":[0.0126117645,0.00024318727,0.0001837622,0.26748055,0.0030506433,0.0017533074,0.00047793856,0.00091515,0.000017688653],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024302837,0.0043865354,0.0032369583,0.0010703104,0.0002170753,0.000038122664,0.009058346,0.00022106634,0.35531437,0.52553165,0.0008356918,0.10006556],"study_design_scores_gemma":[0.0013967085,0.0014104652,0.023802223,0.009275191,0.0007685372,0.0013647693,0.008735174,0.3964834,0.018212806,0.5364727,0.0010674872,0.0010105887],"about_ca_topic_score_codex":0.000002523755,"about_ca_topic_score_gemma":0.000034646757,"teacher_disagreement_score":0.39626232,"about_ca_system_score_codex":0.00054227706,"about_ca_system_score_gemma":0.012025782,"threshold_uncertainty_score":0.99966246},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4396828942","doi":"10.1186/s13019-024-02787-8","title":"Insight into the research history and trends of total anomalous pulmonary venous connection: a bibliometric analysis","year":2024,"lang":"en","type":"article","venue":"Journal of Cardiothoracic Surgery","topic":"Congenital Heart Disease Studies","field":"Medicine","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Total anomalous pulmonary venous connection; Cardiac surgery; Cardiothoracic surgery; Connection (principal bundle); Medicine; Internal medicine; Surgery; Engineering","score_opus":0.06296478435876553,"score_gpt":0.3576639206871282,"score_spread":0.2946991363283627,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396828942","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6982109,0.29783988,0.00001666963,0.0005128092,0.00067825045,0.00005242341,0.0000038839926,0.000013450699,0.002671705],"genre_scores_gemma":[0.99426895,0.004055383,0.000029201432,0.000024627905,0.0006928497,0.000005849662,0.0000019224044,0.000021381466,0.00089985994],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9974284,0.00039195185,0.00063473056,0.0002052322,0.0011030234,0.00023666186],"domain_scores_gemma":[0.9960707,0.0025207081,0.000120515026,0.0002736116,0.00081558345,0.00019892468],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0033706403,0.0001391826,0.00092711597,0.0771037,0.000107058426,0.00003790449,0.00005996157,0.00007754819,0.00012669858],"category_scores_gemma":[0.00081696885,0.000093181,0.0009814715,0.07269785,0.00035324722,0.0001931197,0.000078905294,0.0004950858,0.0000036527258],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000798906,0.00021262263,0.07448001,0.00056228926,0.006606169,0.0050027324,0.0024682018,0.000006453715,0.00074603816,0.000005772888,0.008134577,0.90097624],"study_design_scores_gemma":[0.00019682775,0.00062563375,0.9185494,0.0004237948,0.0052723866,0.0059653926,0.0012212603,0.0005373592,0.00019352284,0.000037872946,0.06677158,0.00020492448],"about_ca_topic_score_codex":0.000069502144,"about_ca_topic_score_gemma":0.000005143322,"teacher_disagreement_score":0.9007713,"about_ca_system_score_codex":0.00041157752,"about_ca_system_score_gemma":0.0005368667,"threshold_uncertainty_score":0.9470124},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4396957358","doi":"10.52768/2691-7785/1150","title":"Bibliometric Analysis of the Relationship Between Breast Cancer and Exercise Based on Citespace","year":2024,"lang":"en","type":"article","venue":"Journal on Surgery","topic":"Cancer Risks and Factors","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Breast cancer; Timeline; Bibliometrics; Web of science; Medicine; Cancer; Meta-analysis; Internal medicine; Library science; Geography; Computer science","score_opus":0.06989499970304987,"score_gpt":0.34760722657394155,"score_spread":0.2777122268708917,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396957358","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.992725,0.0019406447,0.000021917444,0.004655785,0.00034191812,0.000055807002,0.00009277567,0.00001131145,0.0001548364],"genre_scores_gemma":[0.9983948,0.0007142964,0.000004392075,0.00018191438,0.00020716262,0.0000022831346,0.0000034626164,0.000012506651,0.00047916907],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99901485,0.000032123156,0.00026782174,0.00012043647,0.00044640127,0.00011835305],"domain_scores_gemma":[0.9974421,0.0020689252,0.00012614431,0.00015936594,0.000073469455,0.00012994443],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0005549605,0.00008845999,0.00036357035,0.04660227,0.000069826274,0.000056824734,0.000035820012,0.000056073746,0.00021007599],"category_scores_gemma":[0.00019342641,0.000050263978,0.00038415001,0.084832564,0.000033235432,0.00005860454,0.0000082701135,0.00035777324,0.0000026744376],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009340028,0.00002551105,0.97185653,0.000043750126,0.00024053399,0.000015890007,0.000025268984,0.00018777247,0.000012806568,0.000011252673,0.0035585754,0.023928696],"study_design_scores_gemma":[0.00011920507,0.000022093769,0.9957838,0.00084066595,0.0011722414,0.000010200492,0.000011526811,0.0004810419,0.000040966133,0.000015518171,0.0014493105,0.00005344794],"about_ca_topic_score_codex":0.00010198692,"about_ca_topic_score_gemma":0.000007527174,"teacher_disagreement_score":0.03823029,"about_ca_system_score_codex":0.00010826552,"about_ca_system_score_gemma":0.00018307316,"threshold_uncertainty_score":0.96420366},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4398136077","doi":"10.1016/j.jvs.2024.03.239","title":"Global Bibliometric Analysis of Vascular Surgery Research Output (2000-2023)","year":2024,"lang":"en","type":"article","venue":"Journal of Vascular Surgery","topic":"Renal and Vascular Pathologies","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Medicine; Bibliometrics; Vascular surgery; Internal medicine; Library science; Cardiac surgery","score_opus":0.09926066718147775,"score_gpt":0.36838605541888714,"score_spread":0.2691253882374094,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4398136077","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.85847265,0.13732186,0.002062617,0.0003738375,0.0011305133,0.00016449197,0.000026946944,0.000049873765,0.00039719325],"genre_scores_gemma":[0.96955055,0.028876934,0.00043195568,0.0000730048,0.00066901377,0.000007782977,0.000035310335,0.000045875393,0.0003095523],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9922116,0.0009471274,0.0018929348,0.0004992285,0.0036640787,0.000785034],"domain_scores_gemma":[0.993206,0.0030170893,0.0004271979,0.0008896603,0.001998271,0.00046174185],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.019128373,0.00033990073,0.00252006,0.12411338,0.00010414947,0.0001681019,0.00032313398,0.0003177085,0.00031134847],"category_scores_gemma":[0.005535543,0.0002452403,0.0076258476,0.21926378,0.00027124523,0.00039007168,0.00015421545,0.0007880125,0.000061101804],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003309381,0.00101258,0.7063747,0.0018934915,0.047170274,0.00992462,0.00020034553,0.0010849299,0.0010307359,0.00009166616,0.06917928,0.16170642],"study_design_scores_gemma":[0.00039429174,0.00020048575,0.9402901,0.0012084026,0.0075131236,0.00077540195,0.00020677586,0.00056030124,0.0005602933,0.00018340973,0.04779031,0.00031709948],"about_ca_topic_score_codex":0.00019533986,"about_ca_topic_score_gemma":0.000010222685,"teacher_disagreement_score":0.23391537,"about_ca_system_score_codex":0.00037916607,"about_ca_system_score_gemma":0.0017415634,"threshold_uncertainty_score":1},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4398246959","doi":"10.1200/go.23.00287","title":"Global Landscape of the Attack of Predatory Journals in Oncology","year":2024,"lang":"en","type":"article","venue":"JCO Global Oncology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"Genentech; European Society for Medical Oncology","keywords":"Publishing; Publication; Psychological intervention; Descriptive statistics; Medicine; Clinical Oncology; Odds ratio; Feeling; Psychology; Family medicine; Medical education; Political science; Internal medicine; Cancer; Social psychology; Nursing; Law; Statistics","score_opus":0.5602366891606101,"score_gpt":0.6480163093402614,"score_spread":0.08777962017965135,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4398246959","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.85793674,0.008724759,0.00020489408,0.004332144,0.002859119,0.00026339418,0.00030144502,0.00001249677,0.12536499],"genre_scores_gemma":[0.99865204,0.00033392877,0.00028128448,0.00020312372,0.000093060095,0.0000057314255,8.2870605e-7,0.0000037584962,0.00042624382],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99283963,0.00087275123,0.0012886021,0.0005165164,0.003955481,0.0005270214],"domain_scores_gemma":[0.9939727,0.0034149864,0.0004186604,0.0006021831,0.0013746019,0.00021686952],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.016626745,0.00012976222,0.0006294917,0.0075025605,0.000051874507,0.00021903182,0.0026874708,0.00038189284,0.0009630299],"category_scores_gemma":[0.014716134,0.0000735774,0.00027524043,0.13269792,0.00043770668,0.00025235928,0.0011429733,0.00032319606,0.0001233244],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008568843,0.00035994823,0.6272141,0.000018455328,0.00003966454,0.000091624286,0.00008787103,0.00029873356,0.00016847864,0.005480095,0.11933429,0.2468211],"study_design_scores_gemma":[0.0005980243,0.0006628753,0.60482055,0.000030201321,0.000012081048,0.00009642965,0.00035582483,0.0022922088,0.00007564383,0.013563184,0.3774094,0.00008359729],"about_ca_topic_score_codex":0.00017231744,"about_ca_topic_score_gemma":0.0008192861,"teacher_disagreement_score":0.25807512,"about_ca_system_score_codex":0.0009637513,"about_ca_system_score_gemma":0.0024092821,"threshold_uncertainty_score":0.99995023},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W4399050792","doi":"10.5539/ies.v17n3p62","title":"STEAM Learning Environment on Gamification System to Promote Innovators: A Bibliometric Analysis and Systematic Review","year":2024,"lang":"en","type":"article","venue":"International Education Studies","topic":"Technology-Enhanced Education Studies","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Mathematics education; Statistical analysis; Higher education; Psychology; Sociology; Pedagogy; Knowledge management; Computer science; Political science; Statistics; Mathematics","score_opus":0.0445063474333549,"score_gpt":0.4204193496505067,"score_spread":0.3759130022171518,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399050792","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4551424,0.44384965,0.0043919105,0.069630966,0.0093265185,0.004922566,0.000033048196,0.002814478,0.009888431],"genre_scores_gemma":[0.9551049,0.0330488,0.0007038361,0.00045531473,0.00019815637,0.001611641,0.000014705241,0.000014869403,0.008847809],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99798006,0.00016166444,0.00054438424,0.000454748,0.0006795147,0.00017963973],"domain_scores_gemma":[0.99821687,0.00035492316,0.00021642272,0.00021522309,0.0009272734,0.00006926301],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0014188705,0.00016177134,0.00042208083,0.021742266,0.00037957032,0.00015810928,0.00024024659,0.00005414112,0.000033363136],"category_scores_gemma":[0.003715503,0.00014258918,0.00008101737,0.040577643,0.00012558393,0.00020440572,0.00008956733,0.00015735597,0.0003207053],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000032701013,0.0010047518,0.013501981,0.2013848,0.023162462,0.000009003265,0.1249469,0.0002693568,0.00041902054,0.28373963,0.072006434,0.279523],"study_design_scores_gemma":[0.00019418339,0.00024319866,0.029261839,0.12649079,0.0045513995,0.000015341366,0.19690934,0.0002200562,0.0003008947,0.001009317,0.639619,0.0011846694],"about_ca_topic_score_codex":0.00010870288,"about_ca_topic_score_gemma":0.000025590061,"teacher_disagreement_score":0.5676125,"about_ca_system_score_codex":0.0012987119,"about_ca_system_score_gemma":0.0001911863,"threshold_uncertainty_score":0.98934543},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4399140123","doi":"10.18280/ijsdp.190512","title":"Bibliometric Analysis of Sustainable Development Goals (SDGs) Research in India: Trends, Contribution, and Impact","year":2024,"lang":"en","type":"article","venue":"International Journal of Sustainable Development and Planning","topic":"Sustainable Development and Environmental Policy","field":"Environmental Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Sustainable development; Environmental planning; Regional science; Environmental resource management; Political science; Environmental science; Geography","score_opus":0.02224492924977441,"score_gpt":0.334273356332922,"score_spread":0.3120284270831476,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399140123","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99432755,0.0028864662,0.0001954973,0.00018594062,0.00009577483,0.00016346537,0.0000034778466,0.000011450761,0.0021303874],"genre_scores_gemma":[0.9959644,0.0002875133,0.0008839363,0.00003262939,0.000050860952,0.000016395745,0.00004309898,0.000016973174,0.0027041624],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9965662,0.0001092589,0.0009541137,0.00034775457,0.0012346394,0.0007880656],"domain_scores_gemma":[0.998774,0.00036578835,0.0002519312,0.00009062148,0.00029928132,0.00021839948],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0039142664,0.0002461211,0.00044940112,0.061743516,0.00021369966,0.00032356678,0.0003881595,0.00012520533,0.00040561944],"category_scores_gemma":[0.0002707118,0.00021353553,0.0000896146,0.04463562,0.00019065994,0.001074499,0.0006013972,0.00040068367,0.000007281298],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002670092,0.00015030974,0.91985697,0.00011615371,0.0015719973,0.0072056297,0.016360465,0.0023358667,0.00011345849,0.0017368944,0.0016322631,0.048653007],"study_design_scores_gemma":[0.0005866599,0.00006671669,0.9340761,0.00012119816,0.00006189915,0.000058002024,0.02732444,0.00023189811,0.00059548893,0.0005392298,0.03610441,0.00023397754],"about_ca_topic_score_codex":0.0003318636,"about_ca_topic_score_gemma":0.000008551509,"teacher_disagreement_score":0.04841903,"about_ca_system_score_codex":0.0020966362,"about_ca_system_score_gemma":0.00039291737,"threshold_uncertainty_score":0.9756711},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4399423917","doi":"10.21203/rs.3.rs-4443326/v1","title":"The Research Trends of Post-operative Pediatric Cerebellar Mutism Syndrome: A Bibliometric Analysis (1999-2022)","year":2024,"lang":"en","type":"preprint","venue":"Research Square","topic":"Fetal and Pediatric Neurological Disorders","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medulloblastoma; Medicine; Pediatrics; Psychology; Library science; Pathology","score_opus":0.08174718659232176,"score_gpt":0.43465247517610794,"score_spread":0.35290528858378617,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399423917","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8757429,0.100214556,0.000004687586,0.010076121,0.0003830523,0.0017174617,0.00082208647,0.000121356534,0.010917743],"genre_scores_gemma":[0.93522316,0.051253818,0.000030079063,0.000044838027,0.00046736456,0.0004154884,0.00037807276,0.00008174343,0.01210544],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9861109,0.0028656535,0.0010653872,0.0016812993,0.006486447,0.0017902678],"domain_scores_gemma":[0.9879921,0.005666207,0.00017741567,0.0018880011,0.0035885815,0.0006876934],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.009352158,0.0005280307,0.0013321958,0.20579475,0.0007601419,0.00040168586,0.0013378712,0.0008986339,0.0014434544],"category_scores_gemma":[0.005148376,0.0003216803,0.0013329296,0.42968008,0.001043825,0.000058605732,0.0050047045,0.00935125,0.00052773487],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0065321284,0.0076561305,0.30755916,0.038905323,0.019837948,0.017312802,0.0051113004,0.0011752691,0.0011378283,0.0065250574,0.28444105,0.303806],"study_design_scores_gemma":[0.0014096427,0.00857426,0.95888144,0.00030334183,0.0033072091,0.000107037464,0.0015055896,0.0025113455,0.000078924284,0.009669331,0.012875107,0.0007768017],"about_ca_topic_score_codex":0.0018082208,"about_ca_topic_score_gemma":0.00018596827,"teacher_disagreement_score":0.65132225,"about_ca_system_score_codex":0.00028579778,"about_ca_system_score_gemma":0.0015100776,"threshold_uncertainty_score":0.9999235},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4399470549","doi":"10.24252/kah.v12i1a11","title":"Health Education during the COVID-19 Pandemic: A Bibliometric Analysis","year":2024,"lang":"en","type":"article","venue":"Khizanah al-Hikmah Jurnal Ilmu Perpustakaan Informasi dan Kearsipan","topic":"COVID-19 Pandemic Impacts","field":"Economics, Econometrics and Finance","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Pandemic; Bibliometrics; China; Public health; Coronavirus disease 2019 (COVID-19); Political science; Global health; Higher education; Economic growth; Library science; Medicine; Economics; Nursing","score_opus":0.061627672817634864,"score_gpt":0.3317206533818061,"score_spread":0.27009298056417125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399470549","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.919844,0.03322074,0.0063940682,0.024043703,0.002492568,0.0014507106,0.00052401336,0.0008983509,0.011131879],"genre_scores_gemma":[0.9677624,0.0044931704,0.00023074541,0.020679569,0.00045090192,0.0000743431,0.0001920826,0.00009663699,0.0060201334],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9948451,0.00012211311,0.002176142,0.0009741679,0.00041900683,0.0014634874],"domain_scores_gemma":[0.9957491,0.00055442,0.00096869684,0.0012326089,0.00013988836,0.0013552522],"candidate_categories":["metaepi_narrow","bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0036492147,0.00060951273,0.001135186,0.06034703,0.0010319832,0.0013602821,0.00111621,0.00026114628,0.00046233545],"category_scores_gemma":[0.0017433221,0.0005521352,0.0008574011,0.099627726,0.0002452026,0.0016998508,0.00029902052,0.0011955863,0.00052808423],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007901188,0.00019291937,0.8911764,0.0008794127,0.0018752322,0.000035576166,0.008125456,0.0030682376,0.000031041403,0.02619745,0.02752471,0.04081454],"study_design_scores_gemma":[0.0005635584,0.00009443136,0.24065232,0.000058468275,0.0001445146,0.00027437415,0.00074783806,0.0055183102,0.0000052054097,0.000892468,0.7504375,0.00061103306],"about_ca_topic_score_codex":0.0034942876,"about_ca_topic_score_gemma":0.000673117,"teacher_disagreement_score":0.7229128,"about_ca_system_score_codex":0.003205496,"about_ca_system_score_gemma":0.0025596942,"threshold_uncertainty_score":0.99969304},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4399533235","doi":"10.3390/en17122869","title":"Energy Efficiency Trends in Petroleum Extraction: A Bibliometric Study","year":2024,"lang":"en","type":"article","venue":"Energies","topic":"Global Energy and Sustainability Research","field":"Energy","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Sustainability; Descriptive statistics; Thematic analysis; Energy consumption; Computer science; Data science; Engineering; Sociology; Library science; Social science; Qualitative research; Mathematics","score_opus":0.021034936159610516,"score_gpt":0.33071739661597077,"score_spread":0.3096824604563603,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399533235","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8870853,0.005096023,0.00009942881,0.00040623482,0.00066233415,0.000042919295,0.0000026695136,0.00039354092,0.10621155],"genre_scores_gemma":[0.9694206,0.00014267101,0.000024212284,0.00002872277,0.00019566137,0.00011076473,0.000012905595,0.00003144793,0.030033024],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9974246,0.0002673163,0.0003860266,0.0006143838,0.00066395756,0.0006437341],"domain_scores_gemma":[0.9990502,0.00026877955,0.00002470803,0.00045316108,0.00008434975,0.00011882828],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0006249384,0.0002425069,0.00027696972,0.034460682,0.00012775003,0.00029492442,0.00036748403,0.00012921252,0.0010681667],"category_scores_gemma":[0.00011944249,0.00021056877,0.00014624307,0.07152001,0.00009301265,0.0003923071,0.0001531918,0.0002997665,0.000061297476],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010511674,0.0011497022,0.0037662515,0.00004739076,0.000086392036,0.0011859153,0.000713586,0.28239703,0.00024093613,0.17741412,0.0016135368,0.53128004],"study_design_scores_gemma":[0.0018982502,0.0015122668,0.07681208,0.00007212019,0.000057990288,0.00010044141,0.010205556,0.057526242,0.00154334,0.01451788,0.83461964,0.0011341616],"about_ca_topic_score_codex":0.013026028,"about_ca_topic_score_gemma":0.004464077,"teacher_disagreement_score":0.83300614,"about_ca_system_score_codex":0.00028672293,"about_ca_system_score_gemma":0.00011452693,"threshold_uncertainty_score":0.99984497},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4399549168","doi":"10.1177/09718907241253422","title":"Organizational Legitimacy: A Bibliometric Analysis of Web of Science Database","year":2024,"lang":"en","type":"article","venue":"Paradigm A Management Research Journal","topic":"Corporate Social Responsibility Reporting","field":"Business, Management and Accounting","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Legitimacy; Corporate governance; Citation; Political science; Social Sciences Citation Index; Distribution (mathematics); Scope (computer science); Bibliometrics; Public relations; Accounting; Sociology; Library science; Business; Management; Computer science; Economics; Law","score_opus":0.09057195942722031,"score_gpt":0.3878144939995742,"score_spread":0.2972425345723539,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399549168","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9290723,0.0011198344,0.003858144,0.00092180737,0.0004155777,0.000533365,0.000017604538,0.00007488657,0.06398647],"genre_scores_gemma":[0.9984532,0.00025543926,0.0004509269,0.000031742344,0.00030271462,0.00000804702,0.000020597727,0.000020869518,0.00045645135],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99451756,0.00007111445,0.00088636455,0.0004560562,0.0034803615,0.00058853556],"domain_scores_gemma":[0.9970432,0.0003362982,0.00043947747,0.00045080038,0.0016696006,0.00006065686],"candidate_categories":["bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.015940702,0.00014184022,0.000351481,0.20884554,0.0004995366,0.0013544111,0.00091349613,0.00003221089,0.0010094752],"category_scores_gemma":[0.0028091634,0.00012724365,0.00021081809,0.4736714,0.0009738745,0.0022609197,0.00096288545,0.0004182395,0.00006687657],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016287433,0.00050067645,0.17822047,0.0018860966,0.0023178675,0.000868249,0.00016372147,0.00074131496,0.010159548,0.75304985,0.004523501,0.047405843],"study_design_scores_gemma":[0.0014760144,0.0001217332,0.48860452,0.0011426492,0.0037996669,0.00006970011,0.0027236529,0.28667006,0.0010174554,0.12585185,0.08756621,0.0009564883],"about_ca_topic_score_codex":0.00014308296,"about_ca_topic_score_gemma":0.00002615166,"teacher_disagreement_score":0.627198,"about_ca_system_score_codex":0.00021244548,"about_ca_system_score_gemma":0.000845543,"threshold_uncertainty_score":0.99990374},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4399682210","doi":"10.1111/jocd.16427","title":"A bibliometric analysis of alternative drug therapy options in the treatment of androgenetic alopecia","year":2024,"lang":"en","type":"article","venue":"Journal of Cosmetic Dermatology","topic":"Hair Growth and Disorders","field":"Medicine","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Mediprobe Research (Canada); University of Toronto","funders":"","keywords":"Medicine; Minoxidil; Dutasteride; Finasteride; Drug; Dermatology; Randomized controlled trial; Pharmacology; Internal medicine","score_opus":0.026403890357344705,"score_gpt":0.3373238573670578,"score_spread":0.3109199670097131,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399682210","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9571238,0.03974751,0.00048199133,0.0018761106,0.00010730369,0.00015525962,0.000007757029,0.0000026577727,0.0004975801],"genre_scores_gemma":[0.9816227,0.017960176,0.0002436008,0.00011645856,0.000022692117,0.0000059590398,0.0000035491319,0.0000062282356,0.000018627772],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99868035,0.0001813974,0.00063480577,0.00008760853,0.00028345687,0.00013238523],"domain_scores_gemma":[0.99892765,0.0004947643,0.00020122787,0.0001521976,0.00016546657,0.000058691036],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00023757933,0.00009683623,0.0006460346,0.0455325,0.0000124994895,0.000009319473,0.00012223619,0.00004226082,0.00009305414],"category_scores_gemma":[0.00006024041,0.000054234657,0.0003725398,0.04929078,0.00010453517,0.000038957834,0.0000049323276,0.000094394185,0.0000031378217],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011926874,0.0065866415,0.7440074,0.00034985732,0.016508263,0.0034247246,0.023979062,0.0015066767,0.002095177,0.0023579956,0.0028825942,0.19510889],"study_design_scores_gemma":[0.0077620493,0.0034682204,0.93710756,0.00021523664,0.006114645,0.0037563744,0.0032655108,0.005774073,0.006237878,0.0022479154,0.023861155,0.00018935923],"about_ca_topic_score_codex":0.000033900844,"about_ca_topic_score_gemma":0.000037525602,"teacher_disagreement_score":0.19491953,"about_ca_system_score_codex":0.00003388757,"about_ca_system_score_gemma":0.00016804534,"threshold_uncertainty_score":0.970917},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4399795684","doi":"10.56294/pa202432","title":"Community participation in the Latin American context: Bibliometric Analysis","year":2024,"lang":"en","type":"article","venue":"Perspectiva austral.","topic":"Scientific Research and Technology","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Context (archaeology); Latin Americans; Bibliometrics; Regional science; Political science; Sociology; Geography; Library science; Computer science; Archaeology","score_opus":0.07834638113737236,"score_gpt":0.39403082377193344,"score_spread":0.3156844426345611,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399795684","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96070075,0.00020003755,0.028559,0.008263875,0.00010827786,0.00013980099,0.0000033005033,0.00021205533,0.0018128984],"genre_scores_gemma":[0.999019,0.000016951693,0.00057149184,0.000060168062,0.000018995546,0.00003616575,0.0000025481513,0.000003057345,0.0002715878],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9984072,0.00040044816,0.0001599278,0.00028680378,0.0004063776,0.000339235],"domain_scores_gemma":[0.99858135,0.00050562556,0.00003226805,0.00075807143,0.00006782158,0.00005483649],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0014786704,0.0000825327,0.00014372187,0.035550885,0.00016328302,0.0005943701,0.0012796661,0.000032128515,0.000047964546],"category_scores_gemma":[0.00039645136,0.00005608538,0.00009557419,0.28191414,0.00024188845,0.0002957759,0.00015002841,0.00056500605,0.00007912422],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009881633,0.0006565947,0.09835811,0.000024378829,0.0003823883,0.00016188379,0.023172323,0.00025946702,0.001005814,0.2656466,0.01059303,0.59972954],"study_design_scores_gemma":[0.0000922933,0.00014050897,0.9460648,0.0000062495674,0.000022797687,0.0000042680053,0.0044970047,0.039658394,0.0005087487,0.0066078636,0.002282269,0.000114839604],"about_ca_topic_score_codex":0.004590445,"about_ca_topic_score_gemma":0.0017494604,"teacher_disagreement_score":0.8477067,"about_ca_system_score_codex":0.00009661308,"about_ca_system_score_gemma":0.0000461481,"threshold_uncertainty_score":0.9753803},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4400114850","doi":"10.53694/bited.1422271","title":"Bibliometric Analysis of Articles on Distance Education During the Last Two Decades","year":2024,"lang":"en","type":"article","venue":"Bilgi ve İletişim Teknolojileri Dergisi","topic":"Education Practices and Challenges","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Distance education; Quarter (Canadian coin); Coronavirus disease 2019 (COVID-19); Library science; Geography; Distribution (mathematics); General partnership; China; Political science; Sociology; Medicine; Pedagogy; Mathematics; Computer science","score_opus":0.04148856937727385,"score_gpt":0.3175360594851728,"score_spread":0.2760474901078989,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400114850","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9622438,0.011317525,0.00002719524,0.0040571936,0.0010243513,0.00016501277,0.000044399076,0.00012920931,0.020991333],"genre_scores_gemma":[0.9926698,0.0016036347,0.000033664288,0.00020819656,0.00056186673,0.000075447926,0.000018094493,0.000029379906,0.004799922],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984713,0.00010126702,0.00040742444,0.00037636253,0.00038412854,0.0002595134],"domain_scores_gemma":[0.99833846,0.00070562016,0.00019079213,0.00048826096,0.00021080187,0.000066069464],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0003015809,0.00019714904,0.0002513559,0.018925678,0.00033114685,0.00054292555,0.00031231518,0.00004614305,0.002259815],"category_scores_gemma":[0.00014621462,0.00013378735,0.00021516588,0.022150872,0.000245022,0.00039556346,0.000049988685,0.00021688231,0.00018713486],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011874485,0.0014783728,0.00893732,0.0005939667,0.0030308494,0.00001200102,0.09650939,0.0006557054,0.0015510742,0.71811014,0.019534871,0.14946759],"study_design_scores_gemma":[0.00021574867,0.00015024518,0.0766607,0.00023968314,0.0010515548,0.000005544415,0.05971882,0.0006661799,0.0021317883,0.0017776263,0.85695434,0.0004277468],"about_ca_topic_score_codex":0.00020070653,"about_ca_topic_score_gemma":0.001381858,"teacher_disagreement_score":0.8374195,"about_ca_system_score_codex":0.00006536297,"about_ca_system_score_gemma":0.00011292923,"threshold_uncertainty_score":0.9986523},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4400222791","doi":"10.15517/ijds.2024.60813","title":"Scientometric Analysis of Activated Carbon or Probiotics in Mouthwashes or Toothpastes: Dynamicity, Spatiotemporal Evolution and Trends","year":2024,"lang":"en","type":"article","venue":"Odovtos - International Journal of Dental Sciences","topic":"Dental Research and COVID-19","field":"Dentistry","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Food science; Chemistry","score_opus":0.050452617918998986,"score_gpt":0.403115850898346,"score_spread":0.35266323297934704,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400222791","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9976949,0.00033166498,0.00019839598,0.00015200471,0.00082149374,0.000059204434,0.00007576388,0.000010200897,0.00065633643],"genre_scores_gemma":[0.99874496,0.00004562719,0.00016862825,0.00001731736,0.00005986521,0.0000011654505,0.0000070466463,0.0000055817786,0.00094980095],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9970227,0.00008710431,0.00056316546,0.00025134854,0.001859109,0.0002165834],"domain_scores_gemma":[0.99897665,0.0003513799,0.00027338805,0.000068504996,0.00019367652,0.00013638152],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008846471,0.00011877756,0.00026106325,0.0070986105,0.000054460947,0.00041241173,0.00065172807,0.000058408077,0.00036400525],"category_scores_gemma":[0.0007518885,0.00007805505,0.00014285467,0.0094272485,0.00036248236,0.0007321449,0.00018732848,0.00018852226,0.0000030457024],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017547158,0.0006696787,0.9098654,0.00010152543,0.0014882549,0.0015934583,0.0008711246,0.0021866658,0.035386674,0.0020363824,0.00016774407,0.04387842],"study_design_scores_gemma":[0.0015113523,0.0013180064,0.8789626,0.0004862055,0.00025373002,0.00066969794,0.0029271238,0.10545186,0.0073298076,0.00070836575,0.00008980552,0.00029146078],"about_ca_topic_score_codex":0.0009870502,"about_ca_topic_score_gemma":0.006091655,"teacher_disagreement_score":0.103265196,"about_ca_system_score_codex":0.0005593308,"about_ca_system_score_gemma":0.0003654158,"threshold_uncertainty_score":0.63340163},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4400228684","doi":"10.3138/jsp-2023-0063","title":"Hidden Dangers: COVID-19-Based Research in Predatory Journals","year":2024,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"Zoonotic diseases and public health","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Coronavirus disease 2019 (COVID-19); 2019-20 coronavirus outbreak; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Geography; Ecology; Data science; Biology; Computer science; Virology; Medicine; Outbreak","score_opus":0.32110599399205936,"score_gpt":0.47230654877854433,"score_spread":0.15120055478648498,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400228684","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.42966673,0.048978478,0.0013908293,0.49354297,0.0021207533,0.0005996396,0.00004740574,0.000105447885,0.023547728],"genre_scores_gemma":[0.9839655,0.0003951229,0.0023314224,0.010186208,0.0025059662,0.0000071830827,0.000010425173,0.00004751262,0.0005506369],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9938991,0.00080182916,0.001228167,0.00031947144,0.0030524859,0.0006989459],"domain_scores_gemma":[0.99333364,0.0013887159,0.00028362518,0.00038202794,0.002141206,0.002470764],"candidate_categories":["metaresearch","scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","scholarly_communication"],"category_scores_codex":[0.036100354,0.0001643714,0.0004808646,0.003655253,0.00018787761,0.025984028,0.00069486204,0.0003292857,0.0013691119],"category_scores_gemma":[0.028869376,0.00012921411,0.00026687974,0.0023278191,0.00010120191,0.0366259,0.00010813066,0.0065246783,0.000023035642],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012884669,0.0010825491,0.10128735,0.0039290907,0.00040231153,0.017563274,0.004149981,0.0001682155,0.0010316598,0.0057219053,0.5912125,0.27216268],"study_design_scores_gemma":[0.0039997254,0.0008807389,0.030516189,0.005516906,0.00010649687,0.0013859948,0.005633718,0.0011756645,0.000037393726,0.008120911,0.94237804,0.00024823405],"about_ca_topic_score_codex":0.0002528304,"about_ca_topic_score_gemma":0.000022047541,"teacher_disagreement_score":0.55429876,"about_ca_system_score_codex":0.0017398116,"about_ca_system_score_gemma":0.015250023,"threshold_uncertainty_score":0.9995438},"labels":[{"model":"gemma","categories":["metaresearch","scholarly_communication","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics","scholarly_communication","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4400484915","doi":"10.3390/plants13141902","title":"Trends and Directions in Oats Research under Drought and Salt Stresses: A Bibliometric Analysis (1993–2023)","year":2024,"lang":"en","type":"review","venue":"Plants","topic":"Crop Yield and Soil Fertility","field":"Agricultural and Biological Sciences","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Natural Science Foundation of China","keywords":"Agriculture; China; Crop; Climate change; Temperate climate; Geography; Drought stress; Psychological resilience; Agronomy; Political science; Biology; Ecology; Psychology","score_opus":0.21204840333314323,"score_gpt":0.4155762519482906,"score_spread":0.2035278486151474,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400484915","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.068318754,0.9298154,3.4204945e-8,0.00011436592,0.000090829184,0.00011660148,0.000515562,0.000029807743,0.0009986346],"genre_scores_gemma":[0.038086865,0.9599598,0.0000017558104,0.000006315177,0.00010213432,0.000031314965,0.0002581893,0.0000013520165,0.0015522633],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981806,0.0002598413,0.00032113466,0.00059921184,0.00030207823,0.00033713467],"domain_scores_gemma":[0.9986363,0.0010732649,0.000049957587,0.00008164269,0.000029191617,0.00012965965],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00066329533,0.0002122462,0.00079482474,0.013214602,0.00016134046,0.0002751504,0.0001743014,0.00024737295,0.00017952966],"category_scores_gemma":[0.00008750081,0.000073708004,0.00017942584,0.12223448,0.000080123646,0.000067636895,0.000194123,0.00047533473,0.000025986492],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000033614274,0.00004628352,0.0020202182,0.000535014,0.00015110233,0.000019207659,0.000020951256,4.173457e-8,0.0000027972671,0.000008022945,0.00047558045,0.9967174],"study_design_scores_gemma":[0.00004015468,0.000057689147,0.24025354,0.0012441623,0.00066064147,0.000016846889,0.00009129573,0.00001855163,5.5383117e-7,0.000169668,0.75718576,0.00026111948],"about_ca_topic_score_codex":0.0028567256,"about_ca_topic_score_gemma":0.020925868,"teacher_disagreement_score":0.9964563,"about_ca_system_score_codex":0.000041211108,"about_ca_system_score_gemma":0.00001687772,"threshold_uncertainty_score":0.99796975},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4400520506","doi":"10.2139/ssrn.4885729","title":"TECHNOLOGY AND AUTOMATION IN FINANCIAL TRADING: A BIBLIOMETRIC REVIEW","year":2024,"lang":"en","type":"review","venue":"SSRN Electronic Journal","topic":"Stock Market Forecasting Methods","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Automation; Business; Finance; Engineering","score_opus":0.12482163633817543,"score_gpt":0.46178262680540494,"score_spread":0.3369609904672295,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400520506","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000022975868,0.9948005,0.0026796982,0.00046245652,0.0006809693,0.0006233754,0.0000044820317,0.00007203169,0.00065350754],"genre_scores_gemma":[0.000026459022,0.99754894,0.00083919434,0.0000360659,0.00023755112,0.00005788766,0.0000018878283,0.000051984163,0.0012000351],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9919144,0.0013233469,0.0023754532,0.00085563795,0.00129456,0.0022366247],"domain_scores_gemma":[0.99594504,0.002004805,0.0012289395,0.00051600375,0.00019158385,0.0001136129],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","research_integrity"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.04658497,0.0004916699,0.0024836767,0.12875956,0.0001410636,0.00035097238,0.0013931233,0.0005479205,0.000059056943],"category_scores_gemma":[0.03369115,0.00032281256,0.000560769,0.20326327,0.000103536935,0.00026274184,0.0002523132,0.006328271,0.00013124352],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000017680635,0.000012294352,0.00001324197,0.0027067652,0.00004052754,0.00002153599,0.000006783494,3.379029e-8,3.2958198e-8,0.012356265,0.0016790093,0.98316175],"study_design_scores_gemma":[0.00008663467,0.00011538327,0.000011902935,0.021922076,0.00037386932,0.005329193,0.00001499655,0.000041320556,1.8474354e-8,0.3611186,0.6107655,0.00022050548],"about_ca_topic_score_codex":0.0000035759233,"about_ca_topic_score_gemma":0.000052921598,"teacher_disagreement_score":0.9829412,"about_ca_system_score_codex":0.0015207552,"about_ca_system_score_gemma":0.007637492,"threshold_uncertainty_score":0.9999224},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4400524773","doi":"10.2196/48259","title":"Evolution of Digital Health and Exploration of Patented Technologies (2017-2021): Bibliometric Analysis","year":2024,"lang":"en","type":"article","venue":"Interactive Journal of Medical Research","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Health technology; Health informatics; Safeguarding; Digital health; Data science; Intellectual property; Bibliometrics; Public health; Computer science; Political science; Health care; Medicine; Library science","score_opus":0.39001294395469166,"score_gpt":0.5820254790110617,"score_spread":0.19201253505637006,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400524773","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9453615,0.014663783,0.014796845,0.0245999,0.00028479795,0.00016600998,0.000007456576,0.0000098731525,0.0001098328],"genre_scores_gemma":[0.9906695,0.00904313,0.00010011731,0.000010345941,0.00012882524,0.0000038041671,0.000006122172,0.000005515323,0.000032639746],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.99667877,0.00017993066,0.00086636964,0.00013737191,0.0019435016,0.00019403895],"domain_scores_gemma":[0.9952829,0.0019344983,0.00027271538,0.00012849287,0.002134119,0.0002473076],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0037077565,0.00006313952,0.00037439118,0.045184374,0.000039540853,0.00004243244,0.00014199174,0.00012969095,0.000120818615],"category_scores_gemma":[0.01666453,0.00004439602,0.00012521308,0.042500008,0.00031063164,0.0006153432,0.000065854365,0.00088993326,0.000006489952],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005901416,0.00036615317,0.029423507,0.0005131308,0.00050964725,0.0000575895,0.0028412049,0.0000046995256,0.0006805205,0.00037405745,0.001956683,0.96268266],"study_design_scores_gemma":[0.0012337943,0.03903834,0.124174625,0.027011106,0.0011705509,0.0013375971,0.5690583,0.062015075,0.099686354,0.061916362,0.0127819935,0.0005758719],"about_ca_topic_score_codex":0.0004455063,"about_ca_topic_score_gemma":0.000020973745,"teacher_disagreement_score":0.9621068,"about_ca_system_score_codex":0.0003285208,"about_ca_system_score_gemma":0.001749567,"threshold_uncertainty_score":0.9916185},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4400532816","doi":"10.1097/js9.0000000000001946","title":"The top 100 most-cited articles in perioperative stroke: a bibliometric analysis","year":2024,"lang":"en","type":"article","venue":"International Journal of Surgery","topic":"Acute Ischemic Stroke Management","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Perioperative; Bibliometrics; Stroke (engine); General surgery; Library science; Surgery","score_opus":0.03271331698919877,"score_gpt":0.32587625722451163,"score_spread":0.29316294023531286,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400532816","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98313445,0.003794118,0.00093539234,0.0091594225,0.00089560484,0.00006609631,0.000012531345,0.000013719397,0.0019886696],"genre_scores_gemma":[0.99626064,0.0009708528,0.00020848388,0.0003353194,0.0004074296,0.0000041898365,0.0000074410887,0.000009972675,0.0017956557],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9981706,0.000044994904,0.0006135198,0.00012192241,0.0008939812,0.00015496212],"domain_scores_gemma":[0.99812,0.0010349703,0.00014627291,0.00011188326,0.0005137844,0.000073051684],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0010528598,0.000095596086,0.0002730267,0.051837407,0.000028584114,0.00021458002,0.00018763926,0.000034502296,0.0002718084],"category_scores_gemma":[0.00081214757,0.000059961458,0.0003327612,0.040949035,0.000052610914,0.0002283307,0.0000616941,0.00026143147,0.000017347005],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004939253,0.00018530486,0.7774763,0.000016424978,0.008959041,0.0027854203,0.00040450142,0.0003124423,0.0031344763,0.0003704199,0.04340466,0.16245705],"study_design_scores_gemma":[0.0006129836,0.000054521173,0.8940065,0.00026341365,0.00067550933,0.00023580088,0.00072891463,0.006187736,0.0022550465,0.000040277813,0.09481154,0.00012776988],"about_ca_topic_score_codex":0.000029735027,"about_ca_topic_score_gemma":0.000006088614,"teacher_disagreement_score":0.16232929,"about_ca_system_score_codex":0.00022741663,"about_ca_system_score_gemma":0.00017742734,"threshold_uncertainty_score":0.97943604},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4400577762","doi":"10.1097/md.0000000000038831","title":"Research trends and hotspots in exercise interventions for liver cirrhosis: A bibliometric analysis via CiteSpace","year":2024,"lang":"en","type":"article","venue":"Medicine","topic":"Liver Disease Diagnosis and Treatment","field":"Medicine","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Cirrhosis; Web of science; Bibliometrics; Psychological intervention; MEDLINE; Gerontology; Internal medicine; Library science; Meta-analysis; Political science","score_opus":0.11867079245689881,"score_gpt":0.45327899122322246,"score_spread":0.33460819876632364,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400577762","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.74096537,0.23977806,0.0006477258,0.016337596,0.00018217831,0.00094856654,0.000059929516,0.000096227894,0.0009843748],"genre_scores_gemma":[0.9879606,0.008940708,0.00015658858,0.000067395835,0.00013391885,0.0005153112,0.000105732,0.000020407133,0.0020993687],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9984249,0.00005787819,0.0003050231,0.00044877344,0.0004486878,0.00031476034],"domain_scores_gemma":[0.9987307,0.0005349821,0.000026638236,0.00026876194,0.00016685315,0.00027205932],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00094941066,0.00014683837,0.00051035685,0.13255353,0.00006192325,0.00003838688,0.00006749803,0.000067410954,0.000870867],"category_scores_gemma":[0.00016619514,0.00010278544,0.00027424074,0.14889546,0.0001331463,0.000086913315,0.000049557395,0.00017435552,0.000020970423],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038862063,0.001412108,0.26280448,0.002719513,0.0023830426,0.0008954528,0.0016931383,0.000002882203,0.00010655884,0.0003258682,0.028624827,0.6986435],"study_design_scores_gemma":[0.0020596525,0.0010676191,0.98292816,0.0032880383,0.0047484273,0.0000095989835,0.00029626768,0.003276847,0.000044285156,0.00012060084,0.00206752,0.000092957715],"about_ca_topic_score_codex":0.000673651,"about_ca_topic_score_gemma":0.00017770806,"teacher_disagreement_score":0.7201237,"about_ca_system_score_codex":0.00013080853,"about_ca_system_score_gemma":0.000038005604,"threshold_uncertainty_score":0.9535383},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4400641897","doi":"10.3390/jrfm17070304","title":"Mapping Corporate Sustainability and Firm Performance Research: A Scientometric and Bibliometric Examination","year":2024,"lang":"en","type":"article","venue":"Journal of risk and financial management","topic":"Environmental Sustainability in Business","field":"Business, Management and Accounting","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Corporate sustainability; Sustainability; Corporate social responsibility; Scopus; Accounting; Business; Sustainability reporting; Sustainable development; Environmental economics; Environmental resource management; Economics; Political science; Public relations; Ecology","score_opus":0.03913030348832868,"score_gpt":0.2706565585191634,"score_spread":0.23152625503083474,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400641897","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99178773,0.0046389755,0.0022844393,0.00031669645,0.00028206542,0.00037077285,0.0000014826996,0.000020288218,0.00029757718],"genre_scores_gemma":[0.99378955,0.0054566516,0.00025176417,0.00004865273,0.00031504233,0.0000094136085,0.0000012467374,0.000013256142,0.00011440994],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99827236,0.000047241454,0.000403873,0.00031423967,0.000636736,0.0003255254],"domain_scores_gemma":[0.99900436,0.00016843058,0.00026699793,0.00013143414,0.00039251978,0.00003623367],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.005954875,0.00015321931,0.0002168418,0.03257191,0.0004034138,0.0008288981,0.00015207382,0.000059902173,0.000010871977],"category_scores_gemma":[0.0009192709,0.00013131942,0.00003725236,0.039895006,0.00033910314,0.0017968869,0.00055218424,0.00033153273,0.000004486762],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000032868622,0.00006694242,0.2821451,0.0023073277,0.000009823016,0.00008167389,0.00021045691,0.000030337917,0.000005111918,0.0029438848,0.00021386317,0.7119526],"study_design_scores_gemma":[0.0003678574,0.00005917183,0.93849117,0.00014823151,0.000045658307,0.0000150278365,0.001052575,0.0019930012,0.0000021509532,0.010529078,0.047159366,0.00013669833],"about_ca_topic_score_codex":0.00008270135,"about_ca_topic_score_gemma":0.00000410031,"teacher_disagreement_score":0.7118159,"about_ca_system_score_codex":0.00021732999,"about_ca_system_score_gemma":0.000027457041,"threshold_uncertainty_score":0.9805125},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4400747030","doi":"10.7759/cureus.64687","title":"Research Hotspots and Trends of Interventions for Sarcopenic Obesity: A Bibliometric Analysis","year":2024,"lang":"en","type":"review","venue":"Cureus","topic":"Nutrition and Health in Aging","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Sarcopenic obesity; Medicine; Psychological intervention; Obesity; Bibliometrics; Sarcopenia; Gerontology; Library science; Pathology; Nursing","score_opus":0.5312426614795842,"score_gpt":0.618895200093531,"score_spread":0.08765253861394684,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400747030","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000031333337,0.9974085,0.00007697784,0.00022515179,0.00014872766,0.0006990159,0.00022124416,0.000031413176,0.0011576089],"genre_scores_gemma":[0.00024914282,0.9925004,0.00067270506,0.000016814221,0.00016675089,0.00024075962,0.00023451533,0.000034994962,0.0058838637],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99806505,0.00014749999,0.0007550797,0.00038868154,0.00034951454,0.00029419886],"domain_scores_gemma":[0.998544,0.00040520247,0.00016314183,0.0003873551,0.00029641215,0.00020391971],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.001788095,0.00017599492,0.0018375485,0.17392296,0.0000835743,0.00004746341,0.00014340134,0.00021572084,0.00040438463],"category_scores_gemma":[0.00024844124,0.00013446077,0.0013745645,0.20651834,0.00008729285,0.00003735974,0.000123447,0.00052915607,0.000027848002],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000074899995,0.00017313576,0.000058679576,0.17935169,0.0014491754,0.000010311883,0.00002141874,5.573371e-9,1.4328954e-7,0.0003911305,0.013503521,0.80503327],"study_design_scores_gemma":[0.00031660814,0.00030058224,0.0009700338,0.0388074,0.017137738,0.000030008385,0.00003090602,0.000015329537,2.5371605e-7,0.00016865163,0.9421149,0.000107625274],"about_ca_topic_score_codex":0.000049544982,"about_ca_topic_score_gemma":0.000039975574,"teacher_disagreement_score":0.92861134,"about_ca_system_score_codex":0.00016362636,"about_ca_system_score_gemma":0.00019133557,"threshold_uncertainty_score":0.8354399},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4400986784","doi":"10.21037/mhealth-23-20","title":"The evolution and mapping trends of mobile health (m-Health): a bibliometric analysis (1997–2023)","year":2024,"lang":"en","type":"article","venue":"mHealth","topic":"Mobile Health and mHealth Applications","field":"Health Professions","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Environmental science; Geography; Environmental health; Computer science; Medicine","score_opus":0.05901089199329632,"score_gpt":0.4536944523763699,"score_spread":0.39468356038307356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400986784","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09764517,0.7064121,0.044451866,0.11376703,0.005479259,0.019030312,0.002108697,0.0016261662,0.009479417],"genre_scores_gemma":[0.9240054,0.053746253,0.0011196829,0.0036783163,0.0005452841,0.012385593,0.0002842943,0.000072410636,0.0041627497],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9927055,0.0013826916,0.0026081698,0.0008273559,0.0006619248,0.0018143402],"domain_scores_gemma":[0.99443495,0.0020913738,0.0010658569,0.00091960497,0.0002483677,0.0012398303],"candidate_categories":["bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0083261095,0.0003104081,0.0010057933,0.039509084,0.00352897,0.00004430492,0.00034253745,0.00024198845,0.00019186342],"category_scores_gemma":[0.00019174485,0.00023433464,0.00021933523,0.16289014,0.00018807888,0.00015734557,0.00015095441,0.0011492093,0.00012479702],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000346911,0.00009082059,0.014938341,0.005074419,0.00016853996,8.701199e-7,0.0019674113,0.000026445245,0.0000042675542,0.028653316,0.067362554,0.88167834],"study_design_scores_gemma":[0.00047389424,0.000568102,0.19050838,0.00041096087,0.000091763344,0.000004562761,0.0032222534,0.0051050875,3.5809165e-7,0.0010291684,0.798407,0.00017846428],"about_ca_topic_score_codex":0.012606877,"about_ca_topic_score_gemma":0.0047374615,"teacher_disagreement_score":0.8814999,"about_ca_system_score_codex":0.0015901749,"about_ca_system_score_gemma":0.0054936903,"threshold_uncertainty_score":0.9977683},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4401029491","doi":"10.1111/2047-3095.12484","title":"Clinical reasoning and clinical judgment in nursing research: A bibliometric analysis","year":2024,"lang":"en","type":"article","venue":"International Journal of Nursing Knowledge","topic":"Clinical Reasoning and Diagnostic Skills","field":"Medicine","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Trois-Rivières","funders":"","keywords":"Clinical judgment; Thematic analysis; Competence (human resources); Nursing; Psychology; Nursing process; Nurse education; Critical thinking; MEDLINE; Medicine; Medical education; Qualitative research; Pedagogy; Social psychology; Sociology","score_opus":0.19113135818044688,"score_gpt":0.5874985959632787,"score_spread":0.3963672377828318,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401029491","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9558897,0.019817064,0.0014637997,0.0055509303,0.0051798304,0.00012462548,0.0000045355355,0.000028502813,0.0119410185],"genre_scores_gemma":[0.99108326,0.0039508794,0.0023859923,0.00007677318,0.0021627853,0.0000014512028,0.000006661982,0.000021582575,0.00031059954],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9959498,0.0005457089,0.001794607,0.00038999642,0.0010075932,0.00031229726],"domain_scores_gemma":[0.97421277,0.023492778,0.0002509034,0.00019701393,0.0013483049,0.0004982287],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.009917001,0.00014152135,0.00070038217,0.043900274,0.000044902317,0.0002130962,0.0002626255,0.00021328978,0.00006165816],"category_scores_gemma":[0.05089252,0.00011551012,0.0006192264,0.030001182,0.00049357145,0.00016032677,0.000052343512,0.0015535557,0.000034669833],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035044324,0.0020037317,0.3039543,0.0000075582157,0.000788884,0.0006744666,0.00048119604,0.000006827763,0.0000059392605,0.00062823755,0.0054527,0.6856457],"study_design_scores_gemma":[0.0021510439,0.0009939363,0.94747615,0.03365878,0.0011326813,0.0006778654,0.0003291514,0.009122889,0.000008637564,0.0026224453,0.0017016771,0.00012472915],"about_ca_topic_score_codex":0.000010214786,"about_ca_topic_score_gemma":0.000005594855,"teacher_disagreement_score":0.685521,"about_ca_system_score_codex":0.0005961909,"about_ca_system_score_gemma":0.00055265555,"threshold_uncertainty_score":0.9906166},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4401327074","doi":"10.3390/su16156645","title":"Research Trends and Development Patterns in Microgreens Publications: A Bibliometric Study from 2004 to 2023","year":2024,"lang":"en","type":"article","venue":"Sustainability","topic":"Climate Change Communication and Perception","field":"Social Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Comisión Nacional de Investigación Científica y Tecnológica","keywords":"Library science; Analytics; Sustainable development; Web of science; Productivity; Bibliometrics; Period (music); Political science; Data science; Computer science; MEDLINE; Economic growth","score_opus":0.5179455909136175,"score_gpt":0.567033687857759,"score_spread":0.04908809694414151,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401327074","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9736661,0.0007642749,0.000036277383,0.023125507,0.00005842164,0.0006654111,0.000023723485,0.00007680637,0.0015835284],"genre_scores_gemma":[0.9972955,0.00032413582,0.000090937,0.000034417088,0.000049631042,0.0003206158,0.000037004622,0.000006360554,0.0018413931],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9979101,0.00077704404,0.00022598017,0.0003581602,0.00040110628,0.0003276057],"domain_scores_gemma":[0.9983588,0.00052723987,0.000013904496,0.00036978078,0.00058580766,0.00014445273],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0049407184,0.000068481146,0.00009762047,0.017130982,0.00039678958,0.00042027875,0.00029095385,0.00006296608,0.0011520796],"category_scores_gemma":[0.0011897575,0.00006903741,0.000016876415,0.0528446,0.00012117845,0.00020546245,0.00026311178,0.00022322657,0.000055401983],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000005805499,0.00022837242,0.5189707,0.0000231504,0.0000044227804,0.0000018347187,0.08483091,3.5156626e-7,0.0000023871069,0.00019283372,0.00088502694,0.3948542],"study_design_scores_gemma":[0.00007390545,0.000026510848,0.7568359,0.0000111009,0.0000013564892,5.0952128e-8,0.10936803,0.0000070719343,6.453683e-7,0.0006698825,0.13294548,0.000060069866],"about_ca_topic_score_codex":0.038040575,"about_ca_topic_score_gemma":0.17073473,"teacher_disagreement_score":0.39479414,"about_ca_system_score_codex":0.0017545285,"about_ca_system_score_gemma":0.00053473044,"threshold_uncertainty_score":0.999761},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4401345950","doi":"10.3389/fneur.2024.1442145","title":"Tracing the evolving dynamics and research hotspots of spinal cord injury and surgical decompression from 1975 to 2024: a bibliometric analysis","year":2024,"lang":"en","type":"review","venue":"Frontiers in Neurology","topic":"Spinal Cord Injury Research","field":"Medicine","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Spinal cord injury; Tracing; Medicine; Decompression; Surgical decompression; Spinal cord; Physical medicine and rehabilitation; Surgery; Computer science","score_opus":0.09328122166089482,"score_gpt":0.475684452773737,"score_spread":0.3824032311128422,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401345950","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09084027,0.90652764,0.00018003986,0.00036960698,0.00063727197,0.0011554824,0.00010600939,0.00001787888,0.00016582177],"genre_scores_gemma":[0.014036606,0.98440456,0.000795983,0.000051235613,0.00018492328,0.00011417897,0.000040261715,0.00006572827,0.00030652864],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99495876,0.0013655502,0.00092084287,0.001129133,0.0009189936,0.00070670166],"domain_scores_gemma":[0.99743295,0.0011324784,0.00016823692,0.00073564216,0.00021519753,0.00031549478],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0024097376,0.00039872166,0.0026305749,0.1065969,0.000110573594,0.00009333802,0.00050578755,0.0005697957,0.00006616295],"category_scores_gemma":[0.0010165977,0.00026582179,0.00034319784,0.12351355,0.00051744154,0.00006682137,0.0010135388,0.0032492783,0.000008145744],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005160624,0.00005779779,0.011354449,0.004086571,0.0006419299,0.0008437914,0.000017650538,6.31406e-7,0.0000018505034,0.000012403692,0.0057167895,0.9721055],"study_design_scores_gemma":[0.001944403,0.045444075,0.13666523,0.019657398,0.016828854,0.0013066338,0.00025868864,0.026885321,0.0000042807183,0.00082504284,0.74886304,0.0013170529],"about_ca_topic_score_codex":0.0004335009,"about_ca_topic_score_gemma":0.00007648628,"teacher_disagreement_score":0.9707885,"about_ca_system_score_codex":0.0001417027,"about_ca_system_score_gemma":0.00028731543,"threshold_uncertainty_score":0.9999794},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4401451763","doi":"10.1097/md.0000000000039241","title":"Looking backward toward the future: A bibliometric analysis of the last 40 years of meningioma global outcomes","year":2024,"lang":"en","type":"article","venue":"Medicine","topic":"Meningioma and schwannoma management","field":"Medicine","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Citation; Bibliometrics; Informetrics; Citation analysis; Meningioma; Web of science; Data science; Library science; Medical physics; Regional science; Computer science; Radiology; Geography; Pathology; Meta-analysis","score_opus":0.024348398187933128,"score_gpt":0.31084378892656284,"score_spread":0.28649539073862973,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401451763","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97006,0.00854978,0.00049059745,0.0155936675,0.0010550023,0.00033662634,0.00001576339,0.000048713428,0.003849855],"genre_scores_gemma":[0.99886346,0.00006363736,0.0001447197,0.00022670058,0.0002793488,0.000009513849,0.000011861318,0.000012089995,0.0003886524],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9977683,0.00007918074,0.0005678268,0.000271351,0.0010752213,0.00023814122],"domain_scores_gemma":[0.99877596,0.00023041607,0.00016396643,0.00063524116,0.00010944251,0.00008498761],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0009677701,0.00016990717,0.0006932797,0.012003561,0.000039521972,0.000016466169,0.00035049848,0.00006272317,0.0003424055],"category_scores_gemma":[0.00026854477,0.00008056352,0.00041984773,0.11046128,0.00023614409,0.000042845866,0.00016668874,0.00016343922,0.000009381482],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020607609,0.00018097955,0.81370443,0.0010790227,0.014606811,0.000856544,0.0057055578,0.00009540286,0.0017280006,0.007824596,0.030122265,0.12389034],"study_design_scores_gemma":[0.00063293875,0.00020280192,0.9757258,0.00027616016,0.005538904,0.00004756944,0.0011928196,0.0008284851,0.00012249002,0.00010340658,0.015250266,0.00007834442],"about_ca_topic_score_codex":0.00013291772,"about_ca_topic_score_gemma":0.000017400322,"teacher_disagreement_score":0.16202141,"about_ca_system_score_codex":0.000070583876,"about_ca_system_score_gemma":0.000060193302,"threshold_uncertainty_score":0.99919456},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4401545639","doi":"10.3390/jrfm17080352","title":"Bibliometric Analysis of the Machine Learning Applications in Fraud Detection on Crowdfunding Platforms","year":2024,"lang":"en","type":"article","venue":"Journal of risk and financial management","topic":"FinTech, Crowdfunding, Digital Finance","field":"Business, Management and Accounting","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"","funders":"Universidad Nacional de Colombia","keywords":"Computer science; Data science","score_opus":0.009459572617702862,"score_gpt":0.22441669447182735,"score_spread":0.2149571218541245,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401545639","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9413723,0.0010608656,0.05472945,0.00003992409,0.0003613389,0.00020607325,0.000005589111,0.000023369117,0.0022011301],"genre_scores_gemma":[0.9989187,0.0006731029,0.0000657391,0.000043041062,0.00017825614,0.000008338264,0.0000023020407,0.000013233932,0.00009726983],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99881953,0.0000062981503,0.0004950942,0.0001786879,0.000329093,0.00017130237],"domain_scores_gemma":[0.9992319,0.00009355669,0.0004748833,0.0001187971,0.00007085916,0.000010014436],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00072732824,0.0001369895,0.0002755694,0.042354338,0.00016172646,0.00025184962,0.00021461038,0.000053017222,0.000010984805],"category_scores_gemma":[0.00020908534,0.00009476673,0.00021564592,0.0802012,0.000041216474,0.0006513433,0.00015259976,0.00039085728,0.000006336186],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007156167,0.000109036526,0.33290556,0.0003273312,0.00016139213,0.000020681915,0.00008312818,0.004622621,0.00005187279,0.039805133,0.00008556095,0.62175614],"study_design_scores_gemma":[0.00027325613,0.0000418442,0.91144866,0.00024744042,0.0005960042,0.0000017007166,0.0001091252,0.009228465,0.0000863117,0.006240948,0.07158367,0.00014255667],"about_ca_topic_score_codex":0.00014695633,"about_ca_topic_score_gemma":0.00019320333,"teacher_disagreement_score":0.62161356,"about_ca_system_score_codex":0.000070802256,"about_ca_system_score_gemma":0.00000826576,"threshold_uncertainty_score":0.9684998},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4401553417","doi":"10.1111/appy.12564","title":"Bipolar disorder and oxidative stress: A bibliometric perspective","year":2024,"lang":"en","type":"article","venue":"Asia-Pacific Psychiatry","topic":"Tryptophan and brain disorders","field":"Neuroscience","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Universidade de Macau","keywords":"Perspective (graphical); Oxidative stress; Psychology; Chemistry; Computer science; Biochemistry; Artificial intelligence","score_opus":0.014874797055303032,"score_gpt":0.27652415313103884,"score_spread":0.2616493560757358,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401553417","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.77969354,0.11297062,0.0010651347,0.012380666,0.0049682977,0.00089142355,0.00041324247,0.0011719198,0.086445145],"genre_scores_gemma":[0.9956457,0.0020858266,0.0004119316,0.00024265803,0.00021263245,0.00003185433,0.0000047055946,0.000054850338,0.001309841],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99789524,0.00012003251,0.00022395552,0.0009418008,0.00037056833,0.000448387],"domain_scores_gemma":[0.9991263,0.00026971794,0.0000540906,0.0003464987,0.000036626698,0.00016678023],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00012842352,0.00029158616,0.00020964061,0.009407952,0.00022910417,0.0003614685,0.00028403007,0.00012140962,0.00023529906],"category_scores_gemma":[0.00026316696,0.00025260446,0.000120738485,0.029241286,0.00030237137,0.00047908697,0.00009340473,0.0003900139,0.00031597048],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024162851,0.002071098,0.081559785,0.0007456881,0.0003003865,0.0001613668,0.010701278,0.00001207645,0.070281036,0.6249033,0.0672802,0.14174213],"study_design_scores_gemma":[0.004223962,0.0017297714,0.13517202,0.00089028064,0.00019363085,0.0004985456,0.13953196,0.0020312734,0.03553429,0.14147864,0.53498256,0.0037330524],"about_ca_topic_score_codex":0.00006545296,"about_ca_topic_score_gemma":0.000044224238,"teacher_disagreement_score":0.4834247,"about_ca_system_score_codex":0.000046641595,"about_ca_system_score_gemma":0.00014138936,"threshold_uncertainty_score":0.9999926},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4401573182","doi":"10.47863/gojv8413","title":"Social media, body image and mental health: a bibliometric analysis","year":2024,"lang":"en","type":"article","venue":"","topic":"Eating Disorders and Behaviors","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Mental health; Social media; Image (mathematics); Psychology; Data science; Sociology; Computer science; Artificial intelligence; Psychiatry; World Wide Web","score_opus":0.03344260863435955,"score_gpt":0.39649760045438,"score_spread":0.36305499182002043,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401573182","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.973539,0.0051920237,0.00026168342,0.0040200003,0.00044939388,0.00010740271,0.000042567524,0.00018340495,0.016204553],"genre_scores_gemma":[0.9969959,0.00008759893,0.00022409287,0.00022194353,0.000088259265,0.000011318268,0.00005443052,0.000012177973,0.002304271],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9991701,0.000048916896,0.00016439585,0.00025853902,0.00014466235,0.00021335809],"domain_scores_gemma":[0.99972564,0.0000845024,0.000023885003,0.000082729464,0.000013363156,0.00006989722],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00021030725,0.00008391809,0.00017426402,0.029668473,0.00010946718,0.00010379883,0.00005714965,0.000053668482,0.0044281697],"category_scores_gemma":[0.000009408715,0.00006931393,0.00010450152,0.07664453,0.000054882,0.00005604645,0.000029065088,0.00009863218,0.00017015527],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019617892,0.00051712943,0.20594743,0.00005673434,0.0010475154,0.000050702038,0.015861321,3.7540534e-8,0.00041609694,0.0070509315,0.5073281,0.2617044],"study_design_scores_gemma":[0.00045087168,0.00015894083,0.98378813,0.0000058943247,0.00031673055,0.000008572045,0.0035749422,0.00012182122,0.0000066933144,0.00030440922,0.01103712,0.00022588839],"about_ca_topic_score_codex":0.00066484837,"about_ca_topic_score_gemma":0.00025141714,"teacher_disagreement_score":0.7778407,"about_ca_system_score_codex":0.00001856753,"about_ca_system_score_gemma":0.000016108314,"threshold_uncertainty_score":0.9964819},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4401649191","doi":"10.70232/rq1cmx05","title":"Green Chemistry in Education: A Bibliometric Study and Research Trends (2002–2022)","year":2024,"lang":"en","type":"article","venue":"Journal of Education for Sustainable Development Studies.","topic":"Chemistry and Chemical Engineering","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Mathematics education; Chemistry; Library science; Computer science; Psychology","score_opus":0.027858543276737542,"score_gpt":0.35609540484380187,"score_spread":0.32823686156706433,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401649191","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9814787,0.011636024,0.000015186614,0.0007474401,0.00023377987,0.00019373772,3.0439125e-7,0.000009619527,0.005685235],"genre_scores_gemma":[0.9483392,0.00032528094,0.0005556162,0.000020383777,0.00016169241,0.0001367185,0.0000022382258,0.000012616573,0.05044622],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99870765,0.000014811124,0.0003984146,0.0002112051,0.0003797164,0.0002882259],"domain_scores_gemma":[0.9993454,0.0001568687,0.00006697522,0.00009024556,0.00022877683,0.00011170185],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013458289,0.00011031722,0.00016814208,0.0030474346,0.00013558276,0.000074055686,0.00016383361,0.0000431575,0.00028802344],"category_scores_gemma":[0.00044882824,0.00009912764,0.000030969088,0.01733489,0.00005803016,0.00028666446,0.00016274898,0.0002548515,0.0000039497345],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009148353,0.0031151467,0.022519387,0.0031206433,0.00031922624,0.000084606145,0.024272187,0.00011694349,0.003701541,0.00011155876,0.2127854,0.7297619],"study_design_scores_gemma":[0.0007802195,0.00015977751,0.04491287,0.00034153124,0.000049521488,0.00010136269,0.19286965,0.00007554505,0.00767937,0.0015497301,0.75106966,0.00041078284],"about_ca_topic_score_codex":0.000028883253,"about_ca_topic_score_gemma":0.0000031243694,"teacher_disagreement_score":0.7293511,"about_ca_system_score_codex":0.0013719854,"about_ca_system_score_gemma":0.00061577774,"threshold_uncertainty_score":0.832884},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4401651746","doi":"10.70232/nveahh04","title":"Global Trends Research and Application of Green Chemistry and Education: A Bibliometric Analysis (1994–2023)","year":2024,"lang":"en","type":"article","venue":"Journal of Education for Sustainable Development Studies.","topic":"Chemistry and Chemical Engineering","field":"Environmental Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Regional science; Library science; Environmental chemistry; Political science; Chemistry; Sociology; Computer science","score_opus":0.018321944450295644,"score_gpt":0.34515739183843824,"score_spread":0.3268354473881426,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401651746","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98020583,0.013602351,0.0010053024,0.00060929335,0.00008060226,0.00012658475,0.0000025620743,0.00000835337,0.0043591326],"genre_scores_gemma":[0.9864928,0.00036328952,0.0019861,0.000010305499,0.00008084344,0.00005647759,0.000010181994,0.0000052490595,0.010994743],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990194,0.0000061654073,0.00033974616,0.00017956263,0.00028222878,0.00017293448],"domain_scores_gemma":[0.9993444,0.00009870913,0.000109716675,0.00008380256,0.00026133552,0.00010205329],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0006163776,0.00008775703,0.00018047592,0.0021175197,0.00011390122,0.000042001342,0.000109865614,0.0000458921,0.00005102745],"category_scores_gemma":[0.00015400577,0.00007945325,0.000043710872,0.031662468,0.00010976674,0.00018255318,0.00011945051,0.00008702028,5.1625426e-7],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000085223925,0.0008792145,0.06768001,0.005176596,0.0017856113,0.000005973015,0.0035441516,0.0004204094,0.0050787306,0.0009570352,0.045303583,0.86908346],"study_design_scores_gemma":[0.00089509203,0.00021133154,0.28196847,0.0003836293,0.0010498927,0.00015893883,0.06402941,0.0015700684,0.022829577,0.016312022,0.6097675,0.00082409364],"about_ca_topic_score_codex":0.000048854326,"about_ca_topic_score_gemma":0.0000031822622,"teacher_disagreement_score":0.8682594,"about_ca_system_score_codex":0.00067124947,"about_ca_system_score_gemma":0.0002818512,"threshold_uncertainty_score":0.98892003},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4401788540","doi":"10.21037/jtd-24-597","title":"Characteristics of the top 100 cited original studies on extracorporeal membrane oxygenation: a bibliometric analysis","year":2024,"lang":"en","type":"article","venue":"Journal of Thoracic Disease","topic":"Mechanical Circulatory Support Devices","field":"Engineering","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Extracorporeal membrane oxygenation; Internal medicine","score_opus":0.035883432678338524,"score_gpt":0.33581934764073457,"score_spread":0.299935914962396,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401788540","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9924773,0.0049676723,0.0008078598,0.00018476981,0.0012360663,0.00008975275,0.00008523107,0.00004871578,0.000102614176],"genre_scores_gemma":[0.9993039,0.00017969168,0.00011390382,0.000042237356,0.0002573323,0.000002924968,0.0000061161804,0.00002107665,0.00007285138],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9981084,0.00007390291,0.000758188,0.00013071515,0.0007789468,0.00014986926],"domain_scores_gemma":[0.9987139,0.00022687255,0.00027209893,0.00027956098,0.0002786639,0.00022890276],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00067310926,0.0001614861,0.00042735395,0.016728342,0.000038790582,0.0000788467,0.00030487656,0.000045984187,0.00025008933],"category_scores_gemma":[0.00036352748,0.000103762155,0.00048153012,0.046421904,0.000038257585,0.00023082923,0.000029152205,0.00026755343,0.000013077407],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001913099,0.0018870895,0.5220592,0.027361646,0.05258184,0.0058476236,0.001724582,0.1512812,0.007905119,0.009311174,0.008051658,0.2100758],"study_design_scores_gemma":[0.00020926887,0.000115109346,0.93712986,0.0010676897,0.007906619,0.000030446057,0.000092782386,0.04630492,0.0026141377,0.00092360127,0.0032151327,0.00039045032],"about_ca_topic_score_codex":5.3442e-7,"about_ca_topic_score_gemma":5.471103e-7,"teacher_disagreement_score":0.41507065,"about_ca_system_score_codex":0.00009245689,"about_ca_system_score_gemma":0.00009494146,"threshold_uncertainty_score":0.9944162},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4401845731","doi":"10.5530/jscires.13.2.44","title":"Scientometric Insights into the Trends and Evolution of the Underwater Sensing Technologies and their Applications","year":2024,"lang":"en","type":"article","venue":"Journal of Scientometric Research","topic":"Water Quality Monitoring Technologies","field":"Environmental Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Defence Research and Development Canada; National Research Council Canada","funders":"","keywords":"Underwater; Data science; Geography; Computer science; Oceanography; Geology","score_opus":0.08102982158710223,"score_gpt":0.3691248103624888,"score_spread":0.28809498877538653,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401845731","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9853584,0.007710184,0.0018270382,0.0044791247,0.00015896582,0.00015357666,8.255153e-7,0.000042302017,0.0002695705],"genre_scores_gemma":[0.99835294,0.00045155422,0.00090148795,0.000002750907,0.000028261158,0.0000031509323,5.8081362e-8,0.000006320253,0.00025345996],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99753344,0.00017656064,0.00032734455,0.0002593041,0.0014079649,0.0002953907],"domain_scores_gemma":[0.9985677,0.0007328401,0.00011887202,0.00039521308,0.00013339115,0.000051987256],"candidate_categories":["bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.005378349,0.0000958584,0.00014126903,0.008052237,0.00052379613,0.00032786987,0.0009682003,0.0000872613,0.00000518513],"category_scores_gemma":[0.0009620497,0.000041645624,0.000056874745,0.048211083,0.003097925,0.00037941715,0.0015399392,0.00059648306,0.000005704856],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009471361,0.00006948706,0.047137868,0.00006962753,0.00004250962,0.0000043101422,0.005135504,0.00008242368,0.0883909,0.001457139,0.0010345831,0.8565662],"study_design_scores_gemma":[0.00042794002,0.0005901423,0.5783176,0.00029964384,0.000045001965,0.00025495532,0.02998245,0.0048469333,0.224649,0.12855558,0.03170101,0.00032980237],"about_ca_topic_score_codex":0.00014421264,"about_ca_topic_score_gemma":0.000012570168,"teacher_disagreement_score":0.8562364,"about_ca_system_score_codex":0.00052455976,"about_ca_system_score_gemma":0.00004621328,"threshold_uncertainty_score":0.9996151},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4401847930","doi":"10.1108/ci-08-2023-0210","title":"Where lean construction and offsite construction meet: a bibliographic scientometric analysis","year":2024,"lang":"en","type":"article","venue":"Construction Innovation","topic":"BIM and Construction Integration","field":"Engineering","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Lean construction; Computer science; Engineering; Construction engineering; Construction industry","score_opus":0.010108373769357018,"score_gpt":0.23181314956201535,"score_spread":0.22170477579265832,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401847930","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7106038,0.0033260954,0.27237448,0.0002331001,0.0061162165,0.00034911287,0.00006848538,0.0019520893,0.004976666],"genre_scores_gemma":[0.98154163,0.0013658964,0.016378041,0.000028470755,0.0003647119,0.00005933614,0.00013298026,0.000048245554,0.00008065884],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99769884,0.000043716147,0.00086679263,0.0005993931,0.0004568322,0.0003344205],"domain_scores_gemma":[0.9988407,0.00006545474,0.00015236472,0.00029575624,0.00054804643,0.000097671],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00043039868,0.00035459327,0.00035414682,0.027461145,0.00031665977,0.0006496784,0.00010693474,0.00030770962,0.00036688824],"category_scores_gemma":[0.00004534062,0.00037616777,0.0001631529,0.07331111,0.00070994586,0.001331143,0.000026372532,0.00038476332,0.000034829292],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000015501926,0.000009375158,0.048622,0.0002633593,0.00080212223,0.0000031381371,0.000120499324,0.00086809974,0.010002147,0.39800352,0.0010812969,0.54020894],"study_design_scores_gemma":[0.004708643,0.0005157401,0.098927766,0.0015471122,0.0068357443,0.008279701,0.0095841875,0.45536906,0.04452178,0.122690395,0.24084458,0.0061753113],"about_ca_topic_score_codex":0.000026708472,"about_ca_topic_score_gemma":0.00004752824,"teacher_disagreement_score":0.53403366,"about_ca_system_score_codex":0.00016526674,"about_ca_system_score_gemma":0.0000745812,"threshold_uncertainty_score":0.99986905},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4402055545","doi":"10.18280/acsm.480404","title":"Mapping Publications of Cracks Monitoring in Concrete Structures: Bibliometric and Scientometric Review in 2013-2023","year":2024,"lang":"en","type":"article","venue":"Annales de Chimie Science des Matériaux","topic":"Infrastructure Maintenance and Monitoring","field":"Engineering","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Environmental science; Engineering; Library science; Computer science","score_opus":0.03985074345792553,"score_gpt":0.31253312015298595,"score_spread":0.2726823766950604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402055545","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.91595817,0.08072608,0.0010681272,0.0001061382,0.00064890593,0.00021228424,0.000008471035,0.00010508209,0.0011667607],"genre_scores_gemma":[0.95883095,0.03676778,0.004231835,0.000016088221,0.000084814514,0.0000245262,0.0000016240554,0.000016317157,0.000026072341],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99833405,0.000019825902,0.000393305,0.00032132067,0.0003780419,0.0005534406],"domain_scores_gemma":[0.9993429,0.00013036797,0.000042985226,0.00023089386,0.0001298884,0.00012296444],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0014550228,0.0001572576,0.00024501752,0.037952837,0.000084684405,0.0002371141,0.00041252817,0.000055158158,0.000018680337],"category_scores_gemma":[0.0005944077,0.00014480947,0.00004101202,0.10745018,0.0003720256,0.00088499754,0.000104167186,0.00026958477,0.0000069582793],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009524245,0.00002071703,0.5371742,0.020952683,0.00007320692,0.00008194014,0.0067140562,0.016111838,0.10766488,0.006410553,0.0034249516,0.30136144],"study_design_scores_gemma":[0.00017537583,0.000024911096,0.9660355,0.005420573,0.000010690308,0.0000613108,0.00055088685,0.017076531,0.005543353,0.0023001332,0.0024698677,0.00033087254],"about_ca_topic_score_codex":0.000071593364,"about_ca_topic_score_gemma":0.000008722627,"teacher_disagreement_score":0.4288613,"about_ca_system_score_codex":0.00022872684,"about_ca_system_score_gemma":0.00014935528,"threshold_uncertainty_score":0.9729512},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4402071558","doi":"10.3389/fmed.2024.1457366","title":"Assessment of h-index and associated demographic and academic parameters for academic hematologists in Canada","year":2024,"lang":"en","type":"article","venue":"Frontiers in Medicine","topic":"Health and Medical Research Impacts","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Ottawa Hospital; Université de Montréal; McGill University Health Centre","funders":"Canadian Institutes of Health Research","keywords":"Index (typography); Medicine; Demography; Sociology; Computer science","score_opus":0.07571432815318722,"score_gpt":0.4237392277548893,"score_spread":0.3480248996017021,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402071558","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9452196,0.022337269,0.0025280719,0.028245246,0.0005380694,0.0009894139,0.000016770866,0.000017088163,0.00010849536],"genre_scores_gemma":[0.9854382,0.010578746,0.0012473984,0.002551665,0.000044051132,0.00006008834,0.000022034983,0.00001565776,0.000042106993],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9976721,0.00009934506,0.0005971404,0.00030992023,0.0006083759,0.000713132],"domain_scores_gemma":[0.99759334,0.0008761577,0.00006932672,0.000109675864,0.00004314862,0.0013083519],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0028257624,0.00014569095,0.0007763361,0.0006054069,0.000019929095,0.0000023079879,0.00008052508,0.00025625597,0.000006363788],"category_scores_gemma":[0.015114438,0.00010790857,0.000021633845,0.00064646575,0.00034221468,0.00005153767,0.000027872926,0.0013306638,4.2879115e-8],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000116300165,0.000013791013,0.92503977,0.0016235103,0.000053167456,0.00017196717,0.00017510407,0.0000018710796,0.0001379757,0.00007093237,0.043942597,0.028653001],"study_design_scores_gemma":[0.0061072866,0.0007133267,0.9335402,0.006537728,0.000092893664,0.000070790156,0.0020876378,0.04442569,0.000055682103,0.0019529733,0.004251769,0.00016399904],"about_ca_topic_score_codex":0.12425192,"about_ca_topic_score_gemma":0.057962373,"teacher_disagreement_score":0.06628955,"about_ca_system_score_codex":0.0006973762,"about_ca_system_score_gemma":0.0052967994,"threshold_uncertainty_score":0.99318165},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4402094343","doi":"10.4103/joah.joah_27_23","title":"A Bibliometric Analysis of Anemia Research in Children or Adolescents in the Last 10 Years","year":2023,"lang":"en","type":"article","venue":"Journal of Applied Hematology","topic":"Iron Metabolism and Disorders","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Anemia; Pediatrics; Family medicine; Medical education; Psychiatry","score_opus":0.04376658616997898,"score_gpt":0.3705476725145289,"score_spread":0.32678108634454994,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402094343","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.998302,0.00043006305,0.0000025894851,0.0002792343,0.000025664252,0.00019178932,0.0000021228745,0.000003585604,0.00076294475],"genre_scores_gemma":[0.9992372,0.0005030715,0.0000766349,0.00011019257,0.000033476143,0.0000050327762,0.000005764503,0.000008445589,0.000020181726],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99830383,0.00011758674,0.0005761284,0.00012061687,0.00058670883,0.00029514098],"domain_scores_gemma":[0.9992002,0.00024238514,0.00019475655,0.0002050168,0.000098626006,0.00005900224],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0021664298,0.000073804484,0.0006258602,0.10837737,0.000016962387,0.000007511483,0.00025594566,0.000114485905,0.00011508798],"category_scores_gemma":[0.00020903882,0.00004633801,0.000103994695,0.18129314,0.000106488194,0.000027451078,0.000050735438,0.0004925156,0.000036857673],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012538392,0.00044763385,0.98541075,0.00014569945,0.00012553873,0.00030482849,0.0010452993,0.0001221019,0.0002779473,0.00064772944,0.0040984545,0.0061201495],"study_design_scores_gemma":[0.0023462446,0.00009728041,0.99508595,0.000046362744,0.00018697347,0.00012584667,0.0016283211,0.000019793075,0.00007228843,0.00019994673,0.00015196086,0.00003905514],"about_ca_topic_score_codex":0.000036775546,"about_ca_topic_score_gemma":0.0000740364,"teacher_disagreement_score":0.07291578,"about_ca_system_score_codex":0.0000210083,"about_ca_system_score_gemma":0.00014259214,"threshold_uncertainty_score":0.9017284},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4402369529","doi":"10.19040/ecocycles.v10i2.442","title":"Perspective of ecocycles for human well-being and health: A bibliometric analysis","year":2024,"lang":"en","type":"article","venue":"Ecocycles","topic":"Innovative Approaches in Technology and Social Development","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Perspective (graphical); Bibliometrics; Sociology; Data science; Computer science; Library science; Artificial intelligence","score_opus":0.03954235724543098,"score_gpt":0.32448094793386684,"score_spread":0.28493859068843586,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402369529","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9614654,0.0022436415,0.003998846,0.0044906354,0.00022645657,0.000414223,0.00000618079,0.0002613749,0.026893241],"genre_scores_gemma":[0.99799484,0.000062216975,0.0011646184,0.00030705536,0.00021932153,0.00004118384,0.00002335139,0.000017401253,0.00017002739],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9990468,0.0000072663565,0.000297548,0.000308655,0.00011750144,0.0002222178],"domain_scores_gemma":[0.9995292,0.00006683757,0.00013803269,0.0001117425,0.00014713724,0.000007084442],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.000781095,0.00013053075,0.00030752626,0.032190956,0.00029823912,0.00013496587,0.000141706,0.00008784183,0.000059359434],"category_scores_gemma":[0.00006686031,0.0001236973,0.00011483895,0.04612154,0.0001373667,0.00032634556,0.00013675269,0.000113421345,0.000012261995],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000005100435,0.000041188985,0.021557711,0.00038408043,0.0006204209,0.0000017931314,0.00064116134,0.00000449605,0.00010225357,0.95446557,0.0013346209,0.020841587],"study_design_scores_gemma":[0.00030775383,0.000027183189,0.43031082,0.00009185593,0.0003108139,6.961915e-7,0.004259531,0.0016192028,0.0005820313,0.5513213,0.0108476505,0.0003211381],"about_ca_topic_score_codex":0.0004337047,"about_ca_topic_score_gemma":0.00015961406,"teacher_disagreement_score":0.4087531,"about_ca_system_score_codex":0.00008253742,"about_ca_system_score_gemma":0.000021818341,"threshold_uncertainty_score":0.97877836},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4402390234","doi":"10.24818/rfb.23.16.01.05","title":"Mapping the Intellectual Structure of Asset Pricing: A Bibliometric Study","year":2024,"lang":"en","type":"article","venue":"The Review of Finance and Banking","topic":"Complex Systems and Time Series Analysis","field":"Economics, Econometrics and Finance","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Capital asset pricing model; Business; Asset (computer security); Computer science; Finance","score_opus":0.050324134458127744,"score_gpt":0.25964797832192266,"score_spread":0.2093238438637949,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402390234","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.34430024,0.6533676,0.00054091844,0.0003986581,0.0001071278,0.00033259916,0.000030594398,0.0000075336748,0.0009147596],"genre_scores_gemma":[0.89504886,0.10467924,0.000040109022,0.00006677122,0.000050196304,0.000006048765,0.0000016153108,0.0000085088395,0.00009867995],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99881643,0.00003737337,0.00072799355,0.00021858992,0.00006641719,0.00013322213],"domain_scores_gemma":[0.9989569,0.00028149434,0.0003523893,0.00035967724,0.00004027483,0.000009275287],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011641192,0.00011515149,0.00057756645,0.00205604,0.000107061904,0.00005486627,0.00028545316,0.000023384608,0.00027680458],"category_scores_gemma":[0.00017081977,0.0000659721,0.00015482158,0.01746636,0.00006312992,0.000088779045,0.00011562266,0.00013770668,0.0000101070345],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021137943,0.00022732707,0.033898484,0.0385621,0.0019610976,0.0000140699285,0.022227218,0.00010010589,0.00022523514,0.25571388,0.009058801,0.63799053],"study_design_scores_gemma":[0.00047225112,0.0005812164,0.11660063,0.022628779,0.0003654741,0.00007686417,0.0026403149,0.008598667,0.00011458806,0.024095878,0.8230696,0.0007557659],"about_ca_topic_score_codex":0.00021736042,"about_ca_topic_score_gemma":0.000013101514,"teacher_disagreement_score":0.81401074,"about_ca_system_score_codex":0.000014142312,"about_ca_system_score_gemma":0.000014691847,"threshold_uncertainty_score":0.8392007},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4402423426","doi":"10.1016/j.heliyon.2024.e37675","title":"A bibliometric and knowledge-map analysis of psychological violence from 2003 to 2022","year":2024,"lang":"en","type":"review","venue":"Heliyon","topic":"Intimate Partner and Family Violence","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Gannan Medical University","keywords":"Psychology; Applied psychology","score_opus":0.09894655831542816,"score_gpt":0.46535355777083964,"score_spread":0.3664069994554115,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402423426","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000097887656,0.9817916,0.00007180889,0.0000066700945,0.0008337823,0.00045465215,0.00035441795,0.000054804044,0.016334353],"genre_scores_gemma":[0.00015709187,0.9983583,0.00021473764,0.000048270944,0.00020496671,0.000087066386,0.000044145978,0.00001787567,0.0008675586],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99744207,0.00043364454,0.0005711142,0.0007380082,0.00043607526,0.000379076],"domain_scores_gemma":[0.99870706,0.00030050476,0.00016405455,0.0003718016,0.00022726906,0.000229332],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00093815697,0.00029187786,0.0015222827,0.029166514,0.00013483726,0.00012117128,0.0005510646,0.0003982959,0.00064919423],"category_scores_gemma":[0.000392686,0.00022081399,0.00040320013,0.18459462,0.00021245296,0.00006304414,0.00020980659,0.00034359095,0.0010292652],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000028946147,0.00007260848,0.00010382595,0.0040015555,0.00037740378,0.000009149807,0.0004841888,2.1894607e-7,5.61689e-7,0.0014286771,0.002526489,0.9909924],"study_design_scores_gemma":[0.000026724083,0.000049547725,0.0009846346,0.012851608,0.0038166654,3.5265023e-7,0.000119404234,0.000006328775,2.4045684e-7,0.00008426211,0.9817876,0.00027260018],"about_ca_topic_score_codex":0.0004432765,"about_ca_topic_score_gemma":0.000102056525,"teacher_disagreement_score":0.99071985,"about_ca_system_score_codex":0.00008707596,"about_ca_system_score_gemma":0.00020172214,"threshold_uncertainty_score":0.9997485},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4402487954","doi":"10.1177/14604582241283969","title":"Artificial intelligence and health information: A bibliometric analysis of three decades of research","year":2024,"lang":"en","type":"article","venue":"Health Informatics Journal","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Interoperability; Health care; Citation; Web of science; China; Data science; Bibliometrics; Computer science; Library science; MEDLINE; World Wide Web; Political science","score_opus":0.5535763913030445,"score_gpt":0.5754334511636756,"score_spread":0.021857059860631134,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402487954","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.70675796,0.020568568,0.23239852,0.036707286,0.0012528442,0.0013111443,0.00006360636,0.00007263119,0.00086745096],"genre_scores_gemma":[0.9806438,0.0100055905,0.008467643,0.00067824736,0.00015909379,0.0000071759314,0.000024570834,0.0000070941637,0.0000067599976],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9950454,0.00009372778,0.003313161,0.00007571964,0.0010197747,0.0004521951],"domain_scores_gemma":[0.9968613,0.00051318697,0.00072900474,0.00022135032,0.001191576,0.00048363014],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.009949236,0.000108570464,0.0005491576,0.07100009,0.00028803444,0.00013441485,0.00012723937,0.000099622215,0.000106328065],"category_scores_gemma":[0.0007049153,0.00008846585,0.0001260482,0.086091265,0.00020753081,0.0007529492,0.000043623706,0.00077947264,0.000023183758],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000052570314,0.00006164997,0.013257841,0.0038481387,0.00022129925,0.0000010976288,0.021300837,0.00027324236,0.0000012769352,0.006876241,0.0013405165,0.9527653],"study_design_scores_gemma":[0.00013777045,0.01012494,0.19432527,0.008564988,0.0007756396,0.0013136248,0.106604785,0.59297967,0.0028676004,0.056165393,0.02554915,0.00059116277],"about_ca_topic_score_codex":0.00076358515,"about_ca_topic_score_gemma":0.00013681436,"teacher_disagreement_score":0.9521741,"about_ca_system_score_codex":0.00029295505,"about_ca_system_score_gemma":0.0031601551,"threshold_uncertainty_score":0.93952936},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4402505611","doi":"10.1007/s00266-024-04049-3","title":"The Most Influential Publications Regarding Hair Transplantation: A Bibliometric Review","year":2024,"lang":"en","type":"review","venue":"Aesthetic Plastic Surgery","topic":"Hair Growth and Disorders","field":"Medicine","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University Health Centre","funders":"","keywords":"Bibliometrics; Citation; Transplantation; Medicine; Citation analysis; Web of science; Impact factor; Dominance (genetics); Library science; Political science; Pathology; Computer science; Meta-analysis; Surgery; Biology; Law","score_opus":0.057101292200356604,"score_gpt":0.3408860004924637,"score_spread":0.2837847082921071,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402505611","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000008969277,0.9932046,0.00012245883,0.0012654371,0.0017990619,0.0010094154,0.00006372822,0.00025850505,0.0022678345],"genre_scores_gemma":[0.000037051374,0.9972507,0.000015434318,0.0003470365,0.00017964726,0.0006573049,0.00061262556,0.00010440676,0.00079581654],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99656034,0.00020730648,0.0013614482,0.0005813696,0.00071367004,0.00057587813],"domain_scores_gemma":[0.9933268,0.0050201495,0.00034653628,0.00075424084,0.00019967793,0.00035264142],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0011583286,0.00053638563,0.0018904754,0.014454322,0.00027161103,0.00021593177,0.00027436833,0.00026720346,0.000118778844],"category_scores_gemma":[0.0035359124,0.00031980584,0.0012303544,0.05477956,0.00017878633,0.00011265851,0.00004468316,0.0006720835,0.0012434373],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000008025636,0.000038837843,0.000016426513,0.065449595,0.00014301274,0.00011504339,0.0000071719487,1.0416369e-7,8.915195e-9,0.00025464623,0.124377556,0.80958956],"study_design_scores_gemma":[0.000117520794,0.00002246332,0.000023713448,0.0937529,0.0059971614,0.0026906724,0.0000057881907,0.0000054392794,3.2420186e-8,0.000021049125,0.8970848,0.00027849904],"about_ca_topic_score_codex":0.000010468456,"about_ca_topic_score_gemma":0.0000056130516,"teacher_disagreement_score":0.8093111,"about_ca_system_score_codex":0.00009801362,"about_ca_system_score_gemma":0.001945715,"threshold_uncertainty_score":0.9999254},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4402761164","doi":"10.3390/jrfm17100428","title":"Mapping the Knowledge Landscape of Money Laundering for Terrorism Financing: A Bibliometric Analysis","year":2024,"lang":"en","type":"article","venue":"Journal of risk and financial management","topic":"Crime, Illicit Activities, and Governance","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Southwest Jiaotong University","keywords":"Money laundering; Terrorism; Business; Finance; Geography; Archaeology","score_opus":0.029978208336606797,"score_gpt":0.29024722304659933,"score_spread":0.26026901470999253,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402761164","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8014124,0.032768805,0.15907133,0.00034426776,0.0011698605,0.00029312135,0.000027347312,0.000017955825,0.0048949094],"genre_scores_gemma":[0.9848076,0.013586782,0.00042491403,0.000018378601,0.00045759257,0.0000065487056,3.9747545e-7,0.0000065351487,0.0006912429],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989738,0.000053947257,0.00035063125,0.0001289247,0.00029099835,0.00020173195],"domain_scores_gemma":[0.9991122,0.00036165485,0.00028439454,0.000093603805,0.000099925346,0.00004823615],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0015773152,0.000091740185,0.00027855078,0.009072084,0.000306521,0.00012308445,0.00020141696,0.000049465994,0.000008499706],"category_scores_gemma":[0.00024930428,0.00006584358,0.00026001318,0.023378188,0.000079376994,0.00020413513,0.00005507611,0.00014732011,7.44043e-7],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007585254,0.00011981557,0.012555858,0.00041441122,0.00049659703,0.00002791591,0.07929206,0.00022608278,0.000017721084,0.021275152,0.0071581327,0.8783404],"study_design_scores_gemma":[0.0004097176,0.00011004695,0.1978723,0.0002091474,0.00080686715,0.000002028597,0.019647026,0.00071852264,0.000020656998,0.0048421533,0.7752004,0.0001610907],"about_ca_topic_score_codex":0.00041019917,"about_ca_topic_score_gemma":0.0005077891,"teacher_disagreement_score":0.8781793,"about_ca_system_score_codex":0.00005238741,"about_ca_system_score_gemma":0.00007791007,"threshold_uncertainty_score":0.9973804},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4402771677","doi":"10.1504/ijkms.2024.141563","title":"Knowledge management through the lens of business process management: a bibliometric analysis","year":2024,"lang":"en","type":"article","venue":"International Journal of Knowledge Management Studies","topic":"Collaboration in agile enterprises","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Trois-Rivières","funders":"","keywords":"Knowledge management; Process (computing); Business; Business process management; Process management; Computer science; Business process; Marketing; Work in process","score_opus":0.04395351841320131,"score_gpt":0.3510888236325045,"score_spread":0.30713530521930316,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402771677","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012884394,0.055320535,0.030418143,0.0054466557,0.014330658,0.001401939,0.000028393708,0.00027340197,0.87989587],"genre_scores_gemma":[0.9844724,0.008570093,0.00045179273,0.0002862285,0.0015484903,0.00009691995,0.000015894853,0.000048850274,0.0045093284],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99640936,0.000048325594,0.0014044183,0.00047463924,0.0013097054,0.0003535813],"domain_scores_gemma":[0.994672,0.00024133937,0.0008678061,0.00040254992,0.0037984324,0.000017844219],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0011256739,0.000423065,0.00065442733,0.044145092,0.00023286363,0.0007243325,0.0017784806,0.00005060179,0.00014827937],"category_scores_gemma":[0.00012158562,0.000301757,0.0004973727,0.09237607,0.00023909145,0.0021652153,0.0014210151,0.00021919906,0.00022653294],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00039017372,0.0017815845,0.007365254,0.009995223,0.09007632,0.0008395922,0.0039222385,0.0060049063,0.00000980506,0.63229764,0.100598484,0.14671879],"study_design_scores_gemma":[0.0022493645,0.000048794143,0.038004175,0.0030052178,0.012371619,0.000018471059,0.02174588,0.002543643,0.000040036142,0.020825135,0.8983579,0.00078975584],"about_ca_topic_score_codex":0.000008606536,"about_ca_topic_score_gemma":0.000046411173,"teacher_disagreement_score":0.971588,"about_ca_system_score_codex":0.00023281116,"about_ca_system_score_gemma":0.000029469818,"threshold_uncertainty_score":0.99994344},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4402878363","doi":"10.1016/j.heliyon.2024.e38620","title":"Retracted articles in scientific literature: A bibliometric analysis from 2003 to 2022 using the Web of Science","year":2024,"lang":"en","type":"article","venue":"Heliyon","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":25,"is_retracted":true,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Ontario; University of Toronto","funders":"Tzu Chi College of Technology; Tzu Chi University","keywords":"Web of science; Bibliometrics; Data science; Library science; Computer science; Engineering ethics; Management science; MEDLINE; Engineering; Political science","score_opus":0.3699171100309078,"score_gpt":0.5316296822327008,"score_spread":0.16171257220179297,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402878363","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9668082,0.028697183,0.0017613408,0.00047510897,0.0013720492,0.00026665322,0.00019467667,0.000022151842,0.00040261494],"genre_scores_gemma":[0.997238,0.0006360445,0.0012423153,0.000062337735,0.00005715413,0.0000071315803,0.0000045165775,0.000008750993,0.00074378116],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9832885,0.00031478333,0.00092660356,0.0013112507,0.013400916,0.0007579365],"domain_scores_gemma":[0.98870337,0.0033618885,0.0001594203,0.0014239245,0.0059348894,0.00041653693],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.04405121,0.00015472707,0.00036757777,0.85833466,0.00042183144,0.012716362,0.0032331375,0.00011676925,0.00080048916],"category_scores_gemma":[0.057029687,0.000090533315,0.00018265865,0.99393094,0.00086397486,0.0012846689,0.00090225815,0.00050385814,0.00018276116],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004592428,0.00027154371,0.32701877,0.000047010475,0.00009718575,0.00009665251,0.0022311993,0.0012517171,0.50498426,0.00072152156,0.0046653696,0.15856881],"study_design_scores_gemma":[0.0001762917,0.0000805272,0.73137546,0.00017922773,0.000054916367,0.000003874308,0.0006420706,0.22278307,0.029041631,0.0011954224,0.014229294,0.00023821052],"about_ca_topic_score_codex":0.00020153249,"about_ca_topic_score_gemma":0.00018326932,"teacher_disagreement_score":0.47594264,"about_ca_system_score_codex":0.00021917043,"about_ca_system_score_gemma":0.0011817069,"threshold_uncertainty_score":0.98830855},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch","research_integrity"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4402985645","doi":"10.3390/psychiatryint5040045","title":"Interdisciplinary Insights and Global Perspectives on ADHD in Children: A Comprehensive Bibliometric Analysis (2014–2024)","year":2024,"lang":"en","type":"article","venue":"Psychiatry International","topic":"Attention Deficit Hyperactivity Disorder","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Psychology; Engineering ethics; Sociology; Data science; Computer science; Engineering; Library science","score_opus":0.020207055091215172,"score_gpt":0.3644516407314701,"score_spread":0.3442445856402549,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402985645","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9760136,0.0073729786,0.00043258694,0.0071856403,0.0014211988,0.00021856093,0.00007142529,0.000079394624,0.0072046197],"genre_scores_gemma":[0.9980799,0.00030959328,0.00039901643,0.00025763756,0.0003797744,0.000018965844,0.00009408605,0.000015531832,0.00044550587],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9985133,0.000040816776,0.00027216625,0.00057142944,0.00044507362,0.00015725994],"domain_scores_gemma":[0.99946815,0.00006853954,0.000054005002,0.00019474298,0.00012604876,0.00008849437],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00006538414,0.00018936359,0.0002735858,0.02810123,0.00004674896,0.000112186885,0.00011398404,0.000085348336,0.00087867153],"category_scores_gemma":[0.000033224598,0.00016802014,0.00021586657,0.028907731,0.00007378448,0.00021320803,0.00014110428,0.00031612744,0.00017596911],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021107709,0.00084018894,0.97936976,0.000038521954,0.0026878065,0.000038240756,0.0009869727,0.00013406621,0.00002789856,0.007072217,0.0043548676,0.004238392],"study_design_scores_gemma":[0.0008622479,0.000167855,0.9865734,0.00011950384,0.00032279963,0.00017716072,0.0058375946,0.0039394773,7.050503e-7,0.0012428839,0.0005978756,0.00015851855],"about_ca_topic_score_codex":0.00004507902,"about_ca_topic_score_gemma":0.00005816992,"teacher_disagreement_score":0.022066293,"about_ca_system_score_codex":0.00019523772,"about_ca_system_score_gemma":0.00007489549,"threshold_uncertainty_score":0.9917333},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4403013382","doi":"10.1016/j.jtumed.2024.09.009","title":"Orthopedic research productivity of KSA: First bibliometric analysis","year":2024,"lang":"en","type":"article","venue":"Journal of Taibah University Medical Sciences","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Deanship of Scientific Research, King Saud University; King Saud University","keywords":"Productivity; Orthopedic surgery; Medicine; Surgery; Economics","score_opus":0.7085184768933618,"score_gpt":0.6118808293536963,"score_spread":0.09663764753966553,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403013382","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9716052,0.0039594527,0.005708307,0.01144601,0.00094523875,0.000106146814,0.000014829869,0.000011783833,0.0062030344],"genre_scores_gemma":[0.99516624,0.002190862,0.0005612993,0.000011278043,0.00020847844,1.0489372e-7,2.8020597e-7,0.0000033326758,0.0018581252],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.95871186,0.0010055718,0.0009555261,0.0006781842,0.03797089,0.000677974],"domain_scores_gemma":[0.9662472,0.022659117,0.0006700801,0.0006166485,0.008272211,0.0015347254],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.16237631,0.00012455024,0.00062049786,0.7532117,0.00064792787,0.0011891111,0.0058548725,0.00017571758,0.0026450618],"category_scores_gemma":[0.113856986,0.00007621868,0.0005908801,0.94676805,0.0026594575,0.0018842507,0.0012166324,0.0010196427,0.00006482538],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008032344,0.00065788766,0.62832856,0.00007321756,0.0004908441,0.0009412624,0.00042971232,0.000466382,0.0001758209,0.005492912,0.04946505,0.31339803],"study_design_scores_gemma":[0.0007244764,0.0024412388,0.59390295,0.00022127376,0.00023460534,0.00015379234,0.004221419,0.022247262,0.0004846882,0.0066202627,0.36839578,0.00035227076],"about_ca_topic_score_codex":0.0002860041,"about_ca_topic_score_gemma":0.0001380434,"teacher_disagreement_score":0.31893072,"about_ca_system_score_codex":0.00024560755,"about_ca_system_score_gemma":0.0026878545,"threshold_uncertainty_score":0.99984777},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4403143722","doi":"10.1007/s11192-024-05145-6","title":"Interdisciplinarity and expert rating: an analysis based on faculty opinions","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"Interdisciplinary Research and Collaboration","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Social Science Fund of China","keywords":"Sociology; Computer science; Data science; Psychology; Management science; Engineering","score_opus":0.23329725374897026,"score_gpt":0.5451679613832665,"score_spread":0.31187070763429625,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403143722","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7771541,0.00041942636,0.1842975,0.01574253,0.002445636,0.0005896598,0.0009403064,0.00030877872,0.018102078],"genre_scores_gemma":[0.99620783,0.000001859446,0.0013454149,0.00012659661,0.00012371266,0.000019817913,0.0001850868,0.000007927264,0.0019817287],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99466985,0.00025400627,0.0004262364,0.0009364045,0.0033458297,0.00036766348],"domain_scores_gemma":[0.9970459,0.001089293,0.00006586374,0.00067283586,0.0007175007,0.00040861577],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0063840263,0.00015298738,0.00024302842,0.010413038,0.0005646345,0.0033168618,0.0007378678,0.00007602713,0.0003259655],"category_scores_gemma":[0.0030775568,0.00010331301,0.00016191666,0.05474503,0.0001891067,0.00086270866,0.00038787958,0.00022212332,0.00016676322],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00067503477,0.0023931505,0.1585543,0.000083973006,0.00058640965,0.00030338627,0.035203133,0.03080096,0.0020678479,0.056046557,0.25261265,0.46067262],"study_design_scores_gemma":[0.00015513918,0.00054143777,0.040945623,0.000014620872,0.000022496219,0.0000015226243,0.0030293595,0.9425716,0.00021994363,0.002455977,0.009872912,0.00016940701],"about_ca_topic_score_codex":0.000014351169,"about_ca_topic_score_gemma":0.00017134691,"teacher_disagreement_score":0.9117706,"about_ca_system_score_codex":0.00014761907,"about_ca_system_score_gemma":0.00020691042,"threshold_uncertainty_score":0.9977178},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4403187083","doi":"10.7759/cureus.71001","title":"A Century of Mathematical Epidemiology: A Bibliometric Analysis and Visualization of Research Trends","year":2024,"lang":"en","type":"review","venue":"Cureus","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Stephen's University","funders":"","keywords":"Medicine; Visualization; Epidemiology; Bibliometrics; Data science; Library science; Data mining; Pathology","score_opus":0.8547371498408696,"score_gpt":0.7288303240598216,"score_spread":0.12590682578104795,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403187083","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000098659555,0.9939383,0.0024372488,0.000034568744,0.00020538631,0.00029920673,0.0002241812,0.000015805252,0.0027466097],"genre_scores_gemma":[0.00075784774,0.9969106,0.00085981045,0.000002980819,0.00006477018,0.000032175452,0.000043295226,0.000026595111,0.0013019475],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9818671,0.0029317536,0.0034978993,0.0013851809,0.0095189875,0.00079910393],"domain_scores_gemma":[0.94858867,0.043802127,0.0011957261,0.0015548273,0.0043730177,0.0004856251],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.10176125,0.0003242755,0.004058829,0.9227555,0.00009440997,0.00049676705,0.0026788951,0.0006409004,0.0010782575],"category_scores_gemma":[0.17164032,0.00018451524,0.001335097,0.9791619,0.00080194615,0.00022055287,0.0019314224,0.00066081394,0.00014120695],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000027559513,0.00015861136,0.0013323318,0.0038076402,0.000672342,0.000006177494,0.000045696594,3.5319786e-7,3.9156163e-7,0.008527591,0.0050884522,0.98035765],"study_design_scores_gemma":[0.000117214804,0.00035129217,0.0022879564,0.0017911282,0.0027639263,0.00001817854,0.00010657563,0.0014464299,0.0000020546406,0.01778092,0.9730282,0.00030611965],"about_ca_topic_score_codex":0.00008531905,"about_ca_topic_score_gemma":0.000007903714,"teacher_disagreement_score":0.9800515,"about_ca_system_score_codex":0.00015347429,"about_ca_system_score_gemma":0.00040267254,"threshold_uncertainty_score":0.9998349},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4403192993","doi":"10.3390/w16192840","title":"Research Hotspots and Trend Analysis in Modeling Groundwater Dense Nonaqueous Phase Liquid Contamination Based on Bibliometrics","year":2024,"lang":"en","type":"article","venue":"Water","topic":"Groundwater flow and contamination studies","field":"Environmental Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Environmental remediation; Environmental science; Groundwater; Contamination; Groundwater contamination; Multiphase flow; Aquifer; Engineering; Ecology","score_opus":0.04735175123810771,"score_gpt":0.33419297733071024,"score_spread":0.28684122609260254,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403192993","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98471534,0.0000927958,0.013583735,0.00059360475,0.00007303735,0.00014290094,0.000005596478,0.000038614056,0.0007543485],"genre_scores_gemma":[0.99739486,0.000025716836,0.0000662101,0.00011879494,0.000023787032,0.000051718445,0.000036599267,0.0000145569265,0.0022677337],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9981852,0.00014922816,0.00023482223,0.0004638183,0.00058547343,0.00038142342],"domain_scores_gemma":[0.99959636,0.00012861322,0.000009938666,0.00017368737,0.000023087701,0.00006829713],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012309449,0.00014166135,0.00017761068,0.0060998597,0.00014751666,0.00026449643,0.00010730954,0.00007079335,0.0005033503],"category_scores_gemma":[0.000016483193,0.000095727955,0.000059853508,0.006499818,0.00009779871,0.00032432185,0.00012500583,0.00018912416,0.00023772342],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027353186,0.0043021394,0.11561859,0.00039267624,0.001112158,0.0030205192,0.04277392,0.17173685,0.13641647,0.00040199142,0.0032583592,0.51823103],"study_design_scores_gemma":[0.0012930222,0.0007397368,0.01676184,0.000038528233,0.00013276536,0.0000052916057,0.00039755402,0.96570027,0.008996739,0.000113323906,0.005473587,0.00034736234],"about_ca_topic_score_codex":0.0006259549,"about_ca_topic_score_gemma":0.0009895614,"teacher_disagreement_score":0.7939634,"about_ca_system_score_codex":0.0002210817,"about_ca_system_score_gemma":0.000004425778,"threshold_uncertainty_score":0.5511333},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4403247304","doi":"10.9734/ijpss/2024/v36i105070","title":"A Bibliometric Review of the Impacts of Logging in Forests in Semi-Arid Zones","year":2024,"lang":"en","type":"review","venue":"International Journal of Plant & Soil Science","topic":"Forest Management and Policy","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Logging; Arid; Livelihood; Environmental resource management; Geography; Climate change; Ecosystem services; Ecosystem; Environmental planning; Agroforestry; Ecology; Environmental science; Agriculture; Forestry","score_opus":0.03250107826327611,"score_gpt":0.3531186037973316,"score_spread":0.32061752553405554,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403247304","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0054348363,0.9914373,0.0000021871526,0.0002208655,0.00084601535,0.00022077914,0.00005164882,0.0000020773132,0.0017842922],"genre_scores_gemma":[0.014147128,0.98559254,0.000028179678,0.00006522274,0.00006229902,0.0000029647147,0.0000026445302,0.000007198498,0.00009179483],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99677306,0.00007786636,0.0011266792,0.00018953544,0.0016045755,0.00022830509],"domain_scores_gemma":[0.9983644,0.00016214512,0.0011866632,0.0001748913,0.000048835176,0.000063101004],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0027223185,0.0001704664,0.00062681036,0.012159729,0.000018099752,0.000048536396,0.0023186363,0.00004795492,0.00016899456],"category_scores_gemma":[0.000639359,0.000096456344,0.00028101288,0.02652455,0.00036746686,0.00041931926,0.000677784,0.0003611393,0.000042042262],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000029219798,0.0003909612,0.0722181,0.046252575,0.00017569843,0.00050641305,0.0004982742,0.001158283,0.00004917433,0.001061269,0.023478698,0.85418135],"study_design_scores_gemma":[0.0003372103,0.00011398336,0.037051383,0.38830596,0.00028808348,0.00077744684,0.000012184599,0.0006299815,0.000020155487,0.00069895125,0.57138914,0.00037550752],"about_ca_topic_score_codex":0.000816988,"about_ca_topic_score_gemma":0.0005884867,"teacher_disagreement_score":0.85380584,"about_ca_system_score_codex":0.00042029103,"about_ca_system_score_gemma":0.00030801326,"threshold_uncertainty_score":0.9990366},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4403254476","doi":"10.18009/jcer.1461156","title":"Bibliometric Analysis of Studies between 2010-2023 on Leadership in the Field of Educational Sciences","year":2024,"lang":"en","type":"article","venue":"Journal of Computer and Education Research","topic":"Organizational Leadership and Management Strategies","field":"Business, Management and Accounting","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Transformational leadership; Library science; Political science; Sociology; Public relations; MEDLINE; Computer science","score_opus":0.31131239943490485,"score_gpt":0.45362922787229487,"score_spread":0.14231682843739002,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403254476","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.87082684,0.006394906,0.00027461228,0.117056444,0.00087782205,0.00014940032,0.0000010771266,0.000003451923,0.0044154353],"genre_scores_gemma":[0.9980368,0.00015493103,0.00010032439,0.0005663492,0.00091231044,0.0000016899278,0.0000027489864,0.0000024761923,0.00022237],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9988678,0.000049748505,0.00029228753,0.00009219363,0.00058941025,0.00010856393],"domain_scores_gemma":[0.99717265,0.0021367765,0.000118287244,0.000062964515,0.00050226063,0.000007042531],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0025018002,0.00004976365,0.00015152062,0.04960257,0.00007561108,0.00022734982,0.00025751986,0.000020766072,0.00006777022],"category_scores_gemma":[0.00034864413,0.000032107553,0.00005182739,0.06698141,0.00013825938,0.00047523074,0.00004319612,0.00015459716,0.000004993594],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000028044127,0.0004569519,0.42255983,0.002134403,0.00094785576,0.0000017520154,0.0023793594,0.0007398775,0.000021155269,0.23801711,0.29233688,0.04037677],"study_design_scores_gemma":[0.000078546924,0.00011558583,0.96923465,0.00035275175,0.00017823835,6.9650224e-7,0.011819795,0.0005505289,0.000019441239,0.014738191,0.002854221,0.000057345802],"about_ca_topic_score_codex":0.00007274693,"about_ca_topic_score_gemma":0.00001019785,"teacher_disagreement_score":0.54667485,"about_ca_system_score_codex":0.000013438788,"about_ca_system_score_gemma":0.0001663133,"threshold_uncertainty_score":0.96116936},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4403298204","doi":"10.1159/000541099","title":"Trends and Most-Cited Articles on Hemispheric Surgery: A Bibliometric Analysis and Historical Review","year":2024,"lang":"en","type":"review","venue":"Stereotactic and Functional Neurosurgery","topic":"Epilepsy research and treatment","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Hemispherectomy; Disconnection; Epilepsy surgery; Psychology; Medicine; Epilepsy; Neuroscience; Philosophy","score_opus":0.10410266536024533,"score_gpt":0.3385703271225989,"score_spread":0.23446766176235356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403298204","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011487142,0.99700475,0.000003896073,0.0008659737,0.00028848488,0.0002996557,0.000057345216,0.00006063779,0.0002705742],"genre_scores_gemma":[0.0010797618,0.99259704,0.000010218906,0.00052059523,0.00020083998,0.00013205798,0.00022263483,0.00005305243,0.0051838206],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9971624,0.00021555104,0.0007368298,0.00087843643,0.0006275394,0.0003792695],"domain_scores_gemma":[0.9964338,0.0023626056,0.0002178888,0.00033578483,0.000084920764,0.0005649951],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0005008606,0.00050903735,0.0025058254,0.034018826,0.00012306408,0.00012132922,0.000028815628,0.00015965728,0.00024657798],"category_scores_gemma":[0.00067118765,0.00033658373,0.0007136302,0.0785043,0.00009184766,0.00008641589,0.00008276367,0.00054384535,0.000029073233],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009554315,0.00018372982,0.010967176,0.022434957,0.0021859556,0.0007378995,0.000001309895,1.0134332e-8,3.0423644e-7,0.000010518578,0.03169083,0.93169177],"study_design_scores_gemma":[0.00020682965,0.0003080809,0.039785884,0.010225928,0.019686524,0.0014534323,0.0000012208795,0.000014361846,1.4886079e-7,0.000009594199,0.9279947,0.0003132503],"about_ca_topic_score_codex":0.000033013403,"about_ca_topic_score_gemma":0.0000014430125,"teacher_disagreement_score":0.93137854,"about_ca_system_score_codex":0.00018506346,"about_ca_system_score_gemma":0.00020008553,"threshold_uncertainty_score":0.9999086},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4403426829","doi":"10.1016/j.heliyon.2024.e39370","title":"Global research trends in the teacher evaluation of medical college: A bibliometric and visualized study","year":2024,"lang":"en","type":"review","venue":"Heliyon","topic":"Innovations in Medical Education","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Department of Education of Hebei Province","keywords":"Bibliometrics; Medical education; Library science; Data science; Medicine; Engineering ethics; Computer science; Engineering","score_opus":0.37941088702897685,"score_gpt":0.6530657943204756,"score_spread":0.2736549072914988,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403426829","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00827782,0.9848238,0.0000015621597,0.0011718961,0.00036835845,0.002156795,0.000014448774,0.000017031858,0.0031682446],"genre_scores_gemma":[0.0034752837,0.9947485,0.00003492846,0.00006349703,0.00039591323,0.0007843427,0.0001227331,0.000025727064,0.00034907268],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.98956555,0.002600221,0.001027968,0.0004563386,0.0060755513,0.00027435034],"domain_scores_gemma":[0.99824995,0.0003167872,0.00014983969,0.0005268071,0.00068464543,0.000071965675],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.035847295,0.00021381388,0.0009806089,0.0407655,0.000056135574,0.000040834664,0.00034668576,0.0004964102,0.00082108186],"category_scores_gemma":[0.008876207,0.00012210419,0.00011360612,0.19832395,0.00019911156,0.00003680245,0.0001438547,0.0012972516,0.000037201884],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011444358,0.0010505175,0.00020890277,0.01192514,0.0001039488,0.000024162686,0.00046821422,9.629361e-9,1.2831855e-8,0.0005526092,0.01490192,0.97075313],"study_design_scores_gemma":[0.0024124752,0.0007960301,0.008749875,0.036243346,0.003981451,0.00026355212,0.0040936386,0.00024267226,3.2124706e-8,0.00017279145,0.9428513,0.00019282021],"about_ca_topic_score_codex":0.00007701481,"about_ca_topic_score_gemma":0.00006348907,"teacher_disagreement_score":0.9705603,"about_ca_system_score_codex":0.00064581935,"about_ca_system_score_gemma":0.0019062986,"threshold_uncertainty_score":0.99947244},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4403440584","doi":"10.3389/fnut.2024.1485642","title":"Mapping the evolution and impact of ketogenic diet research on diabetes management: a comprehensive bibliometric analysis from 2005 to 2024","year":2024,"lang":"en","type":"article","venue":"Frontiers in Nutrition","topic":"Diet and metabolism studies","field":"Medicine","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Diabetes mellitus; Insulin resistance; Gerontology; Diabetes management; Medicine; China; Environmental health; Type 2 diabetes; Geography; Endocrinology","score_opus":0.05049472749590787,"score_gpt":0.3491593764802707,"score_spread":0.2986646489843628,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403440584","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.89749944,0.09937175,0.00076795084,0.00047448347,0.00040500175,0.00063679775,0.00008040166,0.000019813531,0.0007443688],"genre_scores_gemma":[0.9820728,0.01603483,0.0012631271,0.000029753326,0.00015604727,0.000111102454,0.000046398407,0.000010961338,0.00027503504],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99870574,0.00010878259,0.00021377491,0.00029140516,0.0003933324,0.0002869855],"domain_scores_gemma":[0.99943846,0.00016500827,0.000025350988,0.00019949778,0.00010215602,0.000069504684],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0004530186,0.00011195804,0.000358756,0.056233287,0.00006855229,0.000045006356,0.00007258869,0.00005263944,0.000023077278],"category_scores_gemma":[0.000055565746,0.000078853445,0.00017268336,0.08085084,0.000091705035,0.00007506754,0.000077011195,0.00020111124,0.000008875436],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00077483675,0.00031852038,0.64410853,0.0015095338,0.0055958624,0.000027977465,0.0011078604,0.00009821054,0.0016493389,0.00013676024,0.31637812,0.028294448],"study_design_scores_gemma":[0.0006886569,0.0003648848,0.98416257,0.0009147653,0.0005572083,1.479194e-7,0.0036255408,0.0021080805,0.000060632785,0.0016371609,0.0057987683,0.00008156916],"about_ca_topic_score_codex":0.00020517228,"about_ca_topic_score_gemma":0.000012696318,"teacher_disagreement_score":0.34005404,"about_ca_system_score_codex":0.00024680822,"about_ca_system_score_gemma":0.000015435298,"threshold_uncertainty_score":0.9544635},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4403541284","doi":"10.1177/09610006241285514","title":"Everyday life information seeking: A systematic review with bibliometric analysis","year":2024,"lang":"en","type":"review","venue":"Journal of Librarianship and Information Science","topic":"Information Retrieval and Search Behavior","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Social Science Foundation of Jiangsu Province","keywords":"Information seeking; Everyday life; Sociology; Systematic review; Psychology; Bibliometrics; Citation analysis; Epistemology; Social science; Political science; Information retrieval; Library science; Computer science; MEDLINE; Citation; Philosophy","score_opus":0.048565434520747644,"score_gpt":0.32407140836892184,"score_spread":0.2755059738481742,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403541284","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000018030591,0.9076445,0.089568555,0.00028449405,0.00032727615,0.00090675737,0.000020057723,0.000057633046,0.0011888837],"genre_scores_gemma":[0.000086346336,0.99652225,0.0021180492,0.0011567973,0.000032411896,0.00002785945,0.00001900058,0.0000055071646,0.000031803367],"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","domain_scores_codex":[0.99306005,0.00017619082,0.003406805,0.00016731478,0.0028063,0.0003833472],"domain_scores_gemma":[0.9932953,0.0002778702,0.0038005875,0.00061181746,0.0014769073,0.0005374826],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics","scholarly_communication"],"category_scores_codex":[0.006468339,0.0003908354,0.0021135896,0.11477573,0.00026931392,0.0060372953,0.0024259975,0.00015392137,0.000016391115],"category_scores_gemma":[0.0017055337,0.00021228797,0.000544974,0.26307508,0.00018465763,0.0966526,0.00038376884,0.0006988521,0.00018290075],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000045357965,0.000014417756,0.000007820868,0.8167999,0.00047879075,0.000007573121,0.0010927018,0.000007241611,1.7467881e-8,0.0135279875,0.001048091,0.16701092],"study_design_scores_gemma":[0.00032909212,0.00044183424,0.0001392464,0.35525236,0.012456032,0.0012644905,0.00021163869,0.0053026523,8.0771395e-7,0.00005748813,0.6236964,0.0008480089],"about_ca_topic_score_codex":0.0000018277518,"about_ca_topic_score_gemma":6.677641e-8,"teacher_disagreement_score":0.6226483,"about_ca_system_score_codex":0.0001490254,"about_ca_system_score_gemma":0.0035292988,"threshold_uncertainty_score":0.9949945},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4403544282","doi":"10.1177/23294884241290216","title":"An In-Depth Bibliometric Exploration of Research on “Bullshit” Communication","year":2024,"lang":"en","type":"article","venue":"International Journal of Business Communication","topic":"Misinformation and Its Impacts","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Saint Mary's University","funders":"","keywords":"Scientific communication; Psychology; Sociology; Knowledge management; Computer science; Library science","score_opus":0.2839208738756311,"score_gpt":0.5191620428264646,"score_spread":0.23524116895083352,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403544282","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.88357943,0.004114803,0.007637118,0.036612578,0.0014568478,0.00046407868,0.0000141237915,0.00008417214,0.06603686],"genre_scores_gemma":[0.98707235,0.011484892,0.001102833,0.000077963705,0.00013642055,0.0000027640294,0.000035033638,0.000009829965,0.00007789439],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9966536,0.0008899601,0.00070570037,0.00007761891,0.001536545,0.00013656534],"domain_scores_gemma":[0.9947062,0.0008631314,0.00036825903,0.0004122303,0.0035760882,0.00007409931],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0064503863,0.00007170906,0.00013870158,0.018803451,0.00015515363,0.00041294753,0.0014796437,0.00009309762,0.0001239118],"category_scores_gemma":[0.0015093908,0.00006720641,0.00004927557,0.01712251,0.00023300134,0.004287524,0.000075480086,0.00042711385,0.000031122978],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004615284,0.0011760354,0.0019043243,0.00008351576,0.00011213931,0.000013403802,0.09664316,0.009285989,0.0015728232,0.3694685,0.005508894,0.5137697],"study_design_scores_gemma":[0.0042452407,0.00094989233,0.48607123,0.01001087,0.000064537795,0.00007663488,0.1257967,0.025841119,0.0044162036,0.09436264,0.24724399,0.0009209256],"about_ca_topic_score_codex":0.0007470313,"about_ca_topic_score_gemma":0.0005246544,"teacher_disagreement_score":0.51284873,"about_ca_system_score_codex":0.0003490768,"about_ca_system_score_gemma":0.0004784378,"threshold_uncertainty_score":0.99231756},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4403560077","doi":"10.1016/j.ecolind.2024.112740","title":"Climate warming effects on temperature structure in lentic waters: A bibliometric analysis from the recent 20 years","year":2024,"lang":"en","type":"article","venue":"Ecological Indicators","topic":"Marine and coastal ecosystems","field":"Earth and Planetary Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Lethbridge","funders":"National Natural Science Foundation of China","keywords":"Lake ecosystem; Climate change; Environmental science; Ecology; Global warming; Ecosystem; Biology","score_opus":0.008688982272584453,"score_gpt":0.22169879262269898,"score_spread":0.21300981035011451,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403560077","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9967223,0.00090553856,7.188672e-7,0.00052800955,0.0005829481,0.00021416305,0.00019707368,0.000059692982,0.0007895901],"genre_scores_gemma":[0.99848765,0.00043904025,0.000014894888,0.00052601506,0.00016105424,0.000004078098,0.00031895394,0.0000035588505,0.000044727985],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9984708,0.0002169687,0.00023951141,0.0004070059,0.00029145123,0.00037426004],"domain_scores_gemma":[0.99884504,0.0008094734,0.000047296522,0.00018338756,0.000007011858,0.00010778574],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00036980418,0.00015557178,0.00025836733,0.009549935,0.00010838097,0.0002682351,0.00030213175,0.00017335589,0.0043501155],"category_scores_gemma":[0.000098654,0.00008177247,0.00011791986,0.05688939,0.000036420173,0.00009451962,0.000043790213,0.00045487718,0.0003096093],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019718082,0.000020272097,0.8511829,0.000022796494,0.00016120968,0.00026073575,0.00013587199,0.0005495809,0.000019010835,0.000016791133,0.0010769707,0.14653416],"study_design_scores_gemma":[0.000097460295,0.00013393111,0.9869666,0.000027814327,0.00009568066,0.0000013369321,0.00006230848,0.0016784387,0.000075869844,0.00014501563,0.010591141,0.0001243892],"about_ca_topic_score_codex":0.00052994984,"about_ca_topic_score_gemma":0.0035655284,"teacher_disagreement_score":0.14640976,"about_ca_system_score_codex":0.000016085382,"about_ca_system_score_gemma":0.000022824612,"threshold_uncertainty_score":0.99656004},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4403745910","doi":"10.18280/isi.290538","title":"Text Summarization: A Bibliometric Study and Systematic Literature Review","year":2024,"lang":"en","type":"article","venue":"Ingénierie des systèmes d information","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Automatic summarization; Bibliometrics; Information retrieval; Systematic review; Computer science; Data science; Library science; MEDLINE; Political science","score_opus":0.017356563989845582,"score_gpt":0.2589526443469761,"score_spread":0.24159608035713054,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403745910","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0055318866,0.19490065,0.7936068,0.00022552347,0.000694142,0.001856401,0.000005135071,0.00064111064,0.0025383611],"genre_scores_gemma":[0.9837432,0.009721674,0.005653044,0.0004642809,0.00006168293,0.00020331994,0.000020347032,0.000009975691,0.00012250534],"study_design_codex":"systematic_review","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984551,0.00011358515,0.00068041775,0.00019309763,0.00038412685,0.0001736606],"domain_scores_gemma":[0.9990195,0.00011656889,0.00013920585,0.00042551375,0.00023338605,0.00006580815],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0010686817,0.00015451045,0.00026973154,0.007501879,0.00013856987,0.0021901992,0.00035525011,0.0000577748,0.0000053082467],"category_scores_gemma":[0.00040117884,0.00012240106,0.00004388526,0.029504742,0.00001940302,0.0063315607,0.00017069296,0.00012956944,0.000072885254],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000024653252,0.000046137175,0.00072235113,0.589054,0.00012773475,0.000053464137,0.028782751,0.000096462056,0.0000085415795,0.04360675,0.0015567723,0.33594257],"study_design_scores_gemma":[0.00045970466,0.0003038404,0.002683402,0.26698133,0.00016525378,0.0008076177,0.00091142,0.7194124,0.000022334038,0.0044794357,0.00302272,0.00075052446],"about_ca_topic_score_codex":0.0000060233265,"about_ca_topic_score_gemma":0.000001014282,"teacher_disagreement_score":0.9782113,"about_ca_system_score_codex":0.000096224656,"about_ca_system_score_gemma":0.000055096127,"threshold_uncertainty_score":0.99884564},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4403821628","doi":"10.18280/ijdne.190513","title":"Bibliometric Analysis of Trends and Future Research Directions in Protected Fat Supplementation in Ruminant Nutrition","year":2024,"lang":"en","type":"article","venue":"International Journal of Design & Nature and Ecodynamics","topic":"Food Industry and Aquatic Biology","field":"Agricultural and Biological Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Universitas Hasanuddin","keywords":"Ruminant; Biology; Biotechnology; Agronomy","score_opus":0.04165523606454467,"score_gpt":0.3408573317325757,"score_spread":0.299202095668031,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403821628","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99098,0.0032884234,0.000029934687,0.0053056916,0.00021446933,0.00007743858,0.00005952449,0.0000029362116,0.00004153781],"genre_scores_gemma":[0.99572873,0.0036697509,0.0002667811,0.000013863264,0.00022434177,0.0000049667606,0.00006771945,5.6707415e-7,0.000023267643],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99909896,0.0001460153,0.0003091463,0.00011711504,0.00022918188,0.00009961095],"domain_scores_gemma":[0.99936694,0.000319476,0.00008611338,0.0000146186785,0.0001840653,0.000028777125],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00073868025,0.000058586676,0.00014975187,0.013087952,0.000027455042,0.00005575185,0.000108687724,0.00018738935,0.00007648264],"category_scores_gemma":[0.000054384524,0.000026934187,0.000049047532,0.032586683,0.000039027003,0.00013866831,0.000021554446,0.00049221615,1.4912042e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034267086,0.00032343654,0.029929522,0.000021388305,0.00038360117,0.000077577824,0.00025728132,0.00007631786,0.043479417,0.0013840004,0.00032162364,0.92340314],"study_design_scores_gemma":[0.00050061307,0.0008827872,0.96860933,0.0001782663,0.000087387605,0.00008345591,0.0010259643,0.01858642,0.0009582198,0.0023987435,0.006571161,0.0001176578],"about_ca_topic_score_codex":0.00009124912,"about_ca_topic_score_gemma":0.0009907676,"teacher_disagreement_score":0.9386798,"about_ca_system_score_codex":0.00005145334,"about_ca_system_score_gemma":0.000010962332,"threshold_uncertainty_score":0.99809784},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4403964361","doi":"10.1186/s12245-024-00738-7","title":"Bibliometric analysis of the usage of tenecteplase for stroke","year":2024,"lang":"en","type":"article","venue":"International Journal of Emergency Medicine","topic":"Acute Ischemic Stroke Management","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Tenecteplase; Angiology; Stroke (engine); MEDLINE; Ischaemic stroke; Emergency medicine; Internal medicine; Myocardial infarction; Thrombolysis; Atrial fibrillation","score_opus":0.04703422917965209,"score_gpt":0.38530565268908495,"score_spread":0.33827142350943284,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403964361","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95145375,0.0067153866,0.016519232,0.008074309,0.009146133,0.0002842548,0.00015086199,0.000010102189,0.0076459954],"genre_scores_gemma":[0.9936351,0.0014477344,0.0005241424,0.00008448222,0.0007779572,0.0000036323356,0.000013347464,0.0000126234945,0.00350095],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99732924,0.000022752787,0.0011696848,0.00012154899,0.0012482103,0.00010854556],"domain_scores_gemma":[0.99754524,0.0002628737,0.0006104776,0.00019501628,0.0013150955,0.000071287795],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0008269917,0.00010872534,0.00046834137,0.036565524,0.000011761421,0.0000025754275,0.00043656162,0.00004029001,0.0018898098],"category_scores_gemma":[0.0015790145,0.000061173785,0.0006322173,0.030267313,0.00007513025,0.00007793202,0.00006878756,0.00015299594,9.721946e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001034721,0.0005567788,0.22194697,0.0006857134,0.049746167,0.000110201785,0.00093621376,0.0005262636,0.25128004,0.0013905338,0.43080333,0.040983092],"study_design_scores_gemma":[0.005052075,0.0022495368,0.66375196,0.0022684091,0.028560985,0.00014226313,0.000903189,0.012161729,0.04072734,0.0003037012,0.24363054,0.00024830067],"about_ca_topic_score_codex":0.000036324185,"about_ca_topic_score_gemma":0.000004474259,"teacher_disagreement_score":0.44180498,"about_ca_system_score_codex":0.00007531811,"about_ca_system_score_gemma":0.00009162274,"threshold_uncertainty_score":0.9990226},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4404376050","doi":"10.3390/jrfm17110509","title":"The Role of Digital Technologies in Corporate Sustainability: A Bibliometric Review and Future Research Agenda","year":2024,"lang":"en","type":"review","venue":"Journal of risk and financial management","topic":"Innovation Diffusion and Forecasting","field":"Decision Sciences","cited_by":62,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Sustainability; Corporate sustainability; Political science; Knowledge management; Business; Engineering ethics; Regional science; Environmental resource management; Data science; Sociology; Engineering; Computer science; Public relations; Corporate social responsibility; Environmental science","score_opus":0.14732792808930317,"score_gpt":0.42717434771369855,"score_spread":0.2798464196243954,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404376050","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00008044213,0.9979701,0.000048204434,0.0005231485,0.00021160641,0.00056052697,0.000022635584,0.000005969168,0.0005773387],"genre_scores_gemma":[0.0005833148,0.99897516,0.00006266436,0.000007048515,0.00008335225,0.000013243078,7.4347184e-7,0.00000883713,0.00026562548],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99621415,0.0002555322,0.0018461138,0.00028679598,0.0011547147,0.00024269636],"domain_scores_gemma":[0.9959386,0.0008771468,0.0017414698,0.00032007796,0.0010882072,0.000034487268],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.014168862,0.00018551783,0.0011064463,0.024182536,0.00018298344,0.00050456275,0.0006766045,0.00014249516,0.000004580702],"category_scores_gemma":[0.0076593813,0.000088982044,0.00021318141,0.0641078,0.00027792595,0.00019985909,0.0008388337,0.0009486449,0.0000045788547],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000007729391,0.000021456857,0.0001784702,0.0045283022,0.000013818928,0.000051364248,0.000043938326,3.8257262e-8,2.8684397e-9,0.009752921,0.0027064423,0.9826955],"study_design_scores_gemma":[0.00007013348,0.00007627872,0.00018157683,0.003764431,0.00015454298,0.000030026573,0.0021900064,0.0000024512426,2.4211758e-8,0.11983185,0.8736251,0.00007357548],"about_ca_topic_score_codex":0.0000019735508,"about_ca_topic_score_gemma":0.000003593283,"teacher_disagreement_score":0.98262197,"about_ca_system_score_codex":0.000079353944,"about_ca_system_score_gemma":0.00020615189,"threshold_uncertainty_score":0.9868775},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4404649597","doi":"10.1016/j.wmb.2024.11.006","title":"Harnessing wood waste for sustainable biofuel: A bibliometric analysis and review of valorisation strategies","year":2024,"lang":"en","type":"article","venue":"Waste Management Bulletin","topic":"Forest Biomass Utilization and Management","field":"Engineering","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Valorisation; Biofuel; Biorefinery; Business; Bioenergy; Sustainability; Sustainable development; Waste management; Environmental science; Engineering; Political science; Biology","score_opus":0.01014795545471459,"score_gpt":0.2411097132294289,"score_spread":0.2309617577747143,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404649597","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023044465,0.16423044,0.71207047,0.0018992497,0.0005925556,0.0035683978,0.000029968345,0.0011153436,0.09344913],"genre_scores_gemma":[0.9394051,0.04906921,0.0037979472,0.00014592841,0.00006872277,0.00024945443,0.00013323381,0.00006184396,0.007068577],"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989059,0.00002093032,0.00035335656,0.00025223702,0.00021907003,0.0002485322],"domain_scores_gemma":[0.99958086,0.000037105427,0.000049462244,0.00021116418,0.000079072364,0.000042364212],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00049351755,0.00016503809,0.00025454958,0.019625848,0.000050968545,0.0002755535,0.00011594033,0.000034725836,0.00015855467],"category_scores_gemma":[0.000018448787,0.00015801445,0.00012307598,0.035828415,0.000027227707,0.00015335911,0.00009308384,0.000040297527,0.000011705464],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002829224,0.00010763256,0.00021236968,0.431062,0.005098964,0.00007978901,0.00028442647,0.018681154,0.000182235,0.1966919,0.15646785,0.19110338],"study_design_scores_gemma":[0.0005848708,0.000086274486,0.00067004707,0.0033927911,0.0032005927,9.935188e-7,0.006446885,0.06329931,0.0005607751,0.0011719153,0.9200914,0.00049414265],"about_ca_topic_score_codex":0.0000129988,"about_ca_topic_score_gemma":0.0000034920565,"teacher_disagreement_score":0.9163606,"about_ca_system_score_codex":0.000052832274,"about_ca_system_score_gemma":0.0000061882724,"threshold_uncertainty_score":0.99148583},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4404821066","doi":"10.1108/ijem-09-2023-0439","title":"Decolonizing management education: a bibliometric analysis of Ubuntu","year":2024,"lang":"en","type":"article","venue":"International Journal of Educational Management","topic":"African cultural and philosophical studies","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Saint Mary's University","funders":"","keywords":"Higher education; Sociology; Political science","score_opus":0.03344538743121625,"score_gpt":0.39421207725637175,"score_spread":0.3607666898251555,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404821066","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.17686039,0.015014741,0.0009040356,0.16472623,0.012239856,0.00047286486,0.000049654536,0.00003534555,0.6296969],"genre_scores_gemma":[0.98738784,0.0028401194,0.0022390352,0.000302915,0.0010041714,0.000017583763,0.000014626407,0.000005445045,0.0061882706],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9979965,0.00006022351,0.0005100383,0.00015235847,0.0011467829,0.0001340876],"domain_scores_gemma":[0.9984761,0.00030719728,0.000251892,0.000076185,0.00079462066,0.00009404443],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0006026371,0.00009320664,0.00018624685,0.0235101,0.00012035168,0.00015855148,0.0005864934,0.000029288465,0.0015351849],"category_scores_gemma":[0.00010635335,0.0000769367,0.00028432236,0.03269036,0.00011575962,0.00029569547,0.000099924226,0.000093062,0.00003069676],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000020346732,0.00037031193,0.0066039604,0.00004655628,0.007128603,0.000014706669,0.0017113124,0.00015457356,0.0000030374342,0.8873207,0.012648172,0.08397775],"study_design_scores_gemma":[0.00016052822,0.000045356846,0.25277698,0.00027361154,0.0017289247,0.00000597114,0.0054099755,0.000045043183,0.0000051792786,0.08629373,0.65307903,0.0001756582],"about_ca_topic_score_codex":0.00010276565,"about_ca_topic_score_gemma":0.000041178195,"teacher_disagreement_score":0.81052744,"about_ca_system_score_codex":0.00035958918,"about_ca_system_score_gemma":0.00020260585,"threshold_uncertainty_score":0.99937755},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4404840467","doi":"10.32920/27926781.v1","title":"A cross-sectional description of open access publication costs, policies and impact in emergency medicine and critical care journals","year":2024,"lang":"en","type":"preprint","venue":"","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan; McMaster University","funders":"","keywords":"Cross-sectional study; Medicine; Family medicine; Medical emergency","score_opus":0.8802126189463708,"score_gpt":0.6923778208821509,"score_spread":0.18783479806421988,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404840467","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9267093,0.031970613,0.002050782,0.007074372,0.00095738296,0.0014266744,0.00017006874,0.0000055193836,0.029635284],"genre_scores_gemma":[0.99463594,0.0011376122,0.0003427307,0.00006782851,0.0001423345,0.00007318984,0.00003427319,0.000009983183,0.003556111],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98320854,0.002815996,0.009625195,0.001089416,0.0030714748,0.00018935897],"domain_scores_gemma":[0.98976225,0.0013808748,0.0027141136,0.0014822698,0.004442728,0.00021774501],"candidate_categories":["metaresearch","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.060841683,0.00030527235,0.0027339596,0.0017672299,0.00009315343,0.008508168,0.002421628,0.0002153135,0.037781466],"category_scores_gemma":[0.055918798,0.00014009241,0.00055187393,0.0015494418,0.00018897478,0.0011442659,0.0040080533,0.00042493216,0.000040140214],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000008362115,0.000021350408,0.9578884,0.00066903635,0.00011380304,7.28777e-7,0.0010833113,0.00001965913,0.00007760779,0.0052840095,0.03206183,0.0027718479],"study_design_scores_gemma":[0.000106670865,0.000033041095,0.9179659,0.0002809399,0.00013667047,0.000019408963,0.0011206607,0.0017498806,0.0000046763507,0.077377856,0.0010698718,0.0001344616],"about_ca_topic_score_codex":0.0036692305,"about_ca_topic_score_gemma":0.00094998226,"teacher_disagreement_score":0.072093844,"about_ca_system_score_codex":0.00012904151,"about_ca_system_score_gemma":0.00020645797,"threshold_uncertainty_score":0.9925211},"labels":[{"model":"gemma","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4404855999","doi":"10.5267/j.uscm.2024.8.013","title":"Mapping the landscape: A bibliometric analysis of digital leadership, service quality, and supply chain management","year":2024,"lang":"en","type":"article","venue":"Uncertain Supply Chain Management","topic":"Human Resource and Talent Management","field":"Business, Management and Accounting","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Supply chain; Quality (philosophy); Supply chain management; Service (business); Business; Service quality; Chain (unit); Process management; Computer science; Knowledge management; Operations management; Marketing; Economics","score_opus":0.06665395373874808,"score_gpt":0.27398403857808107,"score_spread":0.207330084839333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404855999","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.66650534,0.011302839,0.008951025,0.05544144,0.0013594863,0.006976339,0.00022026818,0.0016573288,0.24758597],"genre_scores_gemma":[0.98582476,0.00054898695,0.00015435759,0.0045047225,0.00041647645,0.00025136498,0.00045562192,0.00008598306,0.0077577378],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9952168,0.00007343652,0.0010915254,0.0011781583,0.0014539476,0.0009861202],"domain_scores_gemma":[0.99808294,0.00027727056,0.00036735204,0.0010604912,0.00014912162,0.000062838495],"candidate_categories":["metaepi_narrow","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.002353524,0.00065080466,0.0007571461,0.05562458,0.00038616892,0.0023338613,0.0011660071,0.00011111559,0.0009580288],"category_scores_gemma":[0.00003699119,0.00051388814,0.00046647454,0.10111526,0.00017678853,0.0010014334,0.0016905154,0.00029484124,0.00030957046],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020926108,0.0008054232,0.15816367,0.02457918,0.021618245,0.00079862145,0.003410107,0.010530215,0.000025777497,0.39441103,0.055873804,0.32957467],"study_design_scores_gemma":[0.0019238321,0.00005847306,0.34131765,0.0009354056,0.005019753,0.00000408287,0.043383498,0.099566996,0.000003666228,0.005651885,0.50056374,0.0015710122],"about_ca_topic_score_codex":0.0004266945,"about_ca_topic_score_gemma":0.0001768422,"teacher_disagreement_score":0.44468996,"about_ca_system_score_codex":0.00010289514,"about_ca_system_score_gemma":0.000010771646,"threshold_uncertainty_score":0.99995524},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4404905567","doi":"10.32920/27940890.v1","title":"Using Google Scholar to track the scholarly output of research groups","year":2024,"lang":"en","type":"preprint","venue":"","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; University of Saskatchewan","funders":"","keywords":"Track (disk drive); Computer science; World Wide Web; Data science; Information retrieval; Operating system","score_opus":0.9025638254961048,"score_gpt":0.6849236624817889,"score_spread":0.21764016301431588,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404905567","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9300077,0.005909509,0.0038734677,0.007415183,0.0027962888,0.0015436502,0.00018923279,0.00007080386,0.048194177],"genre_scores_gemma":[0.9672466,0.00019475259,0.009612361,0.00020355475,0.00042165825,0.00004235542,0.0000033153567,0.000048735892,0.02222666],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.95058846,0.0022335837,0.0019741892,0.0025582188,0.041124795,0.0015207471],"domain_scores_gemma":[0.9640794,0.011102684,0.0003148505,0.004223121,0.019359922,0.00092005654],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","open_science","insufficient_payload"],"category_scores_codex":[0.21355075,0.0003539001,0.00076347345,0.16067022,0.00079382927,0.033646576,0.013575235,0.0006235628,0.0010818058],"category_scores_gemma":[0.16511072,0.0001890878,0.00057069375,0.3500021,0.0005593324,0.00093879557,0.029978463,0.0077356813,0.0037106818],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018676909,0.0009506322,0.025927572,0.00048289003,0.0003925994,0.0002068549,0.004190054,0.0045802966,0.013057405,0.050208516,0.40872267,0.49109372],"study_design_scores_gemma":[0.00046264232,0.00056780706,0.035377398,0.0005628133,0.000057152545,0.000031482046,0.007175395,0.04024965,0.009352315,0.6663697,0.23874234,0.0010513344],"about_ca_topic_score_codex":0.0012840384,"about_ca_topic_score_gemma":0.00005900373,"teacher_disagreement_score":0.61616117,"about_ca_system_score_codex":0.0004855162,"about_ca_system_score_gemma":0.0018273437,"threshold_uncertainty_score":0.9998313},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4404968867","doi":"10.3390/jrfm17120545","title":"Trends in the Literature About the Adoption of Digital Banking in Emerging Economies: A Bibliometric Analysis","year":2024,"lang":"en","type":"article","venue":"Journal of risk and financial management","topic":"FinTech, Crowdfunding, Digital Finance","field":"Business, Management and Accounting","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Emerging markets; Scopus; Financial inclusion; Business; Digital transformation; Key (lock); Emerging technologies; Financial services; Digital economy; Marketing; Finance; Computer science; Computer security; Political science; World Wide Web","score_opus":0.009667818564677032,"score_gpt":0.23300441672904282,"score_spread":0.2233365981643658,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404968867","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9838979,0.006776125,0.0031990528,0.00043656325,0.00033005004,0.000120200995,0.000011285814,0.00001068123,0.005218121],"genre_scores_gemma":[0.9976116,0.001832631,0.00005251274,0.00008910826,0.0003248861,0.000005209744,0.0000069240914,0.000009370427,0.00006780285],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99873155,0.000013327798,0.0006145685,0.00017574328,0.0002654307,0.00019938806],"domain_scores_gemma":[0.99929625,0.00009811855,0.00040317158,0.00013505547,0.00006137432,0.000006019941],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0012167178,0.00014811287,0.0002870501,0.043224514,0.000073998664,0.001219404,0.0003133119,0.000048486516,0.000012685324],"category_scores_gemma":[0.00011533244,0.00009150941,0.00023277011,0.075757116,0.000045995246,0.0017826441,0.00012853337,0.00033066602,0.0000036777717],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006720269,0.000086746106,0.16716787,0.00027221607,0.00009157222,0.00017042832,0.0010313868,0.0011444,0.0000011134692,0.029463714,0.00090190256,0.79960144],"study_design_scores_gemma":[0.0003410342,0.000027038259,0.9192824,0.0005545583,0.00028649383,0.0000062177664,0.0007241221,0.002029783,7.2890305e-7,0.0075579016,0.06905179,0.00013793755],"about_ca_topic_score_codex":0.000053956326,"about_ca_topic_score_gemma":0.00012774342,"teacher_disagreement_score":0.7994635,"about_ca_system_score_codex":0.000044447544,"about_ca_system_score_gemma":0.000007648279,"threshold_uncertainty_score":0.99981743},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4405046983","doi":"10.1016/j.jtte.2024.09.004","title":"A review of lithium-ion battery state of health and remaining useful life estimation methods based on bibliometric analysis","year":2024,"lang":"en","type":"review","venue":"Journal of Traffic and Transportation Engineering (English Edition)","topic":"Advanced Battery Technologies Research","field":"Engineering","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Estimation; State of health; Lithium (medication); Battery (electricity); Lithium-ion battery; Computer science; State (computer science); Reliability engineering; Psychology; Engineering; Systems engineering; Psychiatry; Algorithm; Physics","score_opus":0.03428161157375896,"score_gpt":0.34524311250764567,"score_spread":0.3109615009338867,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405046983","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00033147374,0.7898158,0.20893812,0.000026586986,0.0003994658,0.00024506482,0.0001350998,0.00010555288,0.0000028387096],"genre_scores_gemma":[0.0031021778,0.9727849,0.023717223,0.000019069295,0.00008492297,0.000021227366,0.00020689447,0.00006236476,0.0000012156701],"study_design_codex":"design_other","study_design_gemma":"systematic_review","domain_scores_codex":[0.9969158,0.00010929677,0.0020362483,0.00022839058,0.0004873327,0.00022289316],"domain_scores_gemma":[0.99775064,0.00073576276,0.0008017109,0.00020503397,0.00034217874,0.00016464684],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0017078457,0.00035741704,0.0020517618,0.0297966,0.000025775002,0.000029807137,0.00015997523,0.00016113094,0.0000107450605],"category_scores_gemma":[0.00072143,0.0003092823,0.00043877325,0.025675045,0.000048293816,0.00031659377,0.000005570881,0.0007894622,3.1611313e-7],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004164845,0.000020250585,0.0000030342292,0.1505236,0.0005591709,0.0000063663406,0.00009077843,0.37913626,5.5143073e-7,0.0000053526805,0.00024561846,0.46940485],"study_design_scores_gemma":[0.0010866808,0.0012662853,0.0016289282,0.46545127,0.010724648,0.000042848133,0.00013591931,0.35403132,0.00003693228,0.000027855474,0.16429326,0.0012740603],"about_ca_topic_score_codex":4.920928e-7,"about_ca_topic_score_gemma":3.6016772e-7,"teacher_disagreement_score":0.46813077,"about_ca_system_score_codex":0.00010172031,"about_ca_system_score_gemma":0.00016437072,"threshold_uncertainty_score":0.9999359},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4405182919","doi":"10.1088/1755-1315/1420/1/012033","title":"Assessment of the influence of the green economy on sustainable development: bibliometric approach","year":2024,"lang":"en","type":"article","venue":"IOP Conference Series Earth and Environmental Science","topic":"Sustainable Supply Chain Management","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Sustainable development; Quarter (Canadian coin); China; Subject (documents); Green economy; Investment (military); Impact factor; Economy; Political science; Business; Regional science; Economics; Geography; Computer science; Library science","score_opus":0.010064638952251325,"score_gpt":0.19995940141832075,"score_spread":0.18989476246606943,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405182919","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9714458,0.00006436256,0.0000419569,0.00038202916,0.00007626119,0.00045043547,0.0000022769775,0.000017688131,0.02751918],"genre_scores_gemma":[0.99714404,0.000021180691,0.00013490634,0.0002376984,0.00002780742,0.000027482349,0.0000021032233,0.000008200286,0.0023965992],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9984376,0.000014328173,0.00025165686,0.0003787315,0.000538903,0.00037879156],"domain_scores_gemma":[0.999398,0.000027285936,0.00014626759,0.00036227025,0.000044411543,0.000021757178],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006763836,0.00016409706,0.00014413647,0.001983152,0.00044674656,0.00036662733,0.0007722314,0.000030017924,0.00008025925],"category_scores_gemma":[0.00004334461,0.00010121866,0.000044281504,0.009383682,0.0012759471,0.0016515602,0.0011980217,0.000130466,0.0000062385234],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000048213195,0.00041536198,0.2246862,0.0045138933,0.000105538595,0.000021905136,0.001725807,0.013905624,0.019005397,0.6796238,0.0001862791,0.05576198],"study_design_scores_gemma":[0.00012673509,0.000023129873,0.95723724,0.000071002454,0.000017576505,0.0000021671517,0.004725247,0.0025388876,0.003911434,0.001265815,0.029912591,0.0001681615],"about_ca_topic_score_codex":0.00030732687,"about_ca_topic_score_gemma":0.000012142185,"teacher_disagreement_score":0.73255104,"about_ca_system_score_codex":0.00007830177,"about_ca_system_score_gemma":0.00017720577,"threshold_uncertainty_score":0.47012803},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4405258924","doi":"10.18805/ag.d-6202","title":"Bibliometric Investigation of Climate Change Literature in Fisheries using Dimensions.AI Database","year":2024,"lang":"en","type":"article","venue":"Agricultural Science Digest - A Research Journal","topic":"Ocean Acidification Effects and Responses","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Climate change; Scope (computer science); Fisheries science; Fisheries management; Fishery; Geography; Bibliometrics; Globe; Political science; Environmental resource management; Environmental science; Fishing; Library science; Ecology; Computer science; Psychology","score_opus":0.110458689332324,"score_gpt":0.352534591842544,"score_spread":0.24207590251022,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405258924","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99124295,0.0062854765,0.0000011317653,0.0016178829,0.00033400068,0.00020330364,0.00010513882,0.000016163369,0.0001939423],"genre_scores_gemma":[0.99482566,0.0046085906,0.00025434175,0.00004472929,0.00019353037,9.781061e-7,0.00004698575,0.0000020605298,0.000023128485],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99712276,0.00037614876,0.00028718045,0.00029765524,0.0013132474,0.0006029761],"domain_scores_gemma":[0.9985307,0.0004412903,0.00006775742,0.0001225344,0.0005266507,0.00031107973],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0040593334,0.00011256164,0.00013227481,0.014081232,0.0006602285,0.0014534693,0.00043046116,0.00005321458,0.00012058416],"category_scores_gemma":[0.00073437864,0.000057209978,0.00004729341,0.086209245,0.00069714873,0.0039854585,0.00005731335,0.0007014981,0.000023800185],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006858106,0.00004709736,0.80580235,0.00019508404,0.000009468099,0.00033790947,0.0029844574,0.00026712022,0.14369763,0.00031112146,0.0007913992,0.045487788],"study_design_scores_gemma":[0.00007472363,0.00012612695,0.9917087,0.0006972439,0.0000032092603,0.00030784137,0.0006774983,0.0042527127,0.0016076209,0.00013317546,0.00031101154,0.0001001332],"about_ca_topic_score_codex":0.0005530569,"about_ca_topic_score_gemma":0.00019777913,"teacher_disagreement_score":0.18590637,"about_ca_system_score_codex":0.000039046456,"about_ca_system_score_gemma":0.0002551666,"threshold_uncertainty_score":0.9995831},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4405398818","doi":"10.22159/ajpcr.2024v17i12.53196","title":"PEDIATRIC ACCEPTANCE AND SWALLOWABILITY OF MINITABLETS: A BIBLIOMETRIC ANALYSIS","year":2024,"lang":"en","type":"article","venue":"Asian Journal of Pharmaceutical and Clinical Research","topic":"Pharmaceutical studies and practices","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Web of science; Library science; Political science; Computer science; MEDLINE; Law","score_opus":0.32203531426283083,"score_gpt":0.6036811630221336,"score_spread":0.28164584875930276,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405398818","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92003727,0.06262929,0.00014294875,0.013783307,0.00018641923,0.0001822454,0.000009369762,0.00000910393,0.0030200402],"genre_scores_gemma":[0.9236907,0.07504577,0.00047541916,0.00018240215,0.00048755854,0.0000019420368,5.25649e-7,0.000009730094,0.00010595916],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9962413,0.0006827479,0.0011921476,0.00034044756,0.0011049748,0.00043839277],"domain_scores_gemma":[0.9905216,0.007448407,0.00014238994,0.00015447642,0.00071169884,0.0010214415],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.011070511,0.00013591032,0.0007813014,0.010808477,0.0000974857,0.000098619385,0.00017208059,0.00013512398,0.0006924666],"category_scores_gemma":[0.007390471,0.00008898592,0.00039871054,0.046474192,0.0007550359,0.00025485005,0.00021919147,0.0019509423,0.000010750527],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013672675,0.00073783676,0.57460916,0.00089255534,0.0024128228,0.00043323622,0.000119959805,2.2923236e-7,0.0001939468,0.0002812233,0.0011854925,0.4177663],"study_design_scores_gemma":[0.0020384598,0.0022045511,0.92782664,0.0001281146,0.005579693,0.00018275403,0.0005264903,0.001456131,0.0001631223,0.001150849,0.058598466,0.00014473466],"about_ca_topic_score_codex":0.000005218601,"about_ca_topic_score_gemma":0.0000017689622,"teacher_disagreement_score":0.41762155,"about_ca_system_score_codex":0.00003287743,"about_ca_system_score_gemma":0.0002289122,"threshold_uncertainty_score":0.97379345},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4405435813","doi":"10.26599/jgse.2024.9280031","title":"Development, hotspots and trend directions of groundwater numerical simulation: A bibliometric and visualization analysis","year":2024,"lang":"en","type":"article","venue":"Journal of Groundwater Science and Engineering","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Visualization; Groundwater; Environmental science; Computer science; Geography; Geology; Data mining; Geotechnical engineering","score_opus":0.07082533311168879,"score_gpt":0.36588950909493956,"score_spread":0.29506417598325074,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405435813","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.69530135,0.00061889965,0.30357885,0.00005822877,0.00037537585,0.000024018296,6.3067654e-7,0.000012285532,0.000030377845],"genre_scores_gemma":[0.9959467,0.00006753082,0.0038156225,0.000009177704,0.000041103576,5.2601314e-7,5.481153e-7,0.000003685929,0.000115120005],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99783593,0.000020777095,0.00056742324,0.00032267612,0.0010873475,0.00016584934],"domain_scores_gemma":[0.9990291,0.0003246744,0.00012012732,0.00014668453,0.00024986544,0.0001295755],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0047942563,0.0000918813,0.00022376482,0.024542734,0.00017764018,0.0015350472,0.0002295871,0.000022256672,0.00001548258],"category_scores_gemma":[0.00047236486,0.00006112262,0.000042855954,0.04523438,0.00011571603,0.0013838636,0.00019338854,0.00006671055,0.000001214714],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000027526736,0.00014004011,0.036237475,0.00013877133,0.00044862152,0.00004427724,0.008026653,0.086043276,0.0058293347,0.0009888996,0.00038662954,0.8616885],"study_design_scores_gemma":[0.00010498686,0.000059846174,0.2279095,0.000048136557,0.00008967939,0.000045627206,0.00023048931,0.756183,0.00019094974,0.00010144799,0.014927627,0.00010867292],"about_ca_topic_score_codex":0.000014492668,"about_ca_topic_score_gemma":0.00000488364,"teacher_disagreement_score":0.86157984,"about_ca_system_score_codex":0.00003713783,"about_ca_system_score_gemma":0.000040467094,"threshold_uncertainty_score":0.99950147},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4405633240","doi":"10.1109/icdabi63787.2024.10800738","title":"Machine Learning for Detecting Financial Statement Fraud: A Bibliometric Analysis","year":2024,"lang":"en","type":"article","venue":"","topic":"Imbalanced Data Classification Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Financial statement; Computer science; Statement (logic); Financial statement analysis; Accounting; Data science; Artificial intelligence; Financial analysis; Business; Political science","score_opus":0.03248137288188185,"score_gpt":0.32059494638824376,"score_spread":0.2881135735063619,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405633240","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00087769434,0.0004821394,0.99610513,0.00040185623,0.00012497787,0.00018568031,0.000012421325,0.0012626561,0.00054745056],"genre_scores_gemma":[0.680861,0.00008351745,0.31793642,0.00011391794,0.000036567195,0.00010453114,0.000031395317,0.000008475903,0.0008241661],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987726,0.00003592061,0.00028592395,0.0004350813,0.00023703765,0.00023343848],"domain_scores_gemma":[0.99916995,0.00026939117,0.00006610023,0.00035420258,0.00009274593,0.000047590143],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00086075993,0.00010156189,0.00014454662,0.041789833,0.00013636755,0.00051433797,0.0005098724,0.000041621664,0.000051624127],"category_scores_gemma":[0.00037136758,0.00008762251,0.00011872995,0.13536423,0.000010610008,0.0005466641,0.00016944547,0.00013876698,0.000034280456],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000002988028,0.000026509217,0.0014438091,0.00005701039,0.00012873687,0.0000047797676,0.00014002321,0.00033344066,0.0014198666,0.046185795,0.002871654,0.9473854],"study_design_scores_gemma":[0.00006791637,0.00007963761,0.0027915465,0.000009792168,0.00004433062,0.000001181879,0.0000054458574,0.8943286,0.009399238,0.002084208,0.09104623,0.00014187157],"about_ca_topic_score_codex":0.000057039866,"about_ca_topic_score_gemma":0.00001644976,"teacher_disagreement_score":0.9472435,"about_ca_system_score_codex":0.00007529418,"about_ca_system_score_gemma":0.000060454757,"threshold_uncertainty_score":0.9690707},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4405795835","doi":"10.18280/ijdne.190638","title":"Unraveling Factors Affecting Food Waste Globally: A PRISMA-Based Systematic Literature Review and Bibliometric Analysis","year":2024,"lang":"en","type":"article","venue":"International Journal of Design & Nature and Ecodynamics","topic":"Food Waste Reduction and Sustainability","field":"Agricultural and Biological Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Systematic review; Food waste; Environmental science; Engineering; Waste management; MEDLINE; Biology","score_opus":0.013660441814205225,"score_gpt":0.2641514020676028,"score_spread":0.25049096025339757,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405795835","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8021666,0.1936701,0.0017251307,0.0016652273,0.00046467478,0.0002340936,0.000036784182,0.000021609769,0.00001579126],"genre_scores_gemma":[0.9884195,0.010843806,0.0003232419,0.00021883208,0.00015245714,0.0000014594245,0.000019759782,0.0000015833917,0.000019386425],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998583,0.00017096548,0.00047153712,0.00020580608,0.00043075616,0.00013791685],"domain_scores_gemma":[0.99845606,0.0005936273,0.00025404475,0.000038656468,0.00055055413,0.00010706986],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011239244,0.0001590315,0.00040293412,0.0031860226,0.00007254903,0.0005089645,0.00023404602,0.00016784084,0.00000873581],"category_scores_gemma":[0.0005136983,0.00006312589,0.00025266744,0.017714445,0.000025606254,0.0002813276,0.000040233845,0.0004573227,1.7625095e-7],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001245957,0.002027206,0.062096424,0.25880745,0.04946216,0.0025292775,0.0053900504,0.016497813,0.023966597,0.008058294,0.0033514497,0.5665673],"study_design_scores_gemma":[0.0034012264,0.011241532,0.18559632,0.26657003,0.030253677,0.008520229,0.054820515,0.41636822,0.0028455837,0.012781281,0.0023093207,0.005292077],"about_ca_topic_score_codex":0.0000028783832,"about_ca_topic_score_gemma":0.000013814356,"teacher_disagreement_score":0.56127524,"about_ca_system_score_codex":0.00009272777,"about_ca_system_score_gemma":0.000029977,"threshold_uncertainty_score":0.85112035},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4405819503","doi":"10.1007/s12144-024-06876-8","title":"Knowledge Atlas and emerging trends on the impact of mindfulness on tumors: a bibliometric analysis","year":2024,"lang":"en","type":"article","venue":"Current Psychology","topic":"Mindfulness and Compassion Interventions","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Psychology; Mindfulness; Atlas (anatomy); Clinical psychology; Medicine","score_opus":0.08791418999850688,"score_gpt":0.4655505774812949,"score_spread":0.37763638748278805,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405819503","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9404499,0.0051779835,0.009530031,0.0008038644,0.0031363245,0.00013491833,0.00013009859,0.00006948151,0.04056738],"genre_scores_gemma":[0.997563,0.0000058054034,0.0000011619655,0.000024290026,0.00017155011,0.00061656506,0.000020905209,0.000021792299,0.0015749031],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9979999,0.00036048063,0.0005025295,0.0006197078,0.00016141415,0.0003559879],"domain_scores_gemma":[0.99850243,0.00050266,0.0001454458,0.0006717024,0.00006540027,0.000112371294],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00046563116,0.00025810747,0.00043181275,0.050474275,0.00010804439,0.000059781487,0.00035617588,0.00010755942,0.028032921],"category_scores_gemma":[0.000040923027,0.00015569168,0.0007310217,0.074165665,0.00015872008,0.000042705353,0.000059491267,0.0004452541,0.0004502449],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001243875,0.00117435,0.0032067206,0.000029235976,0.0012339788,0.000012331184,0.00040242862,0.000014414719,0.000007445756,0.1270391,0.17637719,0.6903784],"study_design_scores_gemma":[0.0013099604,0.0023008473,0.9247608,0.00060221675,0.0009053765,0.000059866306,0.00037148446,0.0017701689,0.00032526007,0.012842876,0.054160435,0.00059068797],"about_ca_topic_score_codex":0.000035204073,"about_ca_topic_score_gemma":0.000007517045,"teacher_disagreement_score":0.9215541,"about_ca_system_score_codex":0.0000333083,"about_ca_system_score_gemma":0.000020168925,"threshold_uncertainty_score":0.97285557},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4405967087","doi":"10.1101/2024.12.26.24319521","title":"Citation Contamination of Systematic Review Literature in the Life Sciences by Paper Mills","year":2024,"lang":"en","type":"preprint","venue":"medRxiv","topic":"Academic Writing and Publishing","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Contamination; Citation; Environmental science; Library science; Computer science; Biology; Ecology","score_opus":0.04335486111841079,"score_gpt":0.27953419382027,"score_spread":0.2361793327018592,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405967087","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.055630725,0.8924693,0.0000048584934,0.01576533,0.0010022137,0.0013922672,0.00016345832,0.000053766453,0.03351812],"genre_scores_gemma":[0.98483133,0.009814827,0.000009523048,0.0032595575,0.00037628095,0.00017053768,0.000108855595,0.000011558305,0.0014175444],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.99826425,0.0004267979,0.0005732651,0.000216849,0.00041095435,0.00010786796],"domain_scores_gemma":[0.99890995,0.00042929366,0.00031109477,0.00018228353,0.00015048901,0.000016878477],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004086751,0.00013363363,0.00037957067,0.00013088592,0.000089732996,0.0004852043,0.00047782488,0.00011248055,0.0000705705],"category_scores_gemma":[0.001755862,0.0000748682,0.00009063842,0.0001287258,0.0001253441,0.00019675054,0.00009050913,0.00071498845,0.000010915105],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[9.1160564e-7,0.000021751826,0.00010958427,0.7138844,0.0000345997,0.0000029584503,0.0971186,0.000003112061,0.000017091974,0.109956004,0.07876916,0.00008180772],"study_design_scores_gemma":[0.000088007364,0.000036698184,0.00013182223,0.948462,0.00046634287,0.000004291333,0.013069879,0.00025966024,0.0000093037015,0.020305809,0.016845334,0.00032082899],"about_ca_topic_score_codex":0.000061073595,"about_ca_topic_score_gemma":0.00003510474,"teacher_disagreement_score":0.9292006,"about_ca_system_score_codex":0.000014372125,"about_ca_system_score_gemma":0.00005694179,"threshold_uncertainty_score":0.46788362},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4406049137","doi":"10.1007/s44279-026-00541-3","title":"Mapping the Evolution of Agriculture 4.0: A Bibliometric Analysis of Research Trends","year":2025,"lang":"en","type":"preprint","venue":"Discover Agriculture","topic":"Digitalization and Economic Development in Agriculture","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Agriculture; Regional science; Bibliometrics; Data science; Geography; Political science; Library science; Computer science; Archaeology","score_opus":0.04163167384498705,"score_gpt":0.2796986028595091,"score_spread":0.23806692901452203,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406049137","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.333521,0.0048832777,0.0012909996,0.0036373774,0.0025746494,0.0017271199,0.0013401412,0.00028110613,0.6507443],"genre_scores_gemma":[0.9706753,0.00017469273,0.000056786288,0.00022793384,0.0006774933,0.00010675203,0.0037365006,0.000013675041,0.024330897],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.996838,0.00005689054,0.000980371,0.0007116307,0.000976345,0.00043678837],"domain_scores_gemma":[0.99680436,0.00013162926,0.0010998106,0.0006082238,0.001333858,0.000022113725],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00094625057,0.00045048934,0.00091462955,0.057326637,0.00023331352,0.0006686616,0.0013290866,0.0005342984,0.00029440777],"category_scores_gemma":[0.00021553301,0.00022907657,0.0007937857,0.24236757,0.00014962567,0.0007417741,0.0016305177,0.0007307774,0.00002592426],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000044302604,0.00074088277,0.1455414,0.0024930602,0.006902104,0.000003870398,0.0012448038,0.03443034,0.00044684106,0.07825738,0.7254378,0.004457204],"study_design_scores_gemma":[0.0002769487,0.0000054171264,0.87722063,0.00050623843,0.0012509683,6.0101536e-7,0.005051945,0.0007230912,0.000059865546,0.0016493482,0.11276473,0.0004902071],"about_ca_topic_score_codex":0.0017018212,"about_ca_topic_score_gemma":0.0006171028,"teacher_disagreement_score":0.73167926,"about_ca_system_score_codex":0.00028036398,"about_ca_system_score_gemma":0.00012459245,"threshold_uncertainty_score":0.95335776},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4406195046","doi":"10.1186/s12961-024-01282-9","title":"Investigating the trustworthiness of research evidence used to inform public health policy: a qualitative interview study on the use of predatory journal citations in policy documents","year":2025,"lang":"en","type":"article","venue":"Health Research Policy and Systems","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Wilfrid Laurier University; Ottawa Hospital; University of Ottawa","funders":"Social Sciences and Humanities Research Council of Canada; Telfer School of Management, University of Ottawa; University of Ottawa","keywords":"Health services research; Public health; Trustworthiness; Health administration; Health policy; Qualitative research; Peer review; Health economics; Public health policy; Social policy; Public policy; Medicine; Political science; Psychology; Nursing; Sociology; Social psychology; Social science; Law","score_opus":0.9789561516613032,"score_gpt":0.7890776357586877,"score_spread":0.1898785159026155,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406195046","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.61303836,0.0021375238,0.00027531033,0.37945098,0.00013626595,0.004500573,0.00008405233,0.0000074723507,0.00036946608],"genre_scores_gemma":[0.99493825,0.0010910936,0.000049985654,0.0018912236,0.00016706654,0.00033812894,9.78632e-7,0.000012414998,0.001510843],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.8479736,0.102214925,0.007009926,0.0014682339,0.036849327,0.0044839876],"domain_scores_gemma":[0.65074694,0.30676112,0.0025138643,0.004054273,0.032396607,0.0035272413],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.45202458,0.00020401517,0.00084055844,0.11287322,0.0023505567,0.003964072,0.0035366716,0.000105711224,0.000009929911],"category_scores_gemma":[0.6065224,0.00010669633,0.00010471061,0.31289807,0.0012987694,0.001144783,0.0019023222,0.0020530357,0.000015007494],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011042464,0.00091316976,0.09692715,0.0013241534,0.00011949482,0.0000038842986,0.41890675,0.00036742206,0.00003490135,0.35172758,0.046317093,0.08324799],"study_design_scores_gemma":[0.0015646872,0.0053514973,0.43376574,0.0052957395,0.0000019874858,0.000012236664,0.47000244,0.004299806,0.0000135373175,0.026077528,0.0533712,0.00024360596],"about_ca_topic_score_codex":0.111512884,"about_ca_topic_score_gemma":0.004136169,"teacher_disagreement_score":0.3818999,"about_ca_system_score_codex":0.002065063,"about_ca_system_score_gemma":0.02634879,"threshold_uncertainty_score":0.9989483},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["research_integrity"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4406214401","doi":"10.1080/16583655.2025.2449616","title":"Global research trends in zero-waste cities: a comprehensive bibliometric analysis","year":2025,"lang":"en","type":"article","venue":"Journal of Taibah University for Science","topic":"Municipal Solid Waste Management","field":"Environmental Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Zero (linguistics); Zero waste; Environmental science; Regional science; Environmental planning; Geography; Waste management; Engineering","score_opus":0.05237163795772578,"score_gpt":0.35433470798610983,"score_spread":0.3019630700283841,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406214401","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95927304,0.00005507746,0.008128757,0.0006099587,0.00015497064,0.0001306506,0.000011261561,0.000004819649,0.031631447],"genre_scores_gemma":[0.99545604,0.000039887727,0.0015660039,0.00006013292,0.000008442546,2.6034544e-7,5.1915947e-7,0.0000017311733,0.002866989],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.99782515,0.00010890336,0.0002703081,0.00029803885,0.0010106078,0.0004870156],"domain_scores_gemma":[0.9989794,0.00016606669,0.00018992434,0.00029526447,0.00020664373,0.0001627147],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0022253583,0.000096301366,0.00026481945,0.075582094,0.00037612885,0.00010193476,0.0016954928,0.000039541068,0.00014447716],"category_scores_gemma":[0.00015167767,0.00009415779,0.00020285614,0.38271475,0.0009620433,0.00073514384,0.0010464919,0.00017513937,0.0000058573405],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00040765834,0.00039033536,0.27946007,0.000033023818,0.0002991331,0.00015716102,0.000809821,0.64735985,0.0003390452,0.0084073,0.017814731,0.044521842],"study_design_scores_gemma":[0.0022235063,0.0005614497,0.8502824,0.00006522074,0.00032848146,0.000007624647,0.01173623,0.08244462,0.00012475705,0.0028850555,0.049044874,0.00029579038],"about_ca_topic_score_codex":0.0010043484,"about_ca_topic_score_gemma":0.00072548987,"teacher_disagreement_score":0.5708223,"about_ca_system_score_codex":0.0020098668,"about_ca_system_score_gemma":0.00009558038,"threshold_uncertainty_score":0.9348954},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4406307690","doi":"10.1097/yco.0000000000000984","title":"Urban air pollution and child neurodevelopmental conditions: a systematic bibliometric review","year":2024,"lang":"en","type":"review","venue":"Current Opinion in Psychiatry","topic":"Air Quality and Health Impacts","field":"Environmental Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Children's Hospital of Eastern Ontario; Carleton University","funders":"","keywords":"Neuroinflammation; Cognition; Medicine; Environmental health; Psychology; Disease; Neuroscience; Psychiatry; Pathology","score_opus":0.10410963114153024,"score_gpt":0.4260448806821304,"score_spread":0.32193524954060015,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406307690","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[9.855423e-7,0.9804578,0.000010483972,0.0017232196,0.014300158,0.0029474513,0.0001469843,0.000077323326,0.00033561958],"genre_scores_gemma":[0.0000043769783,0.99775594,0.00009659803,0.0009939705,0.00031777407,0.0004818966,0.000270885,0.000043559165,0.00003498875],"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","domain_scores_codex":[0.9962821,0.0004940724,0.001552016,0.0007212254,0.0005094642,0.00044109454],"domain_scores_gemma":[0.99870443,0.000073437404,0.00054402015,0.00039619792,0.0000060436164,0.00027588394],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0010739681,0.0005025266,0.0014761934,0.005809488,0.00017763318,0.00005575953,0.00031972222,0.00020508406,0.00022156468],"category_scores_gemma":[0.0001445771,0.00039451447,0.00025824158,0.023775479,0.000117100106,0.0002445846,0.00025943088,0.0008481627,0.0014513222],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[7.7072895e-7,0.00008079423,0.00007475355,0.6695751,0.000012660119,2.4396837e-7,0.000026038726,1.6513654e-7,2.573962e-10,0.00029830073,0.28050724,0.04942392],"study_design_scores_gemma":[0.00011751437,0.000023785871,0.00078466116,0.45636758,0.00017852006,0.00007472069,0.000007281554,0.0000049764685,3.5057315e-10,0.00009583916,0.54210883,0.00023625493],"about_ca_topic_score_codex":0.000012503374,"about_ca_topic_score_gemma":0.0000037307361,"teacher_disagreement_score":0.26160163,"about_ca_system_score_codex":0.000419471,"about_ca_system_score_gemma":0.00020072606,"threshold_uncertainty_score":0.9998507},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4406379670","doi":"10.1080/0886022x.2024.2444383","title":"Bibliometric study and visual analysis of postoperative diabetes mellitus in kidney transplant recipients based on WoSCC database","year":2025,"lang":"en","type":"article","venue":"Renal Failure","topic":"Renal Transplantation Outcomes and Treatments","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Diabetes mellitus; Kidney transplant; Renal transplant; Internal medicine; Database; Intensive care medicine; Kidney transplantation; Kidney; Endocrinology","score_opus":0.014915571955475946,"score_gpt":0.32062597105966517,"score_spread":0.30571039910418923,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406379670","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9979238,0.00018674064,0.00019984512,0.0005634067,0.00003105691,0.0004974295,0.00039565563,0.000012379789,0.00018968791],"genre_scores_gemma":[0.99825466,0.00013552517,0.00028959778,0.0005417458,0.000004137672,0.000025601124,0.00063976157,0.0000074758027,0.00010151497],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9986793,0.00011115176,0.00035323223,0.0003311878,0.0003464151,0.00017875462],"domain_scores_gemma":[0.9992985,0.00024906266,0.00005963377,0.00017633749,0.00007504447,0.00014139723],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00020374835,0.0001769922,0.000551312,0.025759956,0.00004017186,0.000018077248,0.000053924898,0.000056353576,0.00006971506],"category_scores_gemma":[0.000063079526,0.0001272763,0.00011344872,0.03609857,0.000033109613,0.000069710855,0.000009366174,0.00013803749,0.000002391722],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00065084314,0.0015678098,0.991623,0.00013962499,0.0016071214,0.00011716641,0.00022380086,0.00011547616,0.0019137159,0.000020900397,0.0000300504,0.0019904866],"study_design_scores_gemma":[0.0057774517,0.000986305,0.9805923,0.0011344324,0.002438367,7.7233074e-7,0.00009528629,0.0046784226,0.0041226544,0.0000019426006,0.00006971867,0.00010231354],"about_ca_topic_score_codex":0.00014796609,"about_ca_topic_score_gemma":0.00011163799,"teacher_disagreement_score":0.011030671,"about_ca_system_score_codex":0.000044571956,"about_ca_system_score_gemma":0.00008061477,"threshold_uncertainty_score":0.98528224},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4406389386","doi":"10.1177/20426445241310277","title":"Use of wood-based panels in engineered flooring: A bibliometric and systematic review","year":2025,"lang":"en","type":"article","venue":"International Wood Products Journal","topic":"Wood Treatment and Properties","field":"Engineering","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval; Natural Sciences and Engineering Research Council of Canada","funders":"Coordenação de Aperfeiçoamento de Pessoal de Nível Superior; Fundação de Amparo à Pesquisa do Estado de São Paulo","keywords":"Engineered wood; Oriented strand board; Context (archaeology); Veneer; Fiberboard; Hardwood; Solid wood; Woodworking; Engineering; Materials science; Pulp and paper industry; Composite material; Environmental science; Layer (electronics); Mechanical engineering","score_opus":0.036426676284615496,"score_gpt":0.2550308700741131,"score_spread":0.21860419378949764,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406389386","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7397406,0.24990506,0.00058577355,0.0054026265,0.0022105444,0.0012473601,0.00002808257,0.00012765964,0.0007522914],"genre_scores_gemma":[0.99025625,0.008363199,0.0008689461,0.000102681886,0.0000594527,0.000027376194,0.0000048721977,0.000014150264,0.0003030467],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.99891394,0.00004661355,0.0005353236,0.000117539494,0.0002519875,0.00013456288],"domain_scores_gemma":[0.9993989,0.00011953737,0.00008554702,0.00014205421,0.00021735283,0.000036641053],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00036702165,0.00014035948,0.00035089307,0.0075977175,0.000020896854,0.00009532055,0.0001712915,0.00003434596,0.00001880318],"category_scores_gemma":[0.0007577484,0.00010991177,0.00004998646,0.006147326,0.000016094382,0.00029145775,0.000022884573,0.00017556797,0.0000036680765],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024833222,0.0011028756,0.017350351,0.8610673,0.0061949496,0.00039255488,0.0012476294,0.038162425,0.033528477,0.0014248431,0.017938096,0.021342142],"study_design_scores_gemma":[0.0074593774,0.00057979405,0.041385703,0.6649482,0.002152524,0.0009896876,0.00018133377,0.03703409,0.2313329,0.0011866415,0.010823995,0.0019257097],"about_ca_topic_score_codex":0.0000050746607,"about_ca_topic_score_gemma":0.0000010272354,"teacher_disagreement_score":0.25051567,"about_ca_system_score_codex":0.000077945566,"about_ca_system_score_gemma":0.00003569167,"threshold_uncertainty_score":0.67793643},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4406392207","doi":"10.1007/s11255-025-04363-y","title":"Current applications and research trends of ultrasound examination in acute kidney injury assessment: a bibliometric analysis","year":2025,"lang":"en","type":"article","venue":"International Urology and Nephrology","topic":"Acute Kidney Injury Research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Tianjin Science and Technology Program","keywords":"Medicine; Nephrology; Acute kidney injury; Ultrasound; Intensive care medicine; Internal medicine; Medical physics; Radiology","score_opus":0.029691149168706122,"score_gpt":0.44472448742926973,"score_spread":0.4150333382605636,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406392207","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96940595,0.0008785238,0.0030252517,0.01649173,0.00013198808,0.00039102606,0.00034903878,0.000020399835,0.009306099],"genre_scores_gemma":[0.99531496,0.002375365,0.00026994006,0.00093811046,0.000042897525,0.0002536245,0.00035423978,0.0000057583043,0.00044510292],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99832296,0.00033224272,0.0003671144,0.00040183985,0.0003186266,0.00025722507],"domain_scores_gemma":[0.99840915,0.0007468132,0.000079309895,0.00021432112,0.00042359676,0.00012681642],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0013656564,0.000104909035,0.0003386955,0.089109875,0.00006964038,0.000016454842,0.00018571391,0.0002019286,0.00027622122],"category_scores_gemma":[0.00045634928,0.00010049337,0.00005295814,0.045884673,0.0005832289,0.00009300179,0.00018657807,0.0005922447,0.0000032783632],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006175229,0.0004770376,0.8618648,0.000038805796,0.0016354653,0.000012829707,0.000082495186,7.031015e-7,0.008584231,0.022966746,0.008314135,0.095405236],"study_design_scores_gemma":[0.0011387144,0.0003736588,0.94390863,0.000006040507,0.00034983826,0.000035532506,0.000011834264,0.0010932662,0.000174112,0.0024563982,0.05039974,0.000052246272],"about_ca_topic_score_codex":0.000044648168,"about_ca_topic_score_gemma":0.000012084053,"teacher_disagreement_score":0.09535299,"about_ca_system_score_codex":0.000074691714,"about_ca_system_score_gemma":0.00020925365,"threshold_uncertainty_score":0.9743955},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4406493518","doi":"10.56294/hl2023189","title":"Bibliometric study of scientific production in Scopus on psychiatric nutrition","year":2023,"lang":"en","type":"article","venue":"Health Leadership and Quality of Life","topic":"Human Health and Disease","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Production (economics); Psychiatry; Psychology; Medicine; MEDLINE; Political science; Economics","score_opus":0.4170540695267329,"score_gpt":0.47073137133217435,"score_spread":0.053677301805441446,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406493518","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.981951,0.0023482942,8.25212e-7,0.013879859,0.00033951574,0.001392663,0.00000899247,0.000041457388,0.000037412112],"genre_scores_gemma":[0.99826485,0.00024331476,0.000027719172,0.0011538267,0.00015358871,0.000033356668,0.000029688512,0.000008333925,0.00008529927],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99749774,0.0004185424,0.0008497112,0.0003405735,0.00057669525,0.0003167134],"domain_scores_gemma":[0.99876195,0.0001618608,0.00028343956,0.00027647676,0.0001163257,0.00039993686],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0038648075,0.00008965028,0.0004324317,0.023331486,0.00011969812,0.000008804708,0.000041825733,0.000054114505,0.000012521951],"category_scores_gemma":[0.0011942398,0.00008485563,0.000039478906,0.03695454,0.000101426995,0.00006231505,0.000009318033,0.00017971519,0.00001503366],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021516809,0.0068444354,0.9309267,0.03348967,0.000013329348,0.0000020006717,0.006723436,0.0000063251587,0.000059671867,0.0001620744,0.008543655,0.011077052],"study_design_scores_gemma":[0.003361453,0.0019617819,0.97739834,0.00032623755,0.000009060152,5.196254e-7,0.016676938,0.000012813645,0.000025371697,0.0000728038,0.000093302515,0.000061404346],"about_ca_topic_score_codex":0.00063184166,"about_ca_topic_score_gemma":0.0002657141,"teacher_disagreement_score":0.046471644,"about_ca_system_score_codex":0.000043676835,"about_ca_system_score_gemma":0.00049363635,"threshold_uncertainty_score":0.9877382},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4406708607","doi":"10.1051/e3sconf/202560805019","title":"Artificial intelligence in green management and sustainability: A bibliometric survey","year":2025,"lang":"en","type":"article","venue":"E3S Web of Conferences","topic":"Big Data and Business Intelligence","field":"Business, Management and Accounting","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Aston University; London Metropolitan University; Soonchunhyang University; University of Chinese Academy of Sciences; Chinese Academy of Sciences; Simon Fraser University; King Abdulaziz University; Islamic Azad University; RMIT University; National Cheng Kung University; Trường Đại học Duy Tân","keywords":"Sustainability; Business","score_opus":0.08828927339939126,"score_gpt":0.3283797023532198,"score_spread":0.24009042895382854,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406708607","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9000917,0.0013777007,0.01440435,0.002410704,0.0007794644,0.0009039367,0.000020168667,0.00009338648,0.079918586],"genre_scores_gemma":[0.9991312,0.00021520397,0.0000621571,0.00014651856,0.000050640305,0.000018267228,0.000020364096,0.0000047606063,0.00035089086],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9986543,0.000026361862,0.00047569192,0.00035039318,0.00024325875,0.0002500052],"domain_scores_gemma":[0.9989934,0.00017803011,0.00014430127,0.00025560142,0.00041947208,0.000009216826],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0010583604,0.00015844316,0.00026908537,0.013414313,0.00006110956,0.00023784727,0.00045104578,0.0000690638,0.00023379912],"category_scores_gemma":[0.0004355494,0.00014081287,0.00003002891,0.02736838,0.00023132328,0.00060487137,0.0004067359,0.00010608087,0.00001598953],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000048855083,0.000080658785,0.47167033,0.0005174271,0.000015170857,0.000003830284,0.000007664093,0.000007664747,0.0000049934893,0.21805172,0.00019058122,0.30940112],"study_design_scores_gemma":[0.00006752408,0.000010622903,0.854923,0.00014746435,0.000026865657,2.4415394e-7,0.0005940959,0.0047913166,0.00011904286,0.123207286,0.015921943,0.00019059336],"about_ca_topic_score_codex":0.00884267,"about_ca_topic_score_gemma":0.005101139,"teacher_disagreement_score":0.38325268,"about_ca_system_score_codex":0.000018572478,"about_ca_system_score_gemma":0.00019426324,"threshold_uncertainty_score":0.9977678},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4406712574","doi":"10.1136/bmjopen-2024-086982","title":"Diversity in the medical research ecosystem: a descriptive scientometric analysis of over 49 000 studies and 150 000 authors published in high-impact medical journals between 2007 and 2022","year":2025,"lang":"en","type":"article","venue":"BMJ Open","topic":"Diversity and Career in Medicine","field":"Social Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"National Institute of Biomedical Imaging and Bioengineering; Fulbright Portugal; Fundação Luso-Americana para o Desenvolvimento; Broad Institute","keywords":"Medicine; Diversity (politics); Impact factor; Scale (ratio); Medical journal; Bibliometrics; Descriptive statistics; Representation (politics); Family medicine; Library science; Geography; Political science","score_opus":0.3226879829593176,"score_gpt":0.5539282110974338,"score_spread":0.23124022813811623,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406712574","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9816488,0.0013237881,0.000006128444,0.010017108,0.00017908691,0.0012205542,0.0000416903,0.0000038911326,0.0055589606],"genre_scores_gemma":[0.99847895,0.0007890412,0.000008418819,0.0002641502,0.000072194576,0.000020271476,0.0000039319752,0.0000017619363,0.00036125287],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9898944,0.0032931645,0.00046371637,0.00036965957,0.0054963296,0.0004827679],"domain_scores_gemma":[0.9959949,0.0028390023,0.0001413225,0.00024159336,0.00041531996,0.0003678574],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.098842084,0.000098715806,0.0006958103,0.003847412,0.0010365786,0.00036762934,0.002106339,0.00020221774,0.0016523836],"category_scores_gemma":[0.02166878,0.00006557652,0.00006535423,0.017073601,0.0011475125,0.0006018055,0.0044494993,0.00065075455,0.0000013228346],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006894573,0.000057713496,0.93948036,0.000036592384,0.00046871725,0.00008469793,0.038505934,0.0000015029634,1.6004961e-7,0.0006000237,0.018184468,0.0025108755],"study_design_scores_gemma":[0.0008616746,0.00006750513,0.8808322,0.00038125244,0.00020370887,6.280156e-7,0.115730725,0.000047278703,4.4512512e-7,0.0010089732,0.00080071046,0.00006491243],"about_ca_topic_score_codex":0.06475435,"about_ca_topic_score_gemma":0.06555471,"teacher_disagreement_score":0.07722479,"about_ca_system_score_codex":0.00035275525,"about_ca_system_score_gemma":0.00077111635,"threshold_uncertainty_score":0.99926025},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"incentives","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4406796054","doi":"10.1007/978-981-97-4784-9_38","title":"Kalman Filters in IoT: A Bibliometric Analysis","year":2025,"lang":"en","type":"book-chapter","venue":"Lecture notes in electrical engineering","topic":"IoT and Edge/Fog Computing","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Kalman filter; Internet of Things; Computer science; Data science; Artificial intelligence; World Wide Web","score_opus":0.007486544065311275,"score_gpt":0.21743813567042766,"score_spread":0.2099515916051164,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406796054","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00027259,0.0041523636,0.9834338,0.00019922668,0.0016762095,0.00021148294,5.517201e-7,0.0002719639,0.009781806],"genre_scores_gemma":[0.734981,0.001356966,0.23406188,0.0027004187,0.007345778,0.00012456611,0.00013744972,0.00048520917,0.018806722],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9972127,0.000022354727,0.00062535057,0.0008751334,0.0004504735,0.0008139988],"domain_scores_gemma":[0.9978929,0.0011767143,0.00011400434,0.00064304523,0.000059685768,0.00011362461],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00034203904,0.000533707,0.0008904945,0.12967709,0.000042748157,0.0001623526,0.0011667535,0.0005894258,0.000007946897],"category_scores_gemma":[0.00059886,0.0005568397,0.00035877377,0.114931054,0.000014449646,0.00009115075,0.0003266663,0.0016574488,0.000011775453],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000008153803,0.00004119973,0.0007220428,0.00012728413,0.00039315724,0.00031247904,0.00016248818,0.32075825,0.00008919546,0.006706265,0.00032184727,0.67035764],"study_design_scores_gemma":[0.00023597512,0.00005583878,0.0014148769,0.00019629118,0.00008578728,0.000007755229,3.0522234e-8,0.98482215,0.00017238714,0.0032707236,0.009068547,0.00066963513],"about_ca_topic_score_codex":0.000062923486,"about_ca_topic_score_gemma":0.000019807252,"teacher_disagreement_score":0.74937195,"about_ca_system_score_codex":0.00051276514,"about_ca_system_score_gemma":0.000120608216,"threshold_uncertainty_score":0.9996883},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4406807934","doi":"10.18280/ijsdp.200111","title":"Islamic Finance for Sustainable Development: A Mixed-Methods Systematic Review and Bibliometric Analysis of SDG Alignment","year":2025,"lang":"en","type":"article","venue":"International Journal of Sustainable Development and Planning","topic":"Islamic Finance and Banking Studies","field":"Business, Management and Accounting","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Islamic finance; Islam; Sustainable development; Management science; Accounting; Regional science; Economics; Political science; Geography","score_opus":0.011649983435729141,"score_gpt":0.2949244621738237,"score_spread":0.2832744787380945,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406807934","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8954876,0.070751384,0.03060598,0.0006050561,0.00034529463,0.0009020446,0.0000016945716,0.00001801868,0.0012829265],"genre_scores_gemma":[0.9768464,0.0022936899,0.014642064,0.0010459477,0.00009544011,0.000096520125,0.000027067641,0.000015957026,0.0049369093],"study_design_codex":"systematic_review","study_design_gemma":"observational","domain_scores_codex":[0.9981211,0.000021479913,0.0010193321,0.00019311598,0.00033231918,0.0003126476],"domain_scores_gemma":[0.9965084,0.00027984515,0.0009985992,0.00007988284,0.0021284232,0.000004803779],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0025462105,0.00019455068,0.0008181829,0.022698034,0.0002404875,0.00022320321,0.00031278696,0.000050504736,0.0000045908446],"category_scores_gemma":[0.0010098817,0.00016228892,0.00010220134,0.014160848,0.00004057836,0.0007201506,0.00027906097,0.00009408194,4.0411996e-7],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037098117,0.00036243864,0.030752948,0.49900502,0.019460311,0.0009775542,0.0062580355,0.0007103418,0.00004942618,0.3797922,0.011416183,0.050844528],"study_design_scores_gemma":[0.0073995516,0.00016870903,0.464719,0.17754637,0.024166333,0.00008394828,0.10342118,0.003920843,0.0019102453,0.04321899,0.17071688,0.0027279642],"about_ca_topic_score_codex":0.000022849192,"about_ca_topic_score_gemma":0.0000023089667,"teacher_disagreement_score":0.43396604,"about_ca_system_score_codex":0.0001522122,"about_ca_system_score_gemma":0.0001852343,"threshold_uncertainty_score":0.9883788},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4406827606","doi":"10.1177/17531934241313205","title":"The 100 most-cited articles in WALANT hand surgery: A bibliometric and visualized analysis","year":2025,"lang":"en","type":"article","venue":"Journal of Hand Surgery (European Volume)","topic":"Orthopedic Surgery and Rehabilitation","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Tourniquet; Bibliometrics; Randomized controlled trial; Hand surgery; Popularity; MEDLINE; Physical therapy; General surgery; Surgery; Library science","score_opus":0.019754986343015812,"score_gpt":0.29347256998293586,"score_spread":0.27371758363992005,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406827606","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98867613,0.007961919,0.00026019203,0.0019551106,0.00050314557,0.00013606992,0.0000033922759,0.000014265683,0.0004897567],"genre_scores_gemma":[0.9934183,0.0037543396,0.000040306208,0.0002912576,0.00014166866,0.000002464134,0.000008067892,0.000020516767,0.002323105],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99633515,0.0010754683,0.0014651592,0.00023928065,0.0005206089,0.0003643232],"domain_scores_gemma":[0.99293405,0.0055861846,0.000518739,0.00027219366,0.00046852592,0.00022031715],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.008478303,0.00020019882,0.0009885429,0.03228691,0.00033387993,0.00028486596,0.00008709378,0.00006870903,0.000064011096],"category_scores_gemma":[0.005147806,0.0001253425,0.0005685894,0.04809798,0.00031372963,0.00022670045,0.000052508487,0.0003271416,0.000007261551],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00076548447,0.00023724703,0.866072,0.000064806925,0.0006917612,0.00047067757,0.00025504848,0.000018842746,0.0014387928,0.0000068874856,0.0080534285,0.12192502],"study_design_scores_gemma":[0.0007931646,0.000097067554,0.96111614,0.0003561568,0.000677286,0.00012432595,0.00031187793,0.0007682989,0.00051352975,0.000038379174,0.03506566,0.00013810911],"about_ca_topic_score_codex":0.000046848196,"about_ca_topic_score_gemma":0.000028716531,"teacher_disagreement_score":0.12178691,"about_ca_system_score_codex":0.00005737167,"about_ca_system_score_gemma":0.00027853792,"threshold_uncertainty_score":0.97868127},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4406969884","doi":"10.2196/preprints.63775","title":"Evolution of Artificial Intelligence in Medical Education From 2000 to 2024: Bibliometric Analysis (Preprint)","year":2024,"lang":"en","type":"preprint","venue":"","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Preprint; Computer science; Artificial intelligence; Library science; World Wide Web","score_opus":0.11189229394356713,"score_gpt":0.4565599439399822,"score_spread":0.3446676499964151,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406969884","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9415129,0.0035135536,0.038852762,0.007920981,0.00403794,0.0012111168,0.000034404442,0.00008114522,0.002835211],"genre_scores_gemma":[0.9934206,0.0004673782,0.003646342,0.00034904745,0.00093566487,0.00021458912,0.00020538086,0.000026389243,0.0007346227],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9955918,0.00013049347,0.0018480557,0.0010005417,0.0010899409,0.0003391239],"domain_scores_gemma":[0.9973164,0.00034739077,0.00022573429,0.0009016886,0.0007094352,0.00049934734],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0013544608,0.00028237407,0.0007880488,0.09707673,0.000029860492,0.00006008092,0.00035013838,0.000741189,0.009374424],"category_scores_gemma":[0.0028772599,0.0002696347,0.00035010776,0.12294444,0.00008087213,0.00004913487,0.0005216965,0.001236034,0.00087160483],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018304295,0.001140901,0.07331264,0.0005809483,0.00036155494,0.000011505743,0.0030362678,0.0032851086,0.00013870624,0.0025241014,0.002388278,0.91303694],"study_design_scores_gemma":[0.00003311384,0.00047062698,0.318649,0.0052368618,0.0030159054,0.000012719363,0.014941981,0.24562831,0.019011633,0.39110667,0.00077699445,0.001116179],"about_ca_topic_score_codex":0.09434645,"about_ca_topic_score_gemma":0.010447712,"teacher_disagreement_score":0.9119208,"about_ca_system_score_codex":0.0015028005,"about_ca_system_score_gemma":0.009307921,"threshold_uncertainty_score":0.99997556},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4406970010","doi":"10.2196/63775","title":"Evolution of Artificial Intelligence in Medical Education From 2000 to 2024: Bibliometric Analysis","year":2025,"lang":"en","type":"article","venue":"Interactive Journal of Medical Research","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Science Citation Index; Productivity; Medical education; Computer science; Library science; Psychology; Artificial intelligence; Medicine","score_opus":0.20671002867086027,"score_gpt":0.596326466243808,"score_spread":0.3896164375729477,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406970010","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9199029,0.0020334453,0.03684267,0.038230248,0.001535847,0.00028968873,0.0000034372508,0.0000036898666,0.001158031],"genre_scores_gemma":[0.997308,0.00079073565,0.000518827,0.000508042,0.0006328915,0.000017047472,0.000005364037,0.000006811849,0.00021226735],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.992873,0.0006572025,0.0016800712,0.00026580656,0.0041681468,0.00035574404],"domain_scores_gemma":[0.98973614,0.0047762427,0.00029171925,0.0003004408,0.0040218346,0.0008736226],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.008442562,0.000107863576,0.00056293316,0.08464905,0.000054522872,0.000034516062,0.0005411737,0.0003234994,0.006456517],"category_scores_gemma":[0.06488896,0.00008875186,0.00020033024,0.11192091,0.00022183034,0.0001895843,0.00012409744,0.0019394817,0.000065926964],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014069636,0.0017334429,0.053714745,0.00006045372,0.0002659268,0.000057316294,0.0015877552,0.000039288127,0.00037464304,0.000815753,0.00508555,0.93485814],"study_design_scores_gemma":[0.00040605693,0.0039727804,0.71038455,0.015084951,0.0008231859,0.00012491556,0.07478338,0.061477546,0.04659937,0.07324188,0.012643588,0.0004577956],"about_ca_topic_score_codex":0.012620296,"about_ca_topic_score_gemma":0.0019687642,"teacher_disagreement_score":0.9344004,"about_ca_system_score_codex":0.0015670113,"about_ca_system_score_gemma":0.015225165,"threshold_uncertainty_score":0.9944517},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4406984385","doi":"10.3138/jsp-2024-0044","title":"A New Family of Rankings of Universities","year":2025,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"Higher Education Governance and Development","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Business; Geography; Regional science","score_opus":0.01797988313408163,"score_gpt":0.29958796678838506,"score_spread":0.28160808365430345,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406984385","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8993824,0.0018959467,0.0025960328,0.01779056,0.0013839279,0.000055648416,0.000005852538,0.000007540533,0.07688207],"genre_scores_gemma":[0.9714111,0.00023882168,0.0104594845,0.0006241519,0.00020682698,1.5475155e-7,3.1846426e-7,0.0000032046692,0.017055944],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99850506,0.00008564107,0.00040839388,0.000058498816,0.00082683854,0.00011559537],"domain_scores_gemma":[0.9974334,0.00012961314,0.00065057195,0.00007105678,0.0016304497,0.0000849019],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0022719766,0.0000450711,0.00018038247,0.00034329327,0.00009786788,0.0009837126,0.0005595532,0.00008449995,0.00007169384],"category_scores_gemma":[0.000749716,0.00004468059,0.00008523212,0.0006735883,0.00005971008,0.022163803,0.000044693526,0.00034044246,9.628963e-7],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011626778,0.000120670025,0.09516278,0.00006463709,0.0002453989,0.000006374099,0.12606624,0.000024894267,0.0027885018,0.20853494,0.52977026,0.037099026],"study_design_scores_gemma":[0.0008593641,0.00004047998,0.21544938,0.00050426536,0.00004185883,0.0000011109146,0.11019087,3.8083954e-7,0.00057195977,0.0101550715,0.66209775,0.000087546156],"about_ca_topic_score_codex":0.0034207685,"about_ca_topic_score_gemma":0.00007284097,"teacher_disagreement_score":0.19837986,"about_ca_system_score_codex":0.00023861407,"about_ca_system_score_gemma":0.0051444457,"threshold_uncertainty_score":0.9915127},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4407106325","doi":"10.31521/modecon.v46(2024)-11","title":"Activities of Non-governmental Organisations in the Context of Achieving the UN Sustainable Development Goals: a Bibliometric Analysis","year":2024,"lang":"en","type":"article","venue":"Modern Economics","topic":"International Development and Aid","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Context (archaeology); Sustainable development; Regional science; Process management; Business; Knowledge management; Political science; Computer science; Geography","score_opus":0.015725738996607556,"score_gpt":0.26809348726090154,"score_spread":0.252367748264294,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407106325","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98447084,0.00014292829,0.0009152995,0.0012698538,0.000058889353,0.0001421011,0.000010251184,0.000005805777,0.012984007],"genre_scores_gemma":[0.99852574,0.00015927572,0.00015391997,0.000081072634,0.000022302273,0.000015502399,0.0000057238017,0.0000048940756,0.0010315869],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9992819,0.00005019361,0.0002484676,0.000103598715,0.00017455714,0.00014129824],"domain_scores_gemma":[0.9994248,0.00035950856,0.00009026178,0.00007805335,0.000034866356,0.00001252338],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010171714,0.000059294052,0.00012330312,0.0041286275,0.00018184786,0.00011611554,0.00030440523,0.000029099689,0.00008870218],"category_scores_gemma":[0.000051539646,0.000042244938,0.0000663973,0.009558623,0.000095160845,0.0002526076,0.000061412095,0.00006359588,0.000003499121],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024036448,0.00019570036,0.122383416,0.00007431193,0.0013367353,0.000005108754,0.44608048,0.0010553848,0.00023786574,0.37959415,0.00027078798,0.04874203],"study_design_scores_gemma":[0.00031314083,0.000028494887,0.42733815,0.000043011136,0.00018434292,8.436883e-7,0.5247676,0.014317106,0.0044963825,0.009541023,0.018670836,0.00029908342],"about_ca_topic_score_codex":0.001353098,"about_ca_topic_score_gemma":0.0021593655,"teacher_disagreement_score":0.3700531,"about_ca_system_score_codex":0.0004883118,"about_ca_system_score_gemma":0.00034252767,"threshold_uncertainty_score":0.45926014},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4407608597","doi":"10.22452/mojem.vol12no4.1","title":"A BIBLIOMETRIC ANALYSIS OF TECHNOLOGICAL MANAGEMENT TRENDS IN PRESCHOOL EDUCATION AND META-ANALYSIS OF THE UTILIZATION OF TECHNOLOGICAL TOOLS IN CLASSROOM","year":2024,"lang":"en","type":"article","venue":"Malaysian Online Journal of Educational Management","topic":"Gender and Technology in Education","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Education and Early Childhood Development","funders":"","keywords":"Meta-analysis; Knowledge management; Computer science; Mathematics education; Psychology; Medicine","score_opus":0.0882114477981382,"score_gpt":0.3997086856058982,"score_spread":0.31149723780775995,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407608597","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9647726,0.004729614,0.00046687797,0.023495248,0.0001496644,0.0004271413,0.000045653625,0.000014926721,0.0058983215],"genre_scores_gemma":[0.9948452,0.0010477083,0.0033247927,0.00003175464,0.00001312811,0.000042623065,0.000033220444,0.000005469207,0.0006560649],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9975462,0.00022667577,0.0011108313,0.00027335368,0.0006637139,0.00017921424],"domain_scores_gemma":[0.99852705,0.00024864238,0.0006270997,0.0003246545,0.00023201095,0.000040530675],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0016640249,0.00014181582,0.0007051138,0.09288389,0.000046153786,0.00003800059,0.0005999273,0.00016898355,0.0005082732],"category_scores_gemma":[0.0003091619,0.00010262882,0.00054471055,0.16248725,0.00036771005,0.00021177676,0.00013586701,0.00025917424,3.8925992e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000025094936,0.004611289,0.23675294,0.00019866186,0.051963456,0.000003150584,0.00090404804,0.0043421565,0.000047833,0.5425621,0.00030201796,0.15828723],"study_design_scores_gemma":[0.00010776566,0.000040746047,0.9221186,0.000030143243,0.05561164,0.0000012005322,0.00917187,0.0004463787,0.00003258804,0.0112576885,0.0010922919,0.0000890793],"about_ca_topic_score_codex":0.0000912725,"about_ca_topic_score_gemma":0.00036534504,"teacher_disagreement_score":0.6853657,"about_ca_system_score_codex":0.00020902194,"about_ca_system_score_gemma":0.00016172261,"threshold_uncertainty_score":0.9173975},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4407638392","doi":"10.1097/md.0000000000041537","title":"Progress and prospects of Parkinson’s disease with depression research: A global bibliometric analysis based on CiteSpace","year":2025,"lang":"en","type":"article","venue":"Medicine","topic":"Parkinson's Disease Mechanisms and Treatments","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Depression (economics); Parkinson's disease; Dementia; Meta-analysis; Bibliometrics; Confidence interval; Disease; Quality of life (healthcare); Gerontology; Psychiatry; Library science; Pathology; Internal medicine","score_opus":0.034253065474591005,"score_gpt":0.3633214366199554,"score_spread":0.3290683711453644,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407638392","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9819941,0.009417393,0.00082024693,0.0043643564,0.000045296663,0.0010183756,0.000045401735,0.000058735444,0.0022361148],"genre_scores_gemma":[0.99853337,0.00013725588,0.00058998924,0.00024285229,0.00003796979,0.00012773085,0.000053515523,0.000011640279,0.00026565662],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9976626,0.00011027729,0.0002227052,0.0005566478,0.00111693,0.0003308331],"domain_scores_gemma":[0.99836236,0.00016541625,0.00009375698,0.0005510485,0.00032860978,0.00049881963],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00050406874,0.00020406808,0.0005457516,0.026429672,0.000106456726,0.000015491669,0.00009576172,0.000058116395,0.00007280027],"category_scores_gemma":[0.00044694706,0.00012130564,0.00007347826,0.09703956,0.00033576164,0.00003863649,0.00005195138,0.00013881951,0.000002165079],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0032784466,0.00086641655,0.98068404,0.00036005594,0.00055746513,0.0004141892,0.000022288383,0.0000135029095,0.00004605382,0.00026727404,0.00066111435,0.0128291715],"study_design_scores_gemma":[0.005867617,0.0013535853,0.9821069,0.0013476413,0.002315586,0.0000023684916,0.00004413899,0.0034816775,0.00032103338,0.00021024322,0.0028619631,0.000087238834],"about_ca_topic_score_codex":0.00010824541,"about_ca_topic_score_gemma":0.000024944035,"teacher_disagreement_score":0.07060988,"about_ca_system_score_codex":0.00013258378,"about_ca_system_score_gemma":0.0002426739,"threshold_uncertainty_score":0.9846049},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4407706881","doi":"10.5539/hes.v15n2p1","title":"A Bibliometric Analysis of Artificial Intelligence for Multimedia in Education by Dimensions AI","year":2025,"lang":"en","type":"article","venue":"Higher Education Studies","topic":"Education and Learning Interventions","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Higher education; Multimedia; Computer science; Mathematics education; Computer-Assisted Instruction; Trend analysis; Statistical analysis; Psychology; Statistics; Machine learning; Mathematics","score_opus":0.07720784980722495,"score_gpt":0.44503222501644674,"score_spread":0.3678243752092218,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407706881","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.48612776,0.049706366,0.39303893,0.049391825,0.018644862,0.0013256543,0.000029690565,0.00021074757,0.0015241584],"genre_scores_gemma":[0.97120756,0.00023252836,0.019101247,0.0011271029,0.00004258229,0.000486184,0.000037910304,0.0000045577153,0.0077603017],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9988521,0.00008394969,0.0004891337,0.00030041416,0.00013707276,0.00013733281],"domain_scores_gemma":[0.997991,0.0005368773,0.00016787661,0.00032682513,0.0009396386,0.000037749734],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00035670682,0.0000991992,0.00024258533,0.03948975,0.00010988729,0.000057791185,0.0002932475,0.00004212546,0.000051608706],"category_scores_gemma":[0.00029394435,0.00009795624,0.00013295701,0.10598296,0.000060287977,0.0001989751,0.00006556933,0.000084681014,0.000010439361],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011244051,0.004668577,0.075593345,0.00016210332,0.001122731,3.962356e-8,0.0042283917,0.00041373193,0.0004002275,0.4505983,0.21731788,0.24548343],"study_design_scores_gemma":[0.00013734658,0.00009228698,0.80756426,0.0002795238,0.00061207335,2.302895e-7,0.007948852,0.008604079,0.0011330433,0.021850673,0.15141891,0.00035875046],"about_ca_topic_score_codex":0.00013216549,"about_ca_topic_score_gemma":0.000043411466,"teacher_disagreement_score":0.7319709,"about_ca_system_score_codex":0.000111926216,"about_ca_system_score_gemma":0.00047069008,"threshold_uncertainty_score":0.9713968},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4407760747","doi":"10.5539/hes.v15n2p54","title":"A Bibliometric Analysis of Digital Literacy in Remote Learning","year":2025,"lang":"en","type":"article","venue":"Higher Education Studies","topic":"Technology-Enhanced Education Studies","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Mathematics education; Literacy; Higher education; Trend analysis; Technological literacy; Computer science; Educational technology; Psychology; Pedagogy; Political science","score_opus":0.049042702434567716,"score_gpt":0.4588463093699769,"score_spread":0.4098036069354092,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407760747","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9451006,0.01904963,0.0000610384,0.0048871296,0.0011368701,0.00012622029,0.0000019990985,0.00018873268,0.02944779],"genre_scores_gemma":[0.94041175,0.0021945927,0.0003961862,0.00009917605,0.00004600506,0.00003336954,0.0000042206298,0.0000037651046,0.056810956],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99890685,0.00007948135,0.00035474947,0.00024051951,0.00021709618,0.00020127816],"domain_scores_gemma":[0.9982342,0.00056738436,0.00017882888,0.00017508093,0.0008208418,0.000023664068],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00036271827,0.00009537737,0.00035283776,0.06971964,0.0002563662,0.000061390034,0.00017168127,0.000065356064,0.000061280865],"category_scores_gemma":[0.00089233153,0.000095723364,0.00008294902,0.25399125,0.0003252574,0.00028735137,0.00008556828,0.00012719829,0.000013818778],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000067643236,0.0001875602,0.800019,0.000038152793,0.0010654855,1.7277802e-7,0.024608297,0.000034465527,0.000020789359,0.022530708,0.007892691,0.14359589],"study_design_scores_gemma":[0.00006319375,0.0000071751624,0.7726919,0.000045071632,0.00016158444,1.7300007e-8,0.03494028,0.0000023495945,0.000018045816,0.0027480405,0.18924637,0.00007595098],"about_ca_topic_score_codex":0.00056182325,"about_ca_topic_score_gemma":0.00017065404,"teacher_disagreement_score":0.1842716,"about_ca_system_score_codex":0.0003084683,"about_ca_system_score_gemma":0.00034598864,"threshold_uncertainty_score":0.9408243},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4407782959","doi":"10.3390/buildings15050652","title":"Scientometric Analysis on Climate Resilient Retrofit of Residential Buildings","year":2025,"lang":"en","type":"article","venue":"Buildings","topic":"Building Energy and Comfort Optimization","field":"Engineering","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada","funders":"","keywords":"Architectural engineering; Engineering; Environmental science; Civil engineering; Geography","score_opus":0.005627997182172286,"score_gpt":0.23659986712902137,"score_spread":0.2309718699468491,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407782959","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9464331,0.00033526306,0.04394282,0.000075598364,0.0005143319,0.0000995161,0.0000103275,0.00030862621,0.008280411],"genre_scores_gemma":[0.99409664,0.00020338643,0.005101083,0.000068558715,0.00003157363,0.000011854956,0.000013624708,0.000020623273,0.00045265755],"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9985706,0.000019008467,0.0004012994,0.00031446762,0.00035322475,0.00034137155],"domain_scores_gemma":[0.99929273,0.00007353367,0.0000855751,0.00039650634,0.0000923542,0.00005927304],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004055183,0.00017821687,0.00032238726,0.0038488356,0.00011734832,0.000060421808,0.0003305841,0.00012556787,0.00006504804],"category_scores_gemma":[0.000092129376,0.0001853355,0.00019827463,0.01039248,0.000054244265,0.00012792538,0.00008639102,0.00015293165,0.0000036969664],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000042963016,0.000036587786,0.014687467,0.0000854919,0.00033776995,0.0000019119907,0.000037835496,0.93653727,0.0064399717,0.036746558,0.002627777,0.002418389],"study_design_scores_gemma":[0.0021711597,0.00025445278,0.13147914,0.0006717481,0.0022694622,0.0000039043252,0.00010500513,0.4093851,0.42802763,0.0021669532,0.022067787,0.0013976707],"about_ca_topic_score_codex":0.000049351504,"about_ca_topic_score_gemma":0.000007246521,"teacher_disagreement_score":0.5271522,"about_ca_system_score_codex":0.00012210256,"about_ca_system_score_gemma":0.000018183522,"threshold_uncertainty_score":0.7557762},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4407865394","doi":"10.23917/iseth.3845","title":"Bibliometric Investigation of Development and Research Trends in Transformative Education: A Comprehensive Analysis on Scopus Database (1983-2023)","year":2024,"lang":"en","type":"article","venue":"Proceeding ISETH (International Summit on Science Technology and Humanity)","topic":"Ideological and Political Education","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Transformative learning; Database; Data science; Political science; Sociology; Computer science; MEDLINE; Pedagogy","score_opus":0.1604098254020699,"score_gpt":0.4562320449590845,"score_spread":0.2958222195570146,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407865394","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96714807,0.00023370633,0.00002922721,0.004822633,0.00017888294,0.00012597525,0.00000774707,0.000039061153,0.027414706],"genre_scores_gemma":[0.9987477,0.000092973605,0.00035419952,0.00009665989,0.00004812162,0.000045643592,0.000013741994,0.000002292757,0.0005986396],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99844617,0.000043649874,0.00024016234,0.0003582436,0.0006596773,0.00025210466],"domain_scores_gemma":[0.99901634,0.00026931308,0.00004760158,0.00006036914,0.0005111195,0.000095277675],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0016471505,0.00008284219,0.00013051325,0.082061544,0.00053332566,0.00017831085,0.00031209117,0.00008584447,0.00007829594],"category_scores_gemma":[0.0005190849,0.00007205982,0.00001951302,0.080329165,0.0022654762,0.00048997276,0.00006570726,0.00033958536,0.000010086773],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010492024,0.000077456636,0.036386825,0.000027412914,0.000023013396,7.998619e-7,0.0061418493,0.0000020866757,0.00021696021,0.91614634,0.000065202614,0.04090157],"study_design_scores_gemma":[0.0004734429,0.00055239926,0.6490539,0.00087405473,0.00009060358,0.0000027384185,0.051118392,0.001151121,0.014711687,0.24217284,0.0392271,0.00057175866],"about_ca_topic_score_codex":0.00026081028,"about_ca_topic_score_gemma":0.00027238778,"teacher_disagreement_score":0.6739735,"about_ca_system_score_codex":0.00019929098,"about_ca_system_score_gemma":0.00049936643,"threshold_uncertainty_score":0.9392189},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4407924408","doi":"10.15390/eb.2024.12832","title":"Educational Research in Türkiye: A Bibliometric Analysis of the Global Literature","year":2024,"lang":"en","type":"article","venue":"TED EĞİTİM VE BİLİM","topic":"Educational Methods and Analysis","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Universidad Nacional de Luján","keywords":"Humanities; Philosophy","score_opus":0.15925967989527454,"score_gpt":0.5556654527475771,"score_spread":0.3964057728523025,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407924408","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.89722836,0.015429291,0.000119195436,0.037879113,0.0010661935,0.00036234502,0.00026313294,0.000047533587,0.04760485],"genre_scores_gemma":[0.9942122,0.00054580014,0.0008493175,0.00013783493,0.00051410473,0.000041004758,0.000050276274,0.000008994617,0.0036404668],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9960401,0.0012533532,0.00040125469,0.0004445629,0.0014200748,0.00044064698],"domain_scores_gemma":[0.9967514,0.0019931332,0.000087447814,0.0003853187,0.0006380767,0.00014460346],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0043186476,0.00013071622,0.00029631663,0.036503524,0.0003795575,0.00038206042,0.00069886824,0.00018086005,0.0009250976],"category_scores_gemma":[0.0022327239,0.00009661621,0.00036607316,0.5000937,0.00039974813,0.00022587829,0.00010855797,0.00043573,0.000051557967],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019467381,0.00053314207,0.6064583,0.0001073722,0.001154534,0.000007361417,0.010875033,0.00017937532,0.00024796947,0.34027725,0.010472479,0.029667724],"study_design_scores_gemma":[0.000117676165,0.000028276381,0.890568,0.00030044332,0.0005243284,0.0000015475141,0.0078052958,0.0015479629,0.000030241285,0.059919666,0.038913976,0.00024258572],"about_ca_topic_score_codex":0.0050191907,"about_ca_topic_score_gemma":0.0025280572,"teacher_disagreement_score":0.46359015,"about_ca_system_score_codex":0.00047977272,"about_ca_system_score_gemma":0.0014495277,"threshold_uncertainty_score":0.9999882},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4407930396","doi":"10.5539/ells.v15n1p16","title":"Knowledge Mapping of News Translation Studies: A Bibliometric Analysis","year":2025,"lang":"en","type":"article","venue":"English Language and Literature Studies","topic":"Translation Studies and Practices","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Translation (biology); Computer science; Data science; Information retrieval","score_opus":0.052622419902625,"score_gpt":0.34139142931554284,"score_spread":0.28876900941291783,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407930396","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13997026,0.8022055,0.000049540038,0.00034975365,0.0006525048,0.0001461662,0.000064083346,0.00006490243,0.056497317],"genre_scores_gemma":[0.9637211,0.032721512,0.00021278117,0.0001052412,0.00047128613,0.000023197505,0.000023330611,0.00000698948,0.0027145753],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9990464,0.00009130421,0.0003555705,0.00023912093,0.000117253985,0.00015035151],"domain_scores_gemma":[0.9983162,0.0008147096,0.00012323662,0.00014593874,0.0005785191,0.000021345802],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00029728533,0.00017694086,0.0005297899,0.013956214,0.00032217448,0.0001568644,0.00007074287,0.000043386346,0.00004140847],"category_scores_gemma":[0.00021610623,0.00012857145,0.0001575183,0.020095715,0.00019227597,0.00037803844,0.000042294956,0.00014105337,9.453747e-7],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021895214,0.00003659187,0.00094250747,0.00046883657,0.0047073164,0.000004173471,0.8871449,0.000002308793,0.000009614582,0.007187443,0.0023007027,0.09717368],"study_design_scores_gemma":[0.0003462557,0.00003059187,0.00047926724,0.0002124207,0.0007923412,1.5333524e-7,0.5162751,0.000009949799,0.000015916556,0.00019042507,0.48152474,0.00012281013],"about_ca_topic_score_codex":0.000034935027,"about_ca_topic_score_gemma":0.0009808675,"teacher_disagreement_score":0.82375085,"about_ca_system_score_codex":0.000009894292,"about_ca_system_score_gemma":0.000009673654,"threshold_uncertainty_score":0.99721974},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4407935713","doi":"10.19173/irrodl.v26i1.7953","title":"Bibliometric Insights Into the Open Education Landscape","year":2025,"lang":"en","type":"article","venue":"The International Review of Research in Open and Distributed Learning","topic":"Open Education and E-Learning","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Open education; Distance education; Higher education; Data science; Geography; Computer science; Sociology; World Wide Web; Pedagogy; Political science","score_opus":0.0671515032365844,"score_gpt":0.46146022919374263,"score_spread":0.3943087259571582,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407935713","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.041721366,0.106397234,0.004360729,0.22051018,0.0010659499,0.0032391376,0.000003174535,0.000040305094,0.62266195],"genre_scores_gemma":[0.94679254,0.048200976,0.0004562695,0.0011042326,0.00003522327,0.00018126758,0.000053594078,0.0000041902617,0.0031716835],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976952,0.0009313518,0.0003514834,0.000284108,0.00055895624,0.00017891715],"domain_scores_gemma":[0.9975426,0.0012500352,0.00014687485,0.00037692138,0.0006294935,0.00005406774],"candidate_categories":["bibliometrics","scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.0066763232,0.00008329768,0.00018214982,0.0029249408,0.00045235784,0.0015138438,0.0057097003,0.000029050489,0.000120166944],"category_scores_gemma":[0.0035816133,0.00004907637,0.00003088095,0.02344408,0.0001039159,0.0007702126,0.0040318505,0.0006664048,0.00002741363],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000028564165,0.00022647543,0.029550487,0.00050017383,0.000044672113,0.000001640353,0.0010336931,0.000084393105,0.000028578479,0.5325326,0.03547949,0.40048927],"study_design_scores_gemma":[0.00045587856,0.00006182705,0.058298066,0.005803462,0.0000064603464,0.000006018622,0.008818882,0.005237357,0.000063473846,0.013366728,0.90776485,0.00011698715],"about_ca_topic_score_codex":0.0011928416,"about_ca_topic_score_gemma":0.000038552862,"teacher_disagreement_score":0.9050712,"about_ca_system_score_codex":0.00009074391,"about_ca_system_score_gemma":0.0008943188,"threshold_uncertainty_score":0.9996699},"labels":[{"model":"gemma","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4407946995","doi":"10.1016/j.jclepro.2025.145163","title":"Bibliometric analysis on hotspots and trends of cell immobilization technology in bioethanol production (1994–2024)","year":2025,"lang":"en","type":"article","venue":"Journal of Cleaner Production","topic":"Biofuel production and bioconversion","field":"Engineering","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Natural Science Foundation of China","keywords":"Biofuel; Production (economics); Environmental science; Pulp and paper industry; Agricultural economics; Engineering; Waste management; Economics","score_opus":0.009879916685317847,"score_gpt":0.24540165718580778,"score_spread":0.23552174050048993,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407946995","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9942812,0.0011842456,0.00024027609,0.0015766233,0.0022132378,0.000109894754,0.0000011137124,0.000043489363,0.0003499183],"genre_scores_gemma":[0.99654484,0.0018656579,0.00030072455,0.00000861946,0.00016710602,0.0000023594082,0.0000026711782,0.000010396719,0.0010976292],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.998807,0.000045636614,0.00055310054,0.0002443972,0.00021934598,0.00013051221],"domain_scores_gemma":[0.9993047,0.00001189584,0.00024381338,0.00021816284,0.00019202135,0.000029376415],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0005869421,0.00013301868,0.0003044116,0.11947539,0.00003364607,0.000015585156,0.00009013778,0.00016073325,0.000024640483],"category_scores_gemma":[0.00011718172,0.00011788799,0.00008460979,0.10178912,0.00007446487,0.00026187263,0.000017721592,0.00030082205,0.0000012385462],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030327364,0.0004541571,0.0381457,0.0007724556,0.0003540918,0.0000039027777,0.0002161905,0.070999704,0.57386017,0.000026883861,0.0039563747,0.31090707],"study_design_scores_gemma":[0.00054444093,0.00026545802,0.14555115,0.00010243701,0.00038017577,0.00003329496,0.00045969005,0.0010915484,0.84827,0.00025162453,0.0028831942,0.00016700276],"about_ca_topic_score_codex":0.0000064224046,"about_ca_topic_score_gemma":0.0000047511953,"teacher_disagreement_score":0.31074008,"about_ca_system_score_codex":0.00010596714,"about_ca_system_score_gemma":0.000016427653,"threshold_uncertainty_score":0.9173028},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4407979623","doi":"10.12688/f1000research.159117.1","title":"Women's studies in the Latin American context: a bibliometric approach","year":2025,"lang":"en","type":"preprint","venue":"F1000Research","topic":"Gender and Feminist Studies","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Open peer review; Plant biology; Context (archaeology); Latin Americans; Physiology; Neuroscience; Medicine; Biology; Political science; Botany","score_opus":0.36346604103678587,"score_gpt":0.49989369224021624,"score_spread":0.13642765120343037,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407979623","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24475078,0.01886865,0.000101581325,0.01117711,0.00094419654,0.003619102,0.00015752671,0.00019794863,0.7201831],"genre_scores_gemma":[0.9744375,0.009142984,0.00018468566,0.00045086938,0.00029931165,0.0016985656,0.0000090558515,0.000012368713,0.01376467],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9943742,0.0017730377,0.00037272985,0.00063212594,0.0017972686,0.0010506159],"domain_scores_gemma":[0.996284,0.0024737474,0.00011933862,0.00058831886,0.0004252074,0.00010937485],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.007271936,0.0002505589,0.0006378125,0.015284323,0.000879719,0.00028414928,0.0018751341,0.00015034746,0.000056145735],"category_scores_gemma":[0.0033557606,0.00017931707,0.0001316271,0.054399073,0.0025785326,0.000047152385,0.0020661973,0.0013999417,0.000040523035],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005557976,0.00038470048,0.060707014,0.0005760429,0.00037632897,0.00003226397,0.8136911,0.000027473126,7.634722e-7,0.01122806,0.050332494,0.062588185],"study_design_scores_gemma":[0.00021137472,0.00006696493,0.0610922,0.00007041803,0.000014528392,2.2566991e-7,0.8756411,0.000029358282,0.0000011331219,0.00812764,0.054507464,0.0002375989],"about_ca_topic_score_codex":0.0076522157,"about_ca_topic_score_gemma":0.0017625949,"teacher_disagreement_score":0.7296867,"about_ca_system_score_codex":0.00055907865,"about_ca_system_score_gemma":0.00089342205,"threshold_uncertainty_score":0.9989559},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4408034673","doi":"10.21037/jtd-24-1451","title":"Emerging trends and hotspots in animal experimental research on lung transplantation from 2004 to 2023: a bibliometric analysis","year":2025,"lang":"en","type":"article","venue":"Journal of Thoracic Disease","topic":"Transplantation: Methods and Outcomes","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Lung transplantation; Transplantation; Data science; Internal medicine; Computer science","score_opus":0.05728539956910273,"score_gpt":0.5055152807955822,"score_spread":0.44822988122647944,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408034673","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9904392,0.0050266273,0.0017732085,0.0021597834,0.00009982446,0.000112281676,0.00005907497,0.0000065739996,0.00032346838],"genre_scores_gemma":[0.99594516,0.00045934622,0.0029939078,0.00021920845,0.00007912126,0.0000070268384,0.000018955105,0.000008400905,0.0002688989],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99825686,0.00025830138,0.00044444698,0.00021139716,0.00062363484,0.00020537044],"domain_scores_gemma":[0.998942,0.00035261482,0.00008554394,0.00013302453,0.00012722958,0.00035955664],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0009641256,0.00011767188,0.00037748838,0.07781382,0.00006696853,0.00005876263,0.00009460668,0.000044665645,0.00016076208],"category_scores_gemma":[0.00012372743,0.0000967909,0.00018029407,0.063401215,0.000027652764,0.00016606883,0.000011563518,0.0002694328,0.000002074565],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.02324664,0.0007015785,0.89487576,0.00039381912,0.00075476075,0.001967188,0.0012415929,0.00048428774,0.00822617,0.00006781582,0.00051908667,0.067521274],"study_design_scores_gemma":[0.0010938828,0.00028119827,0.9936477,0.00062876206,0.001189071,0.000006635076,0.00045404697,0.00045818824,0.0020769094,0.000047580237,0.00003588789,0.00008014691],"about_ca_topic_score_codex":0.00008644165,"about_ca_topic_score_gemma":0.000021316804,"teacher_disagreement_score":0.09877191,"about_ca_system_score_codex":0.000118057236,"about_ca_system_score_gemma":0.000115820905,"threshold_uncertainty_score":0.95650667},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4408069007","doi":"10.1016/j.jwpe.2025.107327","title":"Chitosan-based materials for emerging contaminants removal: Bibliometric analysis, research progress, and directions","year":2025,"lang":"en","type":"article","venue":"Journal of Water Process Engineering","topic":"Adsorption and biosorption for pollutant removal","field":"Environmental Science","cited_by":28,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"European Regional Development Fund; Conseil régional de Bourgogne-Franche-Comté; National Natural Science Foundation of China","keywords":"Chitosan; Nanotechnology; Contamination; Environmental chemistry; Chemistry; Environmental science; Materials science; Biochemical engineering; Computer science; Engineering; Organic chemistry; Biology","score_opus":0.018064259390317965,"score_gpt":0.319467918902488,"score_spread":0.30140365951217,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408069007","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9919599,0.00031243465,0.00674135,0.00046702207,0.00021851088,0.000168974,0.000007211136,0.000024277508,0.00010031621],"genre_scores_gemma":[0.99791634,0.000083319566,0.0017281037,0.00001889185,0.000056254565,0.000011389803,0.0000026232085,0.000010201448,0.00017290251],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9987004,0.000029707728,0.0003899118,0.0001587482,0.0004338848,0.0002873668],"domain_scores_gemma":[0.9995989,0.00004791144,0.000094793555,0.00008446542,0.000091873626,0.00008203074],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0011732278,0.00010934478,0.00024258113,0.013805755,0.00013978458,0.00010772866,0.00015843032,0.000059891794,0.000088157976],"category_scores_gemma":[0.00010490927,0.00007507822,0.0000850387,0.019130483,0.00006755208,0.00021870248,0.000036984966,0.00013442939,0.000002576519],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018031492,0.00015647207,0.020939881,0.00034181945,0.0003764105,0.000027208547,0.0003375316,0.013725961,0.9470771,0.00005153282,0.00025440863,0.016531367],"study_design_scores_gemma":[0.0010810513,0.00015374881,0.070363455,0.00015420711,0.00026800705,0.00007322416,0.000097442404,0.01732543,0.8907536,0.000104829865,0.019386591,0.00023841174],"about_ca_topic_score_codex":0.000009202906,"about_ca_topic_score_gemma":0.0000034012048,"teacher_disagreement_score":0.056323487,"about_ca_system_score_codex":0.00007218356,"about_ca_system_score_gemma":0.000018071074,"threshold_uncertainty_score":0.9973719},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408076290","doi":"10.5455/rmj.20240807010929","title":"Bibliometric analysis of research on emergency medicine and ultrasound-guided intubation","year":2025,"lang":"en","type":"article","venue":"Rawal Medical Journal","topic":"Medical Research and Treatments","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Intubation; Emergency ultrasound; Ultrasound; Medical physics; Emergency medicine; Intensive care medicine; Radiology; Anesthesia","score_opus":0.15543031941122012,"score_gpt":0.5325309979102845,"score_spread":0.37710067849906437,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408076290","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97593844,0.001676738,0.0014069574,0.008397966,0.00034518295,0.00020835813,0.000003796593,0.0000140813645,0.012008503],"genre_scores_gemma":[0.9798313,0.01816935,0.00006636074,0.00025926734,0.0003058627,0.0000087939,0.000029219706,0.000007647972,0.001322234],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9937761,0.0003998174,0.0008421454,0.00026710663,0.004250891,0.00046392737],"domain_scores_gemma":[0.99415606,0.003124692,0.00012808183,0.00026310165,0.00094343704,0.0013846305],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0043837754,0.000133225,0.00060378487,0.09735183,0.00018457643,0.00001264347,0.00021824609,0.00017322598,0.009245014],"category_scores_gemma":[0.04969294,0.00008200015,0.00015495191,0.1519188,0.0005438164,0.00005511576,0.00007191703,0.0011691088,0.000016825914],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00070995145,0.0010873404,0.6906809,0.0003360712,0.0045804624,0.00041366002,0.00020135994,0.0000023095592,0.0004754878,0.0016647978,0.1310057,0.16884193],"study_design_scores_gemma":[0.0033602754,0.0025598088,0.987716,0.0009397209,0.0009433676,0.00013870533,0.00018818882,0.00072673435,0.00016393099,0.0016808136,0.0015134803,0.000068956935],"about_ca_topic_score_codex":0.00017013802,"about_ca_topic_score_gemma":0.000041719384,"teacher_disagreement_score":0.2970351,"about_ca_system_score_codex":0.00015459035,"about_ca_system_score_gemma":0.00072075205,"threshold_uncertainty_score":0.99166065},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408108363","doi":"10.3138/jsp-2024-0032","title":"Visualization and Bibliometric Analysis of Research Evolution on Metaverse","year":2025,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"Virtual Reality Applications and Impacts","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Visualization; Data science; Metaverse; Computer science; Bibliometrics; Geography; Epistemology; World Wide Web; Human–computer interaction; Data mining; Virtual reality; Philosophy","score_opus":0.08448440436927343,"score_gpt":0.40207569730022935,"score_spread":0.3175912929309559,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408108363","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.43574104,0.00071992446,0.5574598,0.0030443876,0.00011194869,0.000072378345,0.0000030045308,0.000011755881,0.0028357662],"genre_scores_gemma":[0.9965786,0.000167356,0.0029852393,0.00015641282,0.000027363267,9.815336e-7,0.0000013348662,0.0000022977713,0.0000804551],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9976749,0.0002922968,0.00049486506,0.00016266809,0.0012161799,0.00015910389],"domain_scores_gemma":[0.995059,0.00060617743,0.00038927197,0.00032417628,0.0034948299,0.00012653212],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics","scholarly_communication"],"category_scores_codex":[0.009910945,0.000059544047,0.00022729766,0.1429166,0.00017014374,0.01073567,0.00086791563,0.00009471617,0.0000043169925],"category_scores_gemma":[0.01093837,0.00004996502,0.000094848474,0.26403123,0.000044224373,0.0369045,0.0002181862,0.00060077367,9.198098e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000026190728,0.00021532968,0.020126963,0.00002837709,0.0005949983,0.0000026733062,0.00047565368,0.00076643296,0.0023407352,0.908334,0.004641902,0.06244674],"study_design_scores_gemma":[0.0005357858,0.0003483428,0.9353713,0.00014387304,0.0002573183,0.000005678949,0.00047128633,0.0359889,0.0010310563,0.019784704,0.005968754,0.000093042036],"about_ca_topic_score_codex":0.00012677198,"about_ca_topic_score_gemma":0.0000069398725,"teacher_disagreement_score":0.9152443,"about_ca_system_score_codex":0.00018706982,"about_ca_system_score_gemma":0.00025872188,"threshold_uncertainty_score":0.9973929},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408128458","doi":"10.2196/preprints.56692","title":"Use of Artificial Intelligence, Internet of Things, and Edge Intelligence in Long-Term Care for Older People: Comprehensive Analysis Through Bibliometric, Google Trends, and Content Analysis (Preprint)","year":2024,"lang":"en","type":"preprint","venue":"","topic":"Impact of AI and Big Data on Business and Society","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Preprint; Enhanced Data Rates for GSM Evolution; The Internet; Term (time); Computer science; Content analysis; Data science; Internet privacy; World Wide Web; Artificial intelligence; Sociology; Social science","score_opus":0.2619095506350165,"score_gpt":0.4214502357368264,"score_spread":0.1595406851018099,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408128458","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.64989394,0.0048398315,0.34340823,0.00018774135,0.00029028836,0.0005258379,0.00076626765,0.000021603602,0.00006625311],"genre_scores_gemma":[0.9896008,0.0034667337,0.0060617877,0.00007830755,0.000031531083,0.000032552525,0.0004426387,0.00001900238,0.0002666792],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9940345,0.00018186278,0.0025055136,0.0016266708,0.0012456288,0.00040580682],"domain_scores_gemma":[0.99336845,0.002196023,0.0010794707,0.0011760346,0.002020221,0.00015978311],"candidate_categories":["metaepi_narrow","bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0014162549,0.0005387672,0.002475442,0.03170439,0.00007330863,0.0010726686,0.001010352,0.0004756877,0.0004674418],"category_scores_gemma":[0.0009035157,0.00038262093,0.0012301563,0.06740285,0.00044476538,0.0006088808,0.0029440473,0.00050856697,0.0000034110897],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021668995,0.00034126587,0.33897412,0.0010639624,0.004798993,0.000010064435,0.026662974,0.0023697538,0.000069586284,0.0023873667,0.00031394398,0.6227913],"study_design_scores_gemma":[0.00015003553,0.00016979757,0.8911554,0.00038433136,0.0045314096,0.000003115137,0.01960078,0.07547608,0.0028010784,0.0049422923,0.00008798396,0.0006976968],"about_ca_topic_score_codex":0.00867514,"about_ca_topic_score_gemma":0.004319906,"teacher_disagreement_score":0.6220936,"about_ca_system_score_codex":0.00006924494,"about_ca_system_score_gemma":0.00012584895,"threshold_uncertainty_score":0.9999643},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4408132838","doi":"10.1016/j.apenergy.2025.125638","title":"Borehole thermal energy storage systems: A comprehensive review using bibliometric and qualitative tools","year":2025,"lang":"en","type":"review","venue":"Applied Energy","topic":"Geothermal Energy Systems and Applications","field":"Energy","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Borehole; Thermal energy storage; Energy storage; Petroleum engineering; Environmental science; Computer science; Process engineering; Systems engineering; Engineering; Physics; Geotechnical engineering; Thermodynamics","score_opus":0.09154951289978336,"score_gpt":0.3541724801651441,"score_spread":0.2626229672653607,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408132838","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000007570392,0.96824586,0.00185598,0.000015649079,0.00030987465,0.0006718845,0.00021661092,0.0002122701,0.028464306],"genre_scores_gemma":[0.00039661676,0.9920825,0.00025160977,0.0005379352,0.00044808947,0.0026494781,0.0007645302,0.00021337868,0.002655867],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9944181,0.00094542693,0.0018537323,0.0013635894,0.0006047476,0.00081441266],"domain_scores_gemma":[0.99534696,0.0014373832,0.0012788989,0.0013633772,0.0002680178,0.00030537383],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00045809502,0.0012724268,0.0041576573,0.00872422,0.0003483511,0.00026468423,0.0007978658,0.0007369387,0.00009543998],"category_scores_gemma":[0.00005792064,0.0010794908,0.00061604666,0.026516952,0.0001958315,0.00016635262,0.00043412735,0.0003895976,0.000036083256],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004169363,0.00007019801,2.2023855e-8,0.04226308,0.00046724963,0.000012400019,0.000056669953,0.0029495945,0.000012644902,0.33858985,0.0010977881,0.6144763],"study_design_scores_gemma":[0.00027712845,0.000026624448,5.6034895e-7,0.025496498,0.0010461264,0.000040788615,0.00018330252,0.00025651138,0.0000033326314,0.00013786486,0.971528,0.0010032976],"about_ca_topic_score_codex":0.016670413,"about_ca_topic_score_gemma":0.000060496444,"teacher_disagreement_score":0.9704302,"about_ca_system_score_codex":0.0002703974,"about_ca_system_score_gemma":0.00046033613,"threshold_uncertainty_score":0.99916553},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408137166","doi":"10.1080/87559129.2025.2473006","title":"3D Food Printing Technology: A Critical Scientometric and Systematic Review, and Future Research Directions","year":2025,"lang":"en","type":"article","venue":"Food Reviews International","topic":"Additive Manufacturing and 3D Printing Technologies","field":"Engineering","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Biotechnology; Biology","score_opus":0.03524253021993304,"score_gpt":0.3494468382399118,"score_spread":0.3142043080199788,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408137166","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006766293,0.9656322,0.002385751,0.010422257,0.0010688853,0.0015469341,0.000030852436,0.0010708267,0.011075992],"genre_scores_gemma":[0.63014257,0.35606065,0.011546514,0.00026021083,0.00022623879,0.0011180987,0.000008231114,0.00003639901,0.0006010886],"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","domain_scores_codex":[0.998871,0.00007249394,0.00037636296,0.00026908284,0.00019053939,0.00022047243],"domain_scores_gemma":[0.99923575,0.00026042448,0.00004107191,0.00026283457,0.00016809154,0.000031812593],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011291176,0.0001389369,0.0003861721,0.0015271106,0.00018297584,0.00008818649,0.0003130377,0.00010512925,0.000019757337],"category_scores_gemma":[0.00459164,0.00011284283,0.000044378186,0.001620422,0.00020696258,0.00008580002,0.00034057206,0.00048484246,0.000014467607],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000029426649,0.00011367254,0.0009772184,0.4591333,0.0008102746,0.00000595326,0.0001467354,0.0000025057282,0.00016839706,0.25294018,0.010229901,0.27546892],"study_design_scores_gemma":[0.0004470115,0.00031660657,0.0025043718,0.14942755,0.0005815902,0.00020195292,0.00051441486,0.0009663544,0.0037771026,0.020245675,0.82020456,0.00081282156],"about_ca_topic_score_codex":5.103776e-7,"about_ca_topic_score_gemma":0.0000022322163,"teacher_disagreement_score":0.8099747,"about_ca_system_score_codex":0.00007250508,"about_ca_system_score_gemma":0.000010397824,"threshold_uncertainty_score":0.5496954},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408149387","doi":"10.1016/j.frl.2025.107143","title":"Exploring the landscape of green crowdfunding: Trends, themes, and insights from a bibliometric review","year":2025,"lang":"en","type":"review","venue":"Finance research letters","topic":"FinTech, Crowdfunding, Digital Finance","field":"Business, Management and Accounting","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"University of Waterloo; Royal Bank of Canada","keywords":"Regional science; Bibliometrics; Geography; Political science; Data science; Computer science; Library science","score_opus":0.22834967935828362,"score_gpt":0.36308432583219213,"score_spread":0.1347346464739085,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408149387","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00077845744,0.99044347,0.000011905845,0.0017410388,0.00032580076,0.0010040872,0.00006835995,0.0000738131,0.00555305],"genre_scores_gemma":[0.00045590324,0.99626833,0.00003712397,0.00061057106,0.0005764879,0.00065360963,0.00016269837,0.00008160741,0.0011536526],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99584013,0.000135036,0.0010074739,0.0010056723,0.0011456547,0.0008660086],"domain_scores_gemma":[0.9963709,0.0014723854,0.0007482278,0.0011022396,0.00028150534,0.000024707599],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.001291889,0.0006183881,0.0019157014,0.0293395,0.0003765198,0.00058716844,0.0017831683,0.00014861865,0.00005353232],"category_scores_gemma":[0.0015111121,0.00042215994,0.0004722719,0.091629416,0.00040398008,0.001739032,0.0015619741,0.0012434217,0.00013826138],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000068463723,0.00003133407,0.00004888782,0.06368247,0.0000828284,0.000038651226,0.000016993825,1.6650371e-7,0.0000024396445,0.0028682165,0.062079497,0.8711417],"study_design_scores_gemma":[0.00013494547,0.00001384955,0.0006000287,0.10076998,0.00019362634,0.0000025415152,0.000013240282,0.00000728459,0.0000013321517,0.00030058704,0.89759785,0.00036473144],"about_ca_topic_score_codex":0.001302164,"about_ca_topic_score_gemma":0.000066342356,"teacher_disagreement_score":0.87077695,"about_ca_system_score_codex":0.00010733583,"about_ca_system_score_gemma":0.00013462637,"threshold_uncertainty_score":0.99982303},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408158725","doi":"10.3389/fspas.2025.1485785","title":"Progress in the study of typical planetary landforms over the past 30 years: a bibliometric analysis in CiteSpace","year":2025,"lang":"en","type":"article","venue":"Frontiers in Astronomy and Space Sciences","topic":"Planetary Science and Exploration","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Natural Science Foundation of China","keywords":"Physics; Landform; Astronomy; Astrobiology; Geography; Cartography","score_opus":0.00994044624405185,"score_gpt":0.2503895520548777,"score_spread":0.24044910581082582,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408158725","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9973005,0.00030423095,0.0014588763,0.00035742557,0.000071646,0.00026730367,0.0000056180543,0.000001949247,0.00023241632],"genre_scores_gemma":[0.9990338,0.0000063854595,0.00087376084,0.000017968276,0.000018646077,0.000021969932,0.000010230845,8.0747066e-7,0.000016419135],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9990805,0.00010413342,0.00018090788,0.00022120937,0.00020367547,0.0002095735],"domain_scores_gemma":[0.99968624,0.000091011054,0.00006592168,0.0001316896,0.000007177854,0.000017957065],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0008914304,0.00008231132,0.00018500826,0.008002193,0.00007611321,0.00009976763,0.00035424985,0.000018472258,0.00001113033],"category_scores_gemma":[0.0000049774894,0.000046995545,0.000026518996,0.039148036,0.0002535363,0.0002866631,0.00004105682,0.00013800996,0.000001067763],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014152034,0.00011725807,0.9823002,0.0000015249087,0.000023109737,8.415017e-7,0.0015221259,0.0018477418,1.8918796e-7,0.000110839515,0.00031455077,0.013747439],"study_design_scores_gemma":[0.00031894163,0.00008625687,0.96732485,0.0000090447265,0.00002150391,4.8513808e-8,0.026854832,0.0045837895,7.9959557e-7,0.00028840476,0.00046095543,0.000050550207],"about_ca_topic_score_codex":0.0005187339,"about_ca_topic_score_gemma":0.0005425176,"teacher_disagreement_score":0.031145845,"about_ca_system_score_codex":0.000008516249,"about_ca_system_score_gemma":0.000046082343,"threshold_uncertainty_score":0.9812753},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408158802","doi":"10.3389/fnins.2025.1506566","title":"Bibliometric analysis of the intestinal microbiota and demyelinating diseases, particularly multiple sclerosis, since 2014","year":2025,"lang":"en","type":"article","venue":"Frontiers in Neuroscience","topic":"Gut microbiota and health","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Multiple sclerosis; Medicine; Demyelinating disease; Immunology","score_opus":0.012871805881872395,"score_gpt":0.2516583319843879,"score_spread":0.23878652610251552,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408158802","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9882931,0.002542522,0.008359773,0.00024597574,0.00036502743,0.00013783111,0.000031973243,0.0000040841437,0.000019675646],"genre_scores_gemma":[0.99717164,0.00045712243,0.0017311729,0.0005286836,0.000007546747,0.0000050425197,0.000005478179,0.0000039015235,0.00008938988],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9989939,0.000088114124,0.00022240852,0.00038255358,0.000089136454,0.0002239013],"domain_scores_gemma":[0.99948037,0.000026025844,0.00009951747,0.00029107515,0.000053772324,0.000049256658],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00021126734,0.000094315576,0.00016678355,0.00561284,0.00012371768,0.000036730227,0.00033952922,0.000049158396,4.593149e-7],"category_scores_gemma":[0.00069148163,0.000075984266,0.000073243515,0.036008254,0.00031974824,0.0000074536683,0.00025090628,0.00008727366,1.0690287e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000006839526,0.00002771043,0.5258543,0.000013549862,0.000006434628,2.280389e-7,0.0000056448303,0.000090512876,0.4717913,0.000007124778,0.0013653737,0.0008310037],"study_design_scores_gemma":[0.00019494424,0.00003290201,0.9772175,0.000033002783,0.000073242416,0.0000010136215,0.000018659413,0.0045669554,0.016403928,0.00000936237,0.0013784895,0.00006998924],"about_ca_topic_score_codex":0.000047361053,"about_ca_topic_score_gemma":0.000041365583,"teacher_disagreement_score":0.45538738,"about_ca_system_score_codex":0.000016053576,"about_ca_system_score_gemma":0.000114111295,"threshold_uncertainty_score":0.9844819},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4408202934","doi":"10.4102/hsag.v30i0.2776","title":"Impact of a Fulbright award: A bibliometric analysis of persistence","year":2025,"lang":"en","type":"article","venue":"Health SA Gesondheid","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Persistence (discontinuity); Library science; Humanities; Psychology; Art; Engineering; Computer science","score_opus":0.5185590538425793,"score_gpt":0.6212961476712524,"score_spread":0.10273709382867313,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408202934","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97822225,0.011163812,0.0046106926,0.0012802131,0.00027649614,0.00034012165,0.00028582438,0.00001856598,0.0038020497],"genre_scores_gemma":[0.99616784,0.0014669325,0.0008061823,0.00015498565,0.0000144716105,0.0000079778465,0.000008768672,0.000006553116,0.0013662907],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9883493,0.0004588943,0.001883635,0.00084643025,0.0075429794,0.0009187653],"domain_scores_gemma":[0.9860917,0.005774619,0.0010490285,0.0014835516,0.0050108484,0.0005902889],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.024036082,0.00020254795,0.001331271,0.86534745,0.00020217174,0.00036648184,0.0024687196,0.00013789425,0.0010035554],"category_scores_gemma":[0.021191578,0.00013447086,0.0011673699,0.97327375,0.000267047,0.00031595162,0.0005507967,0.0002594576,0.00004361205],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000053080603,0.00032560786,0.8249309,0.000050396575,0.00050295854,0.0000023409207,0.00016922478,0.0002783983,0.00017343163,0.0003661247,0.016987922,0.15615958],"study_design_scores_gemma":[0.0003659744,0.00048573758,0.98639166,0.000023583872,0.00008645539,0.0000010975282,0.00018074979,0.008396437,0.00019273878,0.00047018842,0.003300693,0.00010465735],"about_ca_topic_score_codex":0.002394888,"about_ca_topic_score_gemma":0.00009058295,"teacher_disagreement_score":0.16146076,"about_ca_system_score_codex":0.00038426,"about_ca_system_score_gemma":0.0019813897,"threshold_uncertainty_score":0.99990964},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408226040","doi":"10.1097/ms9.0000000000003089","title":"Global research trends on DPP-4 inhibitors and cardiovascular outcomes: a comprehensive bibliometric analysis","year":2025,"lang":"en","type":"review","venue":"Annals of Medicine and Surgery","topic":"Diabetes Treatment and Management","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Bibliometrics; Intensive care medicine; Library science","score_opus":0.3008204073311083,"score_gpt":0.47466993647867667,"score_spread":0.17384952914756835,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408226040","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012918633,0.99082017,0.00000174061,0.0016479545,0.00016873748,0.00048723747,0.000043578075,0.000026607177,0.0055121267],"genre_scores_gemma":[0.0010188194,0.9964076,0.000007837533,0.00068456354,0.0001599604,0.00006605365,0.0003081867,0.000018660065,0.0013283042],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99626833,0.00036977546,0.0009458781,0.00071586325,0.0012028106,0.00049734995],"domain_scores_gemma":[0.99544406,0.002765581,0.00021770963,0.0008164777,0.0004449404,0.0003112399],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0024716675,0.0004968954,0.0065090405,0.1297539,0.000081712904,0.000023203635,0.000104555154,0.00025599517,0.00005472803],"category_scores_gemma":[0.0007906264,0.00031020035,0.002049924,0.13859461,0.00039965048,0.000047667017,0.00016863631,0.00027089758,0.0000036605745],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000022651826,0.0001321405,0.011220184,0.014315899,0.022847362,0.00016499672,0.000014306537,2.3352206e-7,4.7125046e-9,0.00006280822,0.105425775,0.84579366],"study_design_scores_gemma":[0.0005778537,0.00040611313,0.025601031,0.016617622,0.021392396,0.0000059849885,0.0000914645,9.3438456e-7,1.9075908e-7,0.000033911227,0.93507314,0.00019933666],"about_ca_topic_score_codex":0.00029511953,"about_ca_topic_score_gemma":0.000007869899,"teacher_disagreement_score":0.8455943,"about_ca_system_score_codex":0.000036155598,"about_ca_system_score_gemma":0.00015325377,"threshold_uncertainty_score":0.99993503},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408256239","doi":"10.5267/j.dsl.2024.12.011","title":"Evolution and gaps in data mining research: Identifying the bibliometric landscape of data mining in management","year":2025,"lang":"en","type":"article","venue":"Decision Science Letters","topic":"Big Data and Business Intelligence","field":"Business, Management and Accounting","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"King Faisal University","keywords":"Bibliometrics; Data mining; Data science; Computer science","score_opus":0.36163583625297413,"score_gpt":0.4476337674619509,"score_spread":0.08599793120897675,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408256239","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98045945,0.0007046287,0.013140024,0.0035657024,0.0004399966,0.00026524463,0.000009661078,0.000013600262,0.0014016728],"genre_scores_gemma":[0.99564517,0.00016798158,0.0033917748,0.00066906074,0.00005957773,0.0000057847433,0.000034005836,0.000005280844,0.000021341937],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9970841,0.000030794617,0.0004590591,0.00083391485,0.001179068,0.00041306712],"domain_scores_gemma":[0.99736905,0.00062210846,0.00013536889,0.0017371438,0.00012562727,0.000010685215],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.011826875,0.000106508276,0.0001588856,0.03369233,0.0002955762,0.00089199585,0.0043318598,0.00003305141,0.000025376545],"category_scores_gemma":[0.0020748551,0.000079518235,0.0000101515425,0.10350146,0.00047665412,0.0055590896,0.007468004,0.00017879996,0.00001650283],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006347673,0.000074078795,0.4528228,0.00027827857,0.0000073188867,0.000040667892,0.00012307764,0.00016431288,0.0023796149,0.002101489,0.047885954,0.4940589],"study_design_scores_gemma":[0.00042576343,0.0000024172361,0.89123195,0.0011951827,0.0000151832055,0.0000020355521,0.0035684467,0.09061003,0.000036294798,0.0010847353,0.0116605405,0.00016740747],"about_ca_topic_score_codex":0.00042636777,"about_ca_topic_score_gemma":0.00031369476,"teacher_disagreement_score":0.4938915,"about_ca_system_score_codex":0.000038129358,"about_ca_system_score_gemma":0.00003507094,"threshold_uncertainty_score":0.97725993},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408295417","doi":"10.3390/jrfm18030145","title":"A Selective Systematic Review and Bibliometric Analysis of Gender and Financial Literacy Research in Developing Countries","year":2025,"lang":"en","type":"article","venue":"Journal of risk and financial management","topic":"Financial Literacy, Pension, Retirement Analysis","field":"Business, Management and Accounting","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Financial literacy; Systematic review; Political science; Psychology; Business; MEDLINE; Finance","score_opus":0.02293620829995311,"score_gpt":0.30889915539310925,"score_spread":0.2859629470931561,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408295417","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8881006,0.1073955,0.0021723218,0.00024932998,0.00016675309,0.0011869837,0.000010116301,0.00001172974,0.0007066659],"genre_scores_gemma":[0.90509826,0.09305944,0.00066986715,0.0009951962,0.000084634084,0.00002389824,0.000004273612,0.000010137331,0.000054313663],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9973397,0.00012310254,0.0013017068,0.00032675138,0.00059605384,0.00031265977],"domain_scores_gemma":[0.99751365,0.00029805145,0.0008750209,0.0001945476,0.0010968168,0.000021939413],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0042830645,0.00022270465,0.0013177723,0.057221122,0.00024426315,0.0002601887,0.00021776182,0.000081704944,0.000009607399],"category_scores_gemma":[0.0012929857,0.00018628461,0.0001542876,0.08057246,0.000116812655,0.0008394142,0.00036146364,0.00028588565,0.0000016212624],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017658656,0.00012242013,0.795424,0.14164871,0.00032658532,0.000068761714,0.0004990378,0.000016850505,0.0000020144091,0.054570023,0.0009556327,0.0061894134],"study_design_scores_gemma":[0.0006428022,0.000028692724,0.9673344,0.015632292,0.0056760707,0.0000018157431,0.00012243039,0.00031168276,0.0000021383073,0.0043039024,0.005745125,0.00019863991],"about_ca_topic_score_codex":0.00023505192,"about_ca_topic_score_gemma":0.00012392267,"teacher_disagreement_score":0.17191045,"about_ca_system_score_codex":0.00008678005,"about_ca_system_score_gemma":0.00007258423,"threshold_uncertainty_score":0.95346445},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408331283","doi":"10.1007/s44290-025-00205-z","title":"Big data applications in intelligent transport systems: a bibliometric analysis and review","year":2025,"lang":"en","type":"article","venue":"Discover Civil Engineering","topic":"Traffic Prediction and Management Techniques","field":"Engineering","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Big data; Data science; Computer science; Data mining","score_opus":0.01973147855671384,"score_gpt":0.2510723541345324,"score_spread":0.23134087557781854,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408331283","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012477643,0.044754103,0.95049834,0.000023448025,0.00015015839,0.0004222518,0.000052546868,0.0011654838,0.0016859248],"genre_scores_gemma":[0.9541776,0.045233052,0.00017080842,0.00002472572,0.000017248947,0.00017327379,0.00015328887,0.000013306975,0.000036665115],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992324,0.0000054699167,0.00028832385,0.00022276855,0.00010437572,0.00014662684],"domain_scores_gemma":[0.9994216,0.000030022855,0.000014635433,0.00048536548,0.00001050277,0.000037881306],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00021340772,0.00013179884,0.00025262655,0.019961039,0.000013826065,0.000033093278,0.0002476658,0.00003709879,0.000003035479],"category_scores_gemma":[0.000014009732,0.00013416994,0.000042110645,0.051749848,0.000007808081,0.00016377043,0.00006154156,0.00010639281,0.000001203844],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004187216,0.00019559072,0.063781865,0.026712796,0.0030998928,0.000018682666,0.00015468568,0.6960407,0.00028970113,0.011088149,0.020342438,0.17827134],"study_design_scores_gemma":[0.00016775157,0.0000055818455,0.10908852,0.0015269931,0.0008694566,0.0000010228466,0.00004456029,0.7001426,0.000062801235,0.0000073902816,0.18772955,0.00035376713],"about_ca_topic_score_codex":0.00005891999,"about_ca_topic_score_gemma":0.00020501332,"teacher_disagreement_score":0.95292985,"about_ca_system_score_codex":0.00005174997,"about_ca_system_score_gemma":0.000008473595,"threshold_uncertainty_score":0.99114686},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408361789","doi":"10.5206/cjils-rcsib.v48i1.22439","title":"A Bibliometric Analysis of Canadian LIS Scholars and Practitioners’ Research Contributions","year":2025,"lang":"en","type":"article","venue":"Canadian Journal of Information and Library Science","topic":"Wikis in Education and Collaboration","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"Université du Québec à Montréal; Dalhousie University; Université de Montréal","funders":"","keywords":"Library science; Bibliometrics; Sociology; Political science; Computer science","score_opus":0.025887968119534834,"score_gpt":0.3619810911431926,"score_spread":0.3360931230236578,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408361789","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7998262,0.0008899936,0.00017645626,0.044523366,0.0005034673,0.00017529342,0.000114122755,0.000005735211,0.15378535],"genre_scores_gemma":[0.9979763,0.0006032444,0.00033837766,0.0007769177,0.000019192985,0.000001283461,0.000007599356,6.753427e-7,0.0002764054],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990166,0.000108319124,0.00027996564,0.00005029467,0.00033766433,0.00020716195],"domain_scores_gemma":[0.9975011,0.00013537925,0.00015566956,0.0000693612,0.0013917998,0.0007466938],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0028266076,0.00003256204,0.0000869357,0.19246374,0.001277824,0.0010016479,0.0002368966,0.000048968188,0.00021650038],"category_scores_gemma":[0.0032627499,0.000032588363,0.000020738813,0.27786142,0.00082970335,0.008434814,0.000010762806,0.00015289172,0.0000018176047],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000007656025,0.000009823901,0.14765172,0.000012547884,0.000075272896,0.0000017366843,0.023739181,0.0000737922,0.00002655454,0.7588041,0.035412103,0.03418551],"study_design_scores_gemma":[0.00009967634,0.000021020267,0.43061832,0.000024642519,0.000028929548,0.0000014867555,0.02200475,0.00014009107,0.000060296,0.001017122,0.54594105,0.000042626725],"about_ca_topic_score_codex":0.17317592,"about_ca_topic_score_gemma":0.32122475,"teacher_disagreement_score":0.757787,"about_ca_system_score_codex":0.00016561718,"about_ca_system_score_gemma":0.024108114,"threshold_uncertainty_score":0.98281187},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408469039","doi":"10.52783/jisem.v10i18s.2884","title":"Global Trends in Foreign Direct Investment: Findings from Bibliometric Analysis for Policy Recommendations","year":2025,"lang":"en","type":"article","venue":"Journal of Information Systems Engineering & Management","topic":"International Business and FDI","field":"Business, Management and Accounting","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Foreign direct investment; Bibliometrics; Regional science; Economics; Business; Political science; Geography; Computer science; Library science; Macroeconomics","score_opus":0.01230572657297759,"score_gpt":0.253477430818403,"score_spread":0.24117170424542542,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408469039","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06269458,0.0004866686,0.70168257,0.0031848757,0.0037924342,0.000949055,0.00014957701,0.00014787272,0.2269124],"genre_scores_gemma":[0.99735266,0.000028564631,0.001270696,0.0005934105,0.00029528796,0.00004510973,0.00017320419,0.0000060852426,0.00023497724],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9984413,0.0000047460144,0.00096407754,0.00008701364,0.00031846407,0.0001844401],"domain_scores_gemma":[0.99894434,0.000047600857,0.0004974631,0.00012452535,0.00037190007,0.000014177306],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0006147633,0.0001481883,0.00031555822,0.086840495,0.00005793458,0.00058643485,0.0002733399,0.000049039587,0.000018971055],"category_scores_gemma":[0.000115855095,0.00013784622,0.00019604852,0.067579076,0.000005965402,0.002479298,0.00007984249,0.00006085222,0.000008749401],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004987404,0.00005900773,0.026018342,0.00040481158,0.0014331326,0.000003129653,0.00003462476,0.33931485,0.000001831689,0.6003538,0.013933796,0.018392833],"study_design_scores_gemma":[0.0015670205,0.000009468094,0.53364307,0.00033809408,0.00044823348,0.0000011212069,0.0002644968,0.14212,0.0000021572343,0.0012240398,0.32016936,0.00021292859],"about_ca_topic_score_codex":0.00061383704,"about_ca_topic_score_gemma":0.0000141915125,"teacher_disagreement_score":0.9346581,"about_ca_system_score_codex":0.00042114602,"about_ca_system_score_gemma":0.00001738669,"threshold_uncertainty_score":0.95224},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408636096","doi":"10.1017/rsm.2024.3","title":"Visualization toolkits for enriching meta-analyses through evidence maps, bibliometrics, and alternative impact metrics","year":2025,"lang":"en","type":"article","venue":"Research Synthesis Methods","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Australian Research Council; National Natural Science Foundation of China","keywords":"Bibliometrics; Visualization; Computer science; Meta-analysis; Data science; Information retrieval; Data mining; Medicine","score_opus":0.982746142638343,"score_gpt":0.805248268991111,"score_spread":0.17749787364723202,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408636096","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001713831,0.048691392,0.9437038,0.0013378415,0.0001898894,0.0016460272,0.000069546004,0.0000126415825,0.0026350443],"genre_scores_gemma":[0.12253013,0.0048987484,0.86258554,0.0002664572,0.00014614195,0.0008595919,0.0000070266415,0.000043935863,0.0086624315],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.8034738,0.16681205,0.01066261,0.0031474973,0.014674415,0.0012296609],"domain_scores_gemma":[0.1908421,0.78373504,0.0048310026,0.0062780106,0.01381948,0.0004943533],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.5816067,0.0005222296,0.0063630454,0.026919322,0.0007534629,0.00436651,0.003158907,0.0002025652,0.002600378],"category_scores_gemma":[0.8696154,0.0002434018,0.0038821585,0.08964452,0.00023364996,0.0015205265,0.0007456948,0.00039299444,0.00011309801],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011004592,0.00026643637,0.006843273,0.00076185085,0.027601728,0.000007911289,0.00092468574,0.00068270747,0.005437054,0.050499726,0.148387,0.75847757],"study_design_scores_gemma":[0.00045908566,0.00038359806,0.008023359,0.00042704315,0.022276776,0.0000124529915,0.002605938,0.17205834,0.058541358,0.44629043,0.28779453,0.001127085],"about_ca_topic_score_codex":0.0003299767,"about_ca_topic_score_gemma":0.000013354721,"teacher_disagreement_score":0.7573505,"about_ca_system_score_codex":0.00023062559,"about_ca_system_score_gemma":0.00032468373,"threshold_uncertainty_score":0.9983114},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4408804230","doi":"10.5498/wjp.v15.i4.104600","title":"Comprehensive bibliometric analysis of transcranial magnetic stimulation-based depression treatment from 2003 to 2022: Research hotspots and trends","year":2025,"lang":"en","type":"article","venue":"World Journal of Psychiatry","topic":"Transcranial Magnetic Stimulation Studies","field":"Neuroscience","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Transcranial magnetic stimulation; Depression (economics); Medicine; Stimulation; Internal medicine","score_opus":0.059646626735569135,"score_gpt":0.3823520145630894,"score_spread":0.3227053878275203,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408804230","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9860109,0.0074847527,0.0021282346,0.002367843,0.0008765959,0.00025846148,0.00012544851,0.000013525087,0.0007342196],"genre_scores_gemma":[0.9944887,0.00024153078,0.0044110706,0.00027742446,0.00008719558,0.0000074893583,0.0000021363378,0.000012571102,0.00047183855],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9972744,0.0004048342,0.00081419235,0.0003832119,0.00085486396,0.00026849352],"domain_scores_gemma":[0.99709123,0.001624371,0.00022866299,0.00028329348,0.0006060487,0.0001663939],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00026775288,0.00020414607,0.00063540967,0.09126696,0.00020656773,0.000057581332,0.0002573282,0.00006007917,0.0005330787],"category_scores_gemma":[0.000221321,0.00016997031,0.00022864134,0.18811618,0.00016116345,0.00008656757,0.000025396417,0.00022842946,0.0000028358295],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003997679,0.0013503567,0.56294155,0.00006611646,0.0006393067,0.000022952978,0.00065994775,0.04452378,0.1816907,0.00037409415,0.0026393651,0.20109417],"study_design_scores_gemma":[0.0040588174,0.0008110472,0.9870108,0.00010267453,0.0009364528,0.000001009712,0.00006641228,0.004055049,0.0008945206,0.000312478,0.0016254085,0.00012537079],"about_ca_topic_score_codex":0.00018730524,"about_ca_topic_score_gemma":0.00045546994,"teacher_disagreement_score":0.42406923,"about_ca_system_score_codex":0.00006201572,"about_ca_system_score_gemma":0.00017820511,"threshold_uncertainty_score":0.91903275},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4408834171","doi":"10.31246/mjn-2024-0016","title":"Female athlete triad syndrome: A bibliometric analysis","year":2025,"lang":"en","type":"article","venue":"Malaysian Journal of Nutrition","topic":"Cardiovascular Effects of Exercise","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Triad (sociology); Athletes; Medicine; Psychology; Physical therapy","score_opus":0.01009715864070624,"score_gpt":0.282105433664098,"score_spread":0.27200827502339175,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408834171","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9806088,0.009400953,0.007263856,0.001067263,0.00028572284,0.00037165085,0.0000063067773,0.000023103396,0.00097233575],"genre_scores_gemma":[0.9954236,0.0010675235,0.002852884,0.00013638819,0.00018780823,0.000012336728,0.000014150282,0.000014413983,0.00029085457],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9982192,0.00013785288,0.00062860333,0.00019099661,0.00059870037,0.00022467975],"domain_scores_gemma":[0.99851644,0.00014338952,0.00027304358,0.00038043753,0.0005000941,0.0001866036],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.001192949,0.0001570746,0.0010557526,0.1034657,0.00005368514,0.00006455348,0.00014767503,0.00012128794,0.000114362694],"category_scores_gemma":[0.00046922217,0.0001407996,0.0017690458,0.1278032,0.00004433746,0.00020608277,0.000030407185,0.00031800027,0.000022096043],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.011943869,0.014289635,0.3831719,0.00817528,0.07818432,0.042654797,0.0004906843,0.0018418168,0.043226507,0.0005888602,0.036910072,0.37852228],"study_design_scores_gemma":[0.017351974,0.0018309742,0.9333195,0.002573498,0.024116222,0.0064522773,0.00030900075,0.0005864943,0.0046681357,0.00033743534,0.008136627,0.00031783653],"about_ca_topic_score_codex":0.0000050345884,"about_ca_topic_score_gemma":6.0722164e-7,"teacher_disagreement_score":0.55014765,"about_ca_system_score_codex":0.00016915919,"about_ca_system_score_gemma":0.00008982746,"threshold_uncertainty_score":0.9066957},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4408846332","doi":"10.1097/ceh.0000000000000602","title":"Measuring the Longitudinal Impact of a Health Professions Education Academy on Scholarship: A Bibliometric Analysis","year":2025,"lang":"en","type":"article","venue":"Journal of Continuing Education in the Health Professions","topic":"Innovations in Medical Education","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scholarship; Bibliometrics; Popularity; Faculty development; Health professions; Medical education; Higher education; Political science; Scholarship of Teaching and Learning; Professional development; Sociology; Public relations; Health care; Medicine; Library science; Pedagogy; Teaching method; Computer science","score_opus":0.09174690030629781,"score_gpt":0.5147054759507477,"score_spread":0.42295857564444983,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408846332","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8424269,0.0038009249,0.0010741843,0.14934775,0.0017177537,0.0013095764,0.0000023166294,0.000011265263,0.00030933585],"genre_scores_gemma":[0.9906347,0.00045156694,0.0014691872,0.0061675217,0.00032981235,0.00010803704,0.000017589227,0.000014819854,0.0008067939],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9941134,0.0019979533,0.0022522502,0.00022664558,0.0009949708,0.000414806],"domain_scores_gemma":[0.99401176,0.0009523549,0.0028073406,0.0005641335,0.0015214679,0.00014291989],"candidate_categories":["bibliometrics","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.014165411,0.00019615534,0.00073338207,0.037127323,0.00058829726,0.0000455322,0.0005413961,0.00016564294,0.000047560723],"category_scores_gemma":[0.00724733,0.00010670957,0.00033676365,0.08574717,0.000111793124,0.00025805185,0.000045116998,0.0023467306,0.0000033318356],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017833342,0.0033925725,0.49895126,0.00051118596,0.00032623854,2.6574645e-7,0.015157179,0.000054532644,0.000059151604,0.0027273097,0.04470916,0.4339328],"study_design_scores_gemma":[0.0005356278,0.00032851097,0.9671605,0.0071264883,0.00023681334,0.000026078098,0.022001833,0.00004775548,0.00001735608,0.0012214414,0.0012276679,0.000069926195],"about_ca_topic_score_codex":0.0002070501,"about_ca_topic_score_gemma":0.0000169644,"teacher_disagreement_score":0.46820924,"about_ca_system_score_codex":0.0013024365,"about_ca_system_score_gemma":0.018406883,"threshold_uncertainty_score":0.9999549},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408888505","doi":"10.2196/preprints.71844","title":"Knowledge Mapping and Global Trends in Simulation in Medical Education: Bibliometric and Visual Analysis (Preprint)","year":2025,"lang":"en","type":"preprint","venue":"","topic":"Simulation-Based Education in Healthcare","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Preprint; Computer science; Library science; Data science; Regional science; Geography; World Wide Web","score_opus":0.05574838904828912,"score_gpt":0.48163364858907104,"score_spread":0.42588525954078194,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408888505","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98118806,0.0044049984,0.0018384518,0.0049113007,0.00058841263,0.000675745,0.000013811784,0.00007730105,0.0063019022],"genre_scores_gemma":[0.9969609,0.0002555163,0.00080794044,0.0005542964,0.00020041541,0.0001259348,0.00042824293,0.000010077651,0.00065667334],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99702024,0.0002322172,0.0010427018,0.0009282606,0.0005041643,0.0002724182],"domain_scores_gemma":[0.9977146,0.0008207977,0.00018690176,0.0004089258,0.00048895867,0.00037985205],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0010518181,0.00028939685,0.0007443179,0.07297971,0.00004801805,0.00006696236,0.00011306399,0.0006657178,0.0010906908],"category_scores_gemma":[0.0009901698,0.00030496917,0.000108951725,0.087902136,0.00007493849,0.00008030324,0.00035957477,0.00066711265,0.000004314332],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003412839,0.0003774879,0.81294,0.00061160565,0.000104712424,0.0000025665622,0.0007813081,0.0063092,9.6785094e-8,0.00014753459,0.000098383396,0.178593],"study_design_scores_gemma":[0.0006216892,0.00001758297,0.6919203,0.00052032643,0.00013575493,0.0000028431043,0.00062053977,0.30564862,6.506574e-7,0.00026465167,0.0001020789,0.00014498261],"about_ca_topic_score_codex":0.0037885217,"about_ca_topic_score_gemma":0.0014607022,"teacher_disagreement_score":0.2993394,"about_ca_system_score_codex":0.0009753807,"about_ca_system_score_gemma":0.0034220687,"threshold_uncertainty_score":0.9999402},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4409019216","doi":"10.1109/icietdw61607.2024.10939117","title":"Artificial Intelligence in Gaming: A Bibliometric Analysis of Research Outputs and Trends","year":2024,"lang":"en","type":"article","venue":"","topic":"Sports Analytics and Performance","field":"Economics, Econometrics and Finance","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Computer science; Intelligence analysis; Data science; Artificial intelligence","score_opus":0.18071758407052715,"score_gpt":0.3724763052267386,"score_spread":0.19175872115621143,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409019216","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94352674,0.011857542,0.0071604396,0.00048131455,0.00017516997,0.000056742505,0.00005847553,0.00002242724,0.036661156],"genre_scores_gemma":[0.9960934,0.0016688018,0.00013785741,0.000013845148,0.000031249274,0.0000044568665,0.0000075218322,0.0000076749375,0.0020351752],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9987024,0.0000066814778,0.0006359055,0.00034690535,0.000075244796,0.0002328774],"domain_scores_gemma":[0.99951315,0.0001361461,0.000058453563,0.00019904866,0.000039421546,0.00005376879],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0020378325,0.00007524437,0.00033637777,0.19772679,0.000029764502,0.00012983716,0.00012682588,0.00006187707,0.0013904009],"category_scores_gemma":[0.00007410419,0.000074317686,0.00009459054,0.23691732,0.0000863246,0.00014274944,0.000058429396,0.00016411538,0.00005604563],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000069342555,0.000064835804,0.19930227,0.000051415795,0.00019694756,0.000012521874,0.0004914982,0.00063082506,0.000006297627,0.66511136,0.0001526035,0.1339725],"study_design_scores_gemma":[0.000017855415,0.000057220557,0.37643942,0.000019646981,0.000026412794,7.0243084e-7,0.00012172677,0.601602,0.000092021575,0.01588211,0.005614602,0.00012624086],"about_ca_topic_score_codex":0.0011191749,"about_ca_topic_score_gemma":0.0004436093,"teacher_disagreement_score":0.6492292,"about_ca_system_score_codex":0.000038267342,"about_ca_system_score_gemma":0.0000115377825,"threshold_uncertainty_score":0.99952245},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4409048057","doi":"10.47772/ijriss.2025.90300097","title":"Global Trends in Assessing Social and Emotional Development in Early Childhood Education: A Bibliometric Analysis (2020–2025)","year":2025,"lang":"en","type":"article","venue":"International Journal of Research and Innovation in Social Science","topic":"Psychology of Development and Education","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Social emotional learning; Early childhood education; Psychology; Political science; Developmental psychology","score_opus":0.09075387536008997,"score_gpt":0.5243502633705066,"score_spread":0.43359638801041667,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409048057","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97384673,0.00016801232,0.00014684607,0.005594726,0.0008595706,0.000055652246,0.000001355478,0.0000027415515,0.019324377],"genre_scores_gemma":[0.99864215,0.000025059242,0.0005646758,0.00016945525,0.00019998747,0.0000118516,0.0000086729715,0.0000018173176,0.00037631698],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9977937,0.00012236605,0.00068946596,0.00025456428,0.0008646796,0.00027526703],"domain_scores_gemma":[0.9983106,0.0000893216,0.00021522645,0.000038186394,0.0013084552,0.000038199145],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.004312621,0.00007521709,0.00016148401,0.09637107,0.00022365486,0.00026207374,0.00037033478,0.00009023503,0.000102313694],"category_scores_gemma":[0.0003493678,0.00007841116,0.000021026337,0.20271862,0.00041456494,0.00066848274,0.000089938076,0.00033569336,0.0000015260609],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000034020573,0.00033402507,0.72862214,0.0000015467115,0.000044848694,0.000004755953,0.0041636396,5.193655e-7,0.000021681793,0.025124269,0.0008956714,0.24075289],"study_design_scores_gemma":[0.0006873631,0.000021975191,0.9887087,0.000047411373,0.0000026941748,0.000007224727,0.0021726333,0.000003839822,0.000005351274,0.0076241754,0.00065843604,0.000060237206],"about_ca_topic_score_codex":0.000101280755,"about_ca_topic_score_gemma":0.000068499576,"teacher_disagreement_score":0.2600865,"about_ca_system_score_codex":0.00067812606,"about_ca_system_score_gemma":0.0015211083,"threshold_uncertainty_score":0.9138708},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4409132844","doi":"10.35445/alishlah.v17i1.6695","title":"Mapping of Religious Moderation Literature in Higher Education: A Bibliometric Review","year":2025,"lang":"en","type":"review","venue":"AL-ISHLAH Jurnal Pendidikan","topic":"Islamic Studies and Radicalism","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Moderation; Psychology; Sociology; Social science; Social psychology","score_opus":0.0572340008128816,"score_gpt":0.3962072751740805,"score_spread":0.3389732743611989,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409132844","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000022138677,0.94172597,0.0000031758445,0.003782996,0.0019967568,0.0010366184,0.000033023396,0.000026588816,0.05139267],"genre_scores_gemma":[0.000038245293,0.9859983,0.00010575342,0.0015383495,0.0008166947,0.0001366138,0.000101966754,0.000023265411,0.011240815],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9964592,0.00062583183,0.001236712,0.00049714814,0.00073945674,0.00044167],"domain_scores_gemma":[0.9979894,0.0004349037,0.0007642655,0.00038287553,0.0003235036,0.00010503342],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0013377275,0.00037683683,0.0018456961,0.014212527,0.00031561535,0.00016411084,0.00063839956,0.0004274249,0.000258868],"category_scores_gemma":[0.0005076321,0.000307983,0.0005916161,0.061052836,0.00014089725,0.00022561643,0.00012517914,0.00074437325,0.00002434286],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000119271,0.00008516135,0.000009687476,0.03189403,0.000079249876,0.000015885857,0.00082731305,7.4758006e-8,3.132145e-8,0.006185935,0.21809328,0.74280816],"study_design_scores_gemma":[0.000074658674,0.000012547282,0.00004626265,0.14481148,0.00019416162,0.000010236639,0.00006585501,2.1559173e-7,1.6352173e-8,0.0006258524,0.85392034,0.00023836046],"about_ca_topic_score_codex":0.000632665,"about_ca_topic_score_gemma":0.00010232108,"teacher_disagreement_score":0.7425698,"about_ca_system_score_codex":0.00035831422,"about_ca_system_score_gemma":0.0020709068,"threshold_uncertainty_score":0.99993724},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4409167433","doi":"10.1007/s44337-025-00290-0","title":"Research hotspots and publication trends of high flow nasal oxygen: a bibliometric analysis from 2004 to 2023","year":2025,"lang":"en","type":"article","venue":"Discover Medicine","topic":"Cardiovascular and Diving-Related Complications","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Lanzhou Science and Technology Bureau","keywords":"Environmental science; Bibliometrics; Computer science; Library science","score_opus":0.033369995618828474,"score_gpt":0.36324898399496547,"score_spread":0.329878988376137,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409167433","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92087394,0.0104972,0.018007703,0.026699841,0.00020097055,0.00044504472,0.0001604983,0.000056970977,0.023057831],"genre_scores_gemma":[0.9933603,0.00030246718,0.00038356066,0.00043292082,0.00014714478,0.00006104419,0.0006468118,0.000011621727,0.0046541356],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9978799,0.00009981307,0.00042086837,0.00047855888,0.0008629247,0.00025791992],"domain_scores_gemma":[0.99778146,0.00032609273,0.00005991417,0.0009116178,0.0006828498,0.00023805066],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.000951087,0.00013799775,0.0006279911,0.08198604,0.00008967292,0.000031539847,0.00017100897,0.000107626365,0.00064007676],"category_scores_gemma":[0.0011004729,0.00010926045,0.00019793212,0.22738491,0.00016442788,0.000095331205,0.00012754922,0.00024878368,0.000026186039],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012646222,0.0011087708,0.17378972,0.00028456774,0.01883473,0.000046802874,0.0021036898,0.0011359955,0.013675493,0.00638125,0.41779998,0.36357436],"study_design_scores_gemma":[0.0022675828,0.00015584566,0.98916817,0.0001640349,0.0020234135,0.0000031039415,0.00025648947,0.00086078956,0.00024437043,0.00040559983,0.0043643676,0.000086215274],"about_ca_topic_score_codex":0.005766832,"about_ca_topic_score_gemma":0.00012893799,"teacher_disagreement_score":0.8153785,"about_ca_system_score_codex":0.00009153903,"about_ca_system_score_gemma":0.00013643056,"threshold_uncertainty_score":0.9284189},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4409213925","doi":"10.1080/03155986.2025.2485795","title":"Dynamic analysis of the time-dependent progress of h-index","year":2025,"lang":"en","type":"article","venue":"INFOR Information Systems and Operational Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Index (typography); Computer science; World Wide Web","score_opus":0.2568807318774551,"score_gpt":0.5561487956127815,"score_spread":0.29926806373532644,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409213925","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9305727,0.0011665268,0.012516682,0.0020533025,0.00053934165,0.001903833,0.000511884,0.000013524774,0.050722245],"genre_scores_gemma":[0.99545425,0.000036447993,0.00004861374,0.000032440996,0.0000051180327,0.000036457852,0.000016663718,0.0000014134462,0.004368564],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98851204,0.00028806014,0.0012718943,0.00016757377,0.009510704,0.00024970534],"domain_scores_gemma":[0.9863609,0.0023654934,0.00028465738,0.000532064,0.010368261,0.00008860974],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.02697964,0.0000740358,0.0002993559,0.039272014,0.00033065453,0.0015135399,0.0012225558,0.000094763265,0.00011475575],"category_scores_gemma":[0.010213248,0.000041134634,0.00011967195,0.107039586,0.00036635395,0.0012331869,0.00066111726,0.00019667468,0.000040744875],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015271503,0.00014855775,0.710974,0.00020522632,0.0008071739,4.636099e-7,0.0014258632,0.033750854,0.0005952657,0.12569605,0.0089852065,0.11725862],"study_design_scores_gemma":[0.00031747203,0.0000470895,0.37441918,0.000029799607,0.000016170772,9.577146e-7,0.0012518741,0.6018962,0.00029393416,0.00032818096,0.02134101,0.000058075562],"about_ca_topic_score_codex":0.0002806004,"about_ca_topic_score_gemma":0.000025988225,"teacher_disagreement_score":0.5681454,"about_ca_system_score_codex":0.00009182444,"about_ca_system_score_gemma":0.00061977905,"threshold_uncertainty_score":0.999523},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4409329075","doi":"10.3390/agriculture15080826","title":"Progress and Trends in Coupled Model Intercomparison Project (CMIP) Research: A Bibliometric Analysis","year":2025,"lang":"en","type":"article","venue":"Agriculture","topic":"Climate variability and models","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"","keywords":"Coupled model intercomparison project; Environmental science; Climate change; Biology; Climate model; Ecology","score_opus":0.07746889844524493,"score_gpt":0.3759554840814092,"score_spread":0.29848658563616426,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409329075","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9916688,0.00028611883,0.0002720916,0.0009370332,0.000017213706,0.00025410493,0.000010112233,0.000032455664,0.0065220734],"genre_scores_gemma":[0.9976426,0.000097893426,0.0007514264,0.000044216253,0.000006396237,0.00008223482,0.0000317562,0.0000029778107,0.0013405109],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9985737,0.00010791621,0.00021358913,0.000450005,0.0003317998,0.00032296282],"domain_scores_gemma":[0.99958915,0.000078954654,0.000033663735,0.00021649615,0.000027000511,0.000054765504],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00074670993,0.00012103182,0.00023712705,0.018497396,0.000103346654,0.00009617562,0.0002260969,0.0001261604,0.000103008024],"category_scores_gemma":[0.000056299385,0.00007865118,0.00006982265,0.1966401,0.00015946974,0.00018369,0.00034957472,0.00029007753,0.000008489882],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008910716,0.0010201386,0.8883227,0.000067313566,0.00019813146,0.000009926802,0.0027080912,0.040674664,0.0030354687,0.00082399754,0.03373489,0.029315578],"study_design_scores_gemma":[0.00028680806,0.000036118872,0.57908386,0.000019104651,0.00007934918,7.552208e-7,0.00024723596,0.41848883,0.00004840086,0.00060208776,0.0009850916,0.00012239086],"about_ca_topic_score_codex":0.0017292799,"about_ca_topic_score_gemma":0.0052789934,"teacher_disagreement_score":0.37781414,"about_ca_system_score_codex":0.0001618652,"about_ca_system_score_gemma":0.000008831298,"threshold_uncertainty_score":0.9926271},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4409340083","doi":"10.12688/f1000research.157349.2","title":"A global overview of the use of cone beam computed tomography in dentistry: a bibliometric review focusing on paediatric patients","year":2025,"lang":"en","type":"review","venue":"F1000Research","topic":"Dental Radiography and Imaging","field":"Dentistry","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Fundação de Amparo à Pesquisa e Inovação do Estado de Santa Catarina; Coordenação de Aperfeiçoamento de Pessoal de Nível Superior","keywords":"Open peer review; Cone beam computed tomography; Plant biology; Computed tomography; Medicine; Dentistry; Paediatric dentistry; Cone (formal languages); Medical physics; Physiology; Orthodontics; Radiology; Biology; Computer science","score_opus":0.14078004599601193,"score_gpt":0.4149324573728314,"score_spread":0.27415241137681945,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409340083","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006488109,0.9932132,0.000018197215,0.000010188388,0.0008484906,0.0030354068,0.0018567451,0.000028019707,0.00034096834],"genre_scores_gemma":[0.00041791398,0.9988972,0.000112229194,0.00007906264,0.000046503104,0.00010138579,0.00021735435,0.000042198277,0.000086183434],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9930856,0.0014854107,0.001854948,0.0007226078,0.0021780278,0.0006733665],"domain_scores_gemma":[0.9954242,0.0016096426,0.0009786615,0.0013279131,0.0005143908,0.00014515748],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0011724418,0.0005350009,0.0025898127,0.030391263,0.00009664563,0.0001392665,0.0017834166,0.00029438987,0.000115323346],"category_scores_gemma":[0.0016557853,0.00040029731,0.001996097,0.24661754,0.00029793064,0.0002399335,0.0011639894,0.0008358972,0.000029769928],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016135935,0.0006421918,0.048598073,0.19257891,0.0002594468,0.000051477975,0.0000022087943,0.0000012109529,1.07123384e-7,0.00009069429,0.017249849,0.7405097],"study_design_scores_gemma":[0.0009893008,0.00013575323,0.13955593,0.3400998,0.0011366565,0.000043452525,0.0000027290898,0.000007828021,0.0000033242857,0.000058110458,0.51747924,0.00048786745],"about_ca_topic_score_codex":0.000479535,"about_ca_topic_score_gemma":0.00004743247,"teacher_disagreement_score":0.7400218,"about_ca_system_score_codex":0.00024678855,"about_ca_system_score_gemma":0.00046475913,"threshold_uncertainty_score":0.9998449},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4409563158","doi":"10.46799/adv.v3i1.353","title":"Bibliometric Analysis of Social Media Trends in Hospital Marketing","year":2025,"lang":"en","type":"article","venue":"Advances In Social Humanities Research","topic":"Social Media in Health Education","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Social media; Bibliometrics; Social media marketing; Marketing; Sociology; Business; Library science; Computer science; World Wide Web","score_opus":0.19904606243499703,"score_gpt":0.5453019985949307,"score_spread":0.34625593615993366,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409563158","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.81912696,0.0034063926,0.0000033602732,0.0018667255,0.002080319,0.0004902088,0.000022944458,0.000037327194,0.17296576],"genre_scores_gemma":[0.9956525,0.0022785277,0.000058743994,0.000022328662,0.0009632284,0.00048830674,0.000028116203,0.000012224778,0.0004960101],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9939805,0.0024695564,0.0006779006,0.000359044,0.0015845075,0.00092844485],"domain_scores_gemma":[0.98520744,0.013756744,0.00018323079,0.00012538489,0.0006719613,0.0000552537],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.008505567,0.00012563536,0.0005314913,0.13151187,0.0012712891,0.000094338786,0.0006332521,0.00024985868,0.0008417751],"category_scores_gemma":[0.014773298,0.00015648616,0.00015435854,0.31441182,0.0018583742,0.0005931048,0.0001336295,0.0006696341,0.0000061711944],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007102513,0.0003298293,0.19170189,0.00010856175,0.000067083114,0.000003360435,0.2880435,0.000004154543,0.0000027546487,0.1000909,0.0007996542,0.4187773],"study_design_scores_gemma":[0.00024921997,0.000020333277,0.8050175,0.000034590637,0.000029475521,4.394239e-9,0.14140388,0.000008687778,0.0000035081252,0.0047419136,0.048370607,0.0001202994],"about_ca_topic_score_codex":0.00566892,"about_ca_topic_score_gemma":0.0778074,"teacher_disagreement_score":0.6133156,"about_ca_system_score_codex":0.0014049128,"about_ca_system_score_gemma":0.00061266095,"threshold_uncertainty_score":0.9935257},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4409595261","doi":"10.1016/j.respol.2025.105253","title":"Mobility and its effect on scientific recognition. A prosopographic analysis of Swiss biologists","year":2025,"lang":"en","type":"article","venue":"Research Policy","topic":"Interdisciplinary Research and Collaboration","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"Social Sciences and Humanities Research Council of Canada; Fonds de Recherche du Québec-Société et Culture; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung","keywords":"Data science; Computer science","score_opus":0.26054420895925856,"score_gpt":0.5727699577190036,"score_spread":0.31222574875974507,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409595261","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98704386,0.0002535268,0.000027312646,0.0038294669,0.000057900786,0.0007517228,0.00016387603,0.000015972684,0.007856381],"genre_scores_gemma":[0.9966846,0.000012815154,0.000012069804,0.000029850991,0.000041443283,0.00010179742,0.000020829337,0.0000031711868,0.003093399],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99400437,0.0022951886,0.00046084507,0.000726597,0.0020033752,0.00050961383],"domain_scores_gemma":[0.9918573,0.0040871184,0.00008540106,0.00074343174,0.003044955,0.00018182378],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.020428563,0.000107173655,0.00035245143,0.008652688,0.0005758831,0.0005401622,0.0006601457,0.00009516431,0.00011247344],"category_scores_gemma":[0.025095884,0.00007103527,0.00016209023,0.03135571,0.00094253535,0.00021094798,0.00048299177,0.0003101769,0.00011124357],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.006483391,0.0025740066,0.28612775,0.00060305663,0.0023695603,0.000042448435,0.0027577083,0.00030423177,0.1292265,0.17242454,0.047795057,0.34929174],"study_design_scores_gemma":[0.0012627728,0.0049313144,0.5342885,0.00020736891,0.000113817216,0.0000014757508,0.0010857465,0.018114425,0.059563246,0.377118,0.0029861215,0.00032717548],"about_ca_topic_score_codex":0.00010941045,"about_ca_topic_score_gemma":0.00053930195,"teacher_disagreement_score":0.34896457,"about_ca_system_score_codex":0.00011088205,"about_ca_system_score_gemma":0.0007067811,"threshold_uncertainty_score":0.9892333},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4409701221","doi":"10.52783/eel.v15i2.2919","title":"Bibliometric Analysis of Global Research on Job Content Plateau","year":2025,"lang":"en","type":"article","venue":"","topic":"Regional Development and Environment","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Plateau (mathematics); Regional science; Geography; Mathematics","score_opus":0.23910687143124706,"score_gpt":0.4491546924869163,"score_spread":0.21004782105566924,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409701221","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5991333,0.00006049357,0.00006962438,0.0019986322,0.000054290034,0.00006256608,0.0000014452554,0.0000078038465,0.39861184],"genre_scores_gemma":[0.9742674,0.0004257994,0.000088428926,0.00012728633,0.000009422817,0.000005105484,0.0000019290792,7.874443e-7,0.025073793],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.998699,0.00011634501,0.00013975607,0.00014029667,0.0007048512,0.00019975794],"domain_scores_gemma":[0.9995242,0.00022397657,0.000020642467,0.000101654216,0.00007501653,0.00005451928],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0012367007,0.00003986831,0.00012598894,0.030239118,0.00018018468,0.000028797633,0.00020094149,0.000048494523,0.0004827429],"category_scores_gemma":[0.00013806613,0.000031569274,0.00007857318,0.25981534,0.00019512027,0.000041799303,0.00005491971,0.00004959123,0.000058961206],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016134425,0.00005474901,0.6765516,0.0000016902443,0.00025245303,9.845394e-7,0.00014054458,0.000067725996,0.000013403815,0.30507812,0.010178343,0.007644289],"study_design_scores_gemma":[0.00007668509,0.000016041251,0.9739966,0.000005374714,0.000032657594,5.16053e-9,0.0011296574,0.000025695028,0.00003393818,0.0016732296,0.022977533,0.00003262124],"about_ca_topic_score_codex":0.0056299823,"about_ca_topic_score_gemma":0.002254752,"teacher_disagreement_score":0.37513414,"about_ca_system_score_codex":0.00028510488,"about_ca_system_score_gemma":0.0001031452,"threshold_uncertainty_score":0.9807523},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4409721297","doi":"10.2196/65844","title":"Exploring 97 Years of Aedes aegypti as the Vector for Dengue, Yellow Fever, Zika, and Chikungunya (Diptera: Culicidae): Scientometric Analysis","year":2025,"lang":"en","type":"article","venue":"Interactive Journal of Medical Research","topic":"Mosquito-borne diseases and control","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Chikungunya; Dengue fever; Aedes aegypti; Vector (molecular biology); Bibliometrics; Aedes; Zika virus; Geography; Biology; Library science; Virology; Ecology; Virus","score_opus":0.1147729831946801,"score_gpt":0.46460926003618136,"score_spread":0.34983627684150126,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409721297","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9841174,0.005502338,0.0011621298,0.0078115747,0.00033069937,0.00038284814,0.000007983995,0.000004054887,0.0006809619],"genre_scores_gemma":[0.99683636,0.0016458293,0.000079376456,0.0003324544,0.00032027817,0.000037454498,0.0000019335864,0.000010428228,0.0007358845],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9958862,0.00030221482,0.000591498,0.00023785398,0.0026033716,0.00037886834],"domain_scores_gemma":[0.99323493,0.004367283,0.0002140801,0.00026781773,0.0014517963,0.00046409055],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0045838333,0.00011696078,0.00060177024,0.0028369178,0.000121768986,0.00008609999,0.00050790806,0.000080425656,0.00043631264],"category_scores_gemma":[0.01958123,0.00007474835,0.00045913612,0.0036820783,0.00036519693,0.00020750688,0.00019553123,0.0009372679,0.0000063538446],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005457678,0.0012952841,0.03404998,0.00038981606,0.008153164,0.0004894516,0.0023423238,0.000019893938,0.0056881253,0.0014370902,0.0051512937,0.9355259],"study_design_scores_gemma":[0.009457144,0.0025181898,0.8992278,0.0025367576,0.0034130283,0.00012497348,0.009540411,0.009274305,0.0063981796,0.0007508403,0.056517243,0.00024112323],"about_ca_topic_score_codex":0.00024437023,"about_ca_topic_score_gemma":0.000029961395,"teacher_disagreement_score":0.9352848,"about_ca_system_score_codex":0.00016018306,"about_ca_system_score_gemma":0.0007553738,"threshold_uncertainty_score":0.98867726},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4409728141","doi":"10.5206/cjils-rcsib.v48i1.22390","title":"Towards Sustainable and Coordinated Methods for Estimating Open Access Costs at Canadian Higher Education Institutions","year":2025,"lang":"en","type":"article","venue":"Canadian Journal of Information and Library Science","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"University of British Columbia; University of Ottawa","funders":"","keywords":"Business; Environmental economics; Higher education; Economic growth; Economics","score_opus":0.354533325933165,"score_gpt":0.5916812547562789,"score_spread":0.23714792882311386,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409728141","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19362408,0.0038890508,0.12508462,0.085023195,0.0069793854,0.002560486,0.00020240877,0.000021174626,0.5826156],"genre_scores_gemma":[0.8996482,0.000050003666,0.0854501,0.004717996,0.000037573947,0.000014422666,0.000011429781,0.0000034466955,0.010066854],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982086,0.00006733631,0.0005585899,0.00015846184,0.00057595206,0.00043109216],"domain_scores_gemma":[0.99477494,0.00048444438,0.000292294,0.00018040411,0.0025563594,0.001711537],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["bibliometrics","scholarly_communication"],"category_scores_codex":[0.009494694,0.00007753376,0.000161258,0.040606193,0.0016601903,0.020929761,0.0029339592,0.00005433787,0.00026693047],"category_scores_gemma":[0.014024946,0.000058923015,0.000026713036,0.055569842,0.00043741643,0.027289229,0.0006101858,0.0001241365,0.0000030692931],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010296631,0.0000051991474,0.021980844,0.000019599962,0.000005287919,0.0000024598014,0.000255452,0.00010119572,0.0000077408395,0.22137366,0.07058112,0.68565714],"study_design_scores_gemma":[0.00029486822,0.000048325648,0.11048402,0.000037810318,0.0000042031247,0.000028891996,0.0009757962,0.01026248,0.00012960623,0.016093694,0.8615426,0.000097709075],"about_ca_topic_score_codex":0.06665031,"about_ca_topic_score_gemma":0.009360983,"teacher_disagreement_score":0.7909615,"about_ca_system_score_codex":0.0005924775,"about_ca_system_score_gemma":0.041599613,"threshold_uncertainty_score":0.9996395},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"incentives","study_design":"not_applicable","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W4409795104","doi":"10.61091/jcmcc127b-428","title":"Current status and bibliometric analysis of carbon emission reduction pathway research based on artificial intelligence and Web of science platforms","year":2025,"lang":"en","type":"article","venue":"Journal of Combinatorial Mathematics and Combinatorial Computing","topic":"Medical Research and Treatments","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"National Social Science Fund of China","keywords":"Web of science; Web intelligence; Computer science; Reduction (mathematics); Artificial intelligence; Data science; World Wide Web; Political science; Web service; Mathematics; MEDLINE; Web development","score_opus":0.07312378524088864,"score_gpt":0.40148547829589,"score_spread":0.32836169305500135,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409795104","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99623406,0.000957562,0.0003900154,0.000062491425,0.0017962988,0.00026081485,0.0000022240706,0.0000046618807,0.00029189175],"genre_scores_gemma":[0.99896663,0.00049581623,0.00040238808,0.0000017473745,0.00012251128,0.0000011237779,0.0000012194337,0.0000067956244,0.0000017457822],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9966136,0.00007891767,0.0009531461,0.00022735672,0.0017961888,0.00033078052],"domain_scores_gemma":[0.996302,0.0011850365,0.00051833014,0.00019953291,0.001411196,0.00038388596],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0046606646,0.00014585035,0.00075964245,0.023699582,0.00019802102,0.00006738276,0.00014926509,0.00009460514,0.0000029586831],"category_scores_gemma":[0.0046834485,0.00010537343,0.00009490554,0.030639572,0.0008064179,0.00007958282,0.00017413041,0.0005350891,5.218783e-8],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0032167686,0.006399017,0.025982121,0.0029753959,0.0010297147,0.000036570873,0.0013632884,0.00006653376,0.021326676,0.6984071,0.000027103179,0.23916969],"study_design_scores_gemma":[0.0068065943,0.009560647,0.015525813,0.0054039727,0.0012837508,0.000021247697,0.0014928773,0.25125822,0.06970827,0.63857764,0.00007475922,0.00028618283],"about_ca_topic_score_codex":0.000021817847,"about_ca_topic_score_gemma":1.8556875e-7,"teacher_disagreement_score":0.25119168,"about_ca_system_score_codex":0.00013084333,"about_ca_system_score_gemma":0.0011317312,"threshold_uncertainty_score":0.98996466},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4409888461","doi":"10.53766/rola/2024.19.2.04","title":"Análisis bibliométrico de trabajos especiales de grado de periodoncia de la Facultad de Odontología de la Universidad de Los Andes","year":2024,"lang":"es","type":"article","venue":"Revista Odontológica de Los Andes","topic":"Health and Medical Education","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Humanities; Political science; Philosophy","score_opus":0.03402133231895909,"score_gpt":0.45418865466364405,"score_spread":0.42016732234468496,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409888461","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.91434443,0.044685856,0.016612394,0.010493277,0.00031429745,0.0014433541,0.00051943446,0.00052592834,0.011061049],"genre_scores_gemma":[0.93116426,0.051858004,0.008386127,0.0040503214,0.0023657696,0.00021660296,0.000093160466,0.0001722579,0.0016935087],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.98574525,0.007339803,0.0013509762,0.001083108,0.00069008116,0.0037907586],"domain_scores_gemma":[0.9887897,0.0061610565,0.00047848016,0.00069508184,0.00029300342,0.0035827095],"candidate_categories":["metaepi_narrow","sts","research_integrity","insufficient_payload"],"consensus_categories":["research_integrity"],"category_scores_codex":[0.008955145,0.00087383616,0.0013236222,0.0019856726,0.0016574009,0.0009625522,0.0009381689,0.00226933,0.0014369375],"category_scores_gemma":[0.0064355074,0.0008289887,0.00059983874,0.003615559,0.0008960377,0.00045554768,0.00017597285,0.0030887134,0.00007902343],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.00094996503,0.0017214132,0.59922785,0.021699183,0.000533249,0.0026321006,0.053997476,0.00004454982,0.007546963,0.036592595,0.14478374,0.13027091],"study_design_scores_gemma":[0.0020873728,0.00038778607,0.41433883,0.007151787,0.00097545364,0.0008826759,0.014239866,0.0033064727,0.000481317,0.0019904338,0.5531,0.0010579982],"about_ca_topic_score_codex":0.007949529,"about_ca_topic_score_gemma":0.0011638553,"teacher_disagreement_score":0.40831628,"about_ca_system_score_codex":0.006467852,"about_ca_system_score_gemma":0.02331738,"threshold_uncertainty_score":0.9996423},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4410008454","doi":"10.55131/jphd/2025/230225","title":"Global bibliometric analysis of pharmacists’ competency exam and educational trends","year":2025,"lang":"en","type":"article","venue":"Journal of Public Health and Development","topic":"Pharmacy and Medical Practices","field":"Pharmacology, Toxicology and Pharmaceutics","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medical education; Library science; Medicine; Computer science","score_opus":0.268821173066603,"score_gpt":0.5509102021413718,"score_spread":0.28208902907476874,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410008454","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9145545,0.033152245,0.0018345305,0.041348122,0.0015955801,0.00012669775,0.000037286216,0.000011620006,0.0073394314],"genre_scores_gemma":[0.9789334,0.008592335,0.0020607319,0.009981391,0.000085291074,0.0000052413297,0.00001910268,0.0000028786574,0.00031964053],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977079,0.0004940541,0.0009823446,0.00015548665,0.00032294082,0.0003372911],"domain_scores_gemma":[0.99717766,0.0007772639,0.00068762986,0.0000579145,0.0003073057,0.0009922185],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.004177091,0.00012917738,0.0004804081,0.018621163,0.00020459291,0.000030762963,0.00017193303,0.00009348885,0.0007820163],"category_scores_gemma":[0.00065186573,0.00010663299,0.00006867376,0.029572727,0.0001487498,0.000269973,0.00007149528,0.00037381894,0.0000013257272],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010770293,0.0005371604,0.2701968,0.00016967942,0.0016741649,0.000006058956,0.0005750269,0.0000065420513,0.000025300878,0.0012401852,0.014952596,0.71050876],"study_design_scores_gemma":[0.0015214073,0.000083737104,0.46309152,0.000015545615,0.0002534045,0.00004853151,0.00015656318,0.00017800629,0.000039619128,0.0001617795,0.53437334,0.00007657298],"about_ca_topic_score_codex":0.000008863187,"about_ca_topic_score_gemma":0.00000584694,"teacher_disagreement_score":0.71043223,"about_ca_system_score_codex":0.00015299249,"about_ca_system_score_gemma":0.0038306804,"threshold_uncertainty_score":0.9925019},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410079009","doi":"10.70844/jmhrp.2025.2.1.39","title":"A Bibliometric Analysis on Mental Health Research Over the Past Two Decades","year":2025,"lang":"en","type":"article","venue":"","topic":"Health, Environment, Cognitive Aging","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Mental health; Bibliometrics; Psychology; Regional science; Geography; Library science; Psychiatry; Computer science","score_opus":0.06992464208212767,"score_gpt":0.42465163339026213,"score_spread":0.35472699130813445,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410079009","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8831502,0.00016600882,0.0015510916,0.016911348,0.000064318265,0.00072263775,0.000012824107,0.00004176923,0.097379774],"genre_scores_gemma":[0.9865276,0.00043496906,0.0002811062,0.008517083,0.000028967701,0.000049479684,0.000008878097,0.000011034863,0.004140889],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99621916,0.0007824227,0.0003139065,0.00070498855,0.0012075054,0.00077202456],"domain_scores_gemma":[0.9983318,0.00076133234,0.00006764474,0.0006248151,0.000008553401,0.00020586605],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.004741705,0.00015136244,0.00020696958,0.027825415,0.0009110596,0.00010394846,0.00047228968,0.00004428897,0.0052664136],"category_scores_gemma":[0.00017218768,0.000102053506,0.00010846759,0.2078656,0.0003828147,0.00015565273,0.00051030307,0.0004784191,0.0013126874],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021624754,0.00023475509,0.81686014,0.000009217545,0.00013662499,0.0000026821024,0.00037551086,0.0004930509,0.00025804035,0.0006111161,0.040088385,0.14090884],"study_design_scores_gemma":[0.00031053438,0.00007989637,0.95584285,0.000012317247,0.000022511198,4.5524473e-7,0.00037466607,0.0014538054,0.00019585426,0.00026185677,0.041350294,0.00009494171],"about_ca_topic_score_codex":0.0050085667,"about_ca_topic_score_gemma":0.0014681823,"teacher_disagreement_score":0.18004018,"about_ca_system_score_codex":0.0010295659,"about_ca_system_score_gemma":0.000041155246,"threshold_uncertainty_score":0.9994649},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4410192603","doi":"10.24083/apjhm.v20i1.3607","title":"Exploring COVID-19's Impact on Mental Health in the Workplace: A bibliometric analysis","year":2025,"lang":"en","type":"article","venue":"Asia Pacific Journal of Health Management","topic":"COVID-19 Pandemic Impacts","field":"Economics, Econometrics and Finance","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Collège Mérici","funders":"","keywords":"Coronavirus disease 2019 (COVID-19); Mental health; Publishing; Project commissioning; 2019-20 coronavirus outbreak; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Pandemic; Sociology; Psychology; Political science; Medicine; Virology; Psychiatry","score_opus":0.13612468748822934,"score_gpt":0.37059708273056047,"score_spread":0.23447239524233113,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410192603","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20444772,0.04420043,0.053285226,0.657368,0.0031730433,0.0035604262,0.00022257071,0.00008078976,0.03366179],"genre_scores_gemma":[0.96789074,0.016740277,0.00026453944,0.014737069,0.00004998016,0.000025495092,0.00001063975,0.000014928305,0.0002663077],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99621713,0.00031463313,0.0020790342,0.00038801928,0.00024318181,0.00075797556],"domain_scores_gemma":[0.9970483,0.00033515383,0.00167969,0.0005733333,0.000027503913,0.0003360145],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.012482332,0.00025620038,0.0010318742,0.086596996,0.0003237702,0.00016783818,0.00064427615,0.00004695156,0.000119488235],"category_scores_gemma":[0.00032557404,0.00020526214,0.0004298107,0.08818066,0.000038018254,0.00035314853,0.000085499436,0.0005067747,0.000048254216],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007886118,0.0016342363,0.40260485,0.0012792872,0.00428274,0.0001339057,0.013214189,0.028375436,2.486517e-7,0.057464413,0.27488124,0.21534082],"study_design_scores_gemma":[0.0022985232,0.00063546014,0.6752299,0.00019587738,0.00005251803,0.0000143863135,0.0065442827,0.0005950448,2.861403e-7,0.00424861,0.3099502,0.0002348895],"about_ca_topic_score_codex":0.0010504726,"about_ca_topic_score_gemma":0.00004979288,"teacher_disagreement_score":0.76344305,"about_ca_system_score_codex":0.0039239367,"about_ca_system_score_gemma":0.00043910724,"threshold_uncertainty_score":0.9998998},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4410215454","doi":"10.3389/fpsyt.2025.1435612","title":"A bibliometric analysis of the current state of research on family interventions for ASD","year":2025,"lang":"en","type":"review","venue":"Frontiers in Psychiatry","topic":"Family and Disability Support Research","field":"Psychology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Psychological intervention; Current (fluid); Psychology; State (computer science); Medicine; Psychiatry; Clinical psychology; Computer science; Engineering","score_opus":0.2290140537032326,"score_gpt":0.5535457878108615,"score_spread":0.3245317341076289,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410215454","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0003109603,0.97478724,0.0011846626,0.0000608816,0.015973069,0.0025440874,0.0018038274,0.000011440674,0.0033238526],"genre_scores_gemma":[0.0001773777,0.9959064,0.00065514894,0.000014779352,0.0000649617,0.0010044466,0.00019140038,0.000032012707,0.0019534659],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9941773,0.0017550095,0.0017950567,0.000787081,0.00084579556,0.000639801],"domain_scores_gemma":[0.99569947,0.0015104167,0.0005455822,0.0018108791,0.00034601477,0.00008762245],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0042969203,0.0003159382,0.002324618,0.18918177,0.000096920805,0.000024851897,0.0016478436,0.00025050878,0.0000863645],"category_scores_gemma":[0.00075498683,0.0002276877,0.0036558334,0.30622685,0.00051136554,0.000036742407,0.0002761714,0.0014199096,0.000009946372],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007919262,0.001319938,0.011551413,0.039843008,0.0019393961,2.4738304e-7,0.00005836496,0.000005117894,6.6342802e-9,0.00058547343,0.20072949,0.7438884],"study_design_scores_gemma":[0.0010117783,0.00064054073,0.0962723,0.052686322,0.00553979,3.6219393e-7,0.0012394456,0.000031781226,1.9678723e-7,0.009753288,0.83235955,0.0004646427],"about_ca_topic_score_codex":0.00012830929,"about_ca_topic_score_gemma":0.00014398103,"teacher_disagreement_score":0.7434237,"about_ca_system_score_codex":0.00020335036,"about_ca_system_score_gemma":0.0011702965,"threshold_uncertainty_score":0.9284835},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410250007","doi":"10.28991/hef-2025-06-01-015","title":"Global Trends in Agricultural Waste-Based Bioplastic Research: A Scientometric Review","year":2025,"lang":"en","type":"review","venue":"Journal of Human Earth and Future","topic":"Microplastics and Plastic Pollution","field":"Environmental Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bioplastic; Agriculture; Environmental science; Engineering; Waste management; Biology; Ecology","score_opus":0.049050726027897645,"score_gpt":0.3530761393913413,"score_spread":0.3040254133634437,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410250007","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00023211798,0.9966477,0.000010001953,0.000081376405,0.0004552563,0.00017063455,0.00007636806,0.0000031574093,0.0023233702],"genre_scores_gemma":[0.00031485627,0.99866813,0.00009169164,0.000040920633,0.00042055236,0.0000027859649,0.000025627654,0.0000055205,0.00042990802],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976568,0.00029414392,0.00072166475,0.00026947205,0.00067707873,0.00038084053],"domain_scores_gemma":[0.99897724,0.00018128725,0.00043761457,0.0001430943,0.00004033265,0.00022044808],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011671764,0.00026172036,0.00097448926,0.0011576093,0.0001691512,0.00008108277,0.0003383999,0.00019691962,0.0007848252],"category_scores_gemma":[0.00016623484,0.00015764937,0.00028401194,0.0056176065,0.00015701594,0.00009427587,0.00011851205,0.0006888288,0.000021675913],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000863197,0.00008761382,0.00011004802,0.00864983,0.000023514236,0.00007941027,0.000010876904,0.000017148306,0.0000022080765,0.00008539211,0.020930704,0.9699946],"study_design_scores_gemma":[0.0003319832,0.00026910828,0.0033053467,0.029423144,0.00025014378,0.00016852407,0.000014035568,0.0000060721695,2.1509628e-7,0.000021114527,0.96601224,0.00019809685],"about_ca_topic_score_codex":0.000013367688,"about_ca_topic_score_gemma":0.00012531744,"teacher_disagreement_score":0.96979654,"about_ca_system_score_codex":0.00021063643,"about_ca_system_score_gemma":0.00012935126,"threshold_uncertainty_score":0.8593285},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410250670","doi":"10.1145/3724154.3724351","title":"A Bibliometric Analysis and Visualization Regarding Language Impairment Pertinent to Autism Based on VOSviewer and CiteSpace from 2014 to 2022","year":2024,"lang":"en","type":"article","venue":"","topic":"Autism Spectrum Disorder Research","field":"Neuroscience","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Visualization; Autism; Computer science; Information visualization; Data visualization; Data science; Psychology; Artificial intelligence; Developmental psychology","score_opus":0.017055733424497035,"score_gpt":0.3346265221198983,"score_spread":0.3175707886954013,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410250670","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9346717,0.00081225025,0.04608855,0.01682623,0.0000795371,0.0005751001,0.000050947372,0.00019066683,0.00070502865],"genre_scores_gemma":[0.99543226,0.00016935756,0.0004065582,0.0017987607,0.0000195032,0.000056097524,0.000008318332,0.000027563327,0.0020816065],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99771947,0.00016820409,0.00019312423,0.0008472276,0.00071741355,0.0003545464],"domain_scores_gemma":[0.99889666,0.0004481466,0.000023667906,0.00032444016,0.000009967934,0.0002971427],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00043766416,0.00017273233,0.00021873503,0.056256883,0.0001238705,0.00051064545,0.00014987691,0.00004370756,0.0013406732],"category_scores_gemma":[0.0005230282,0.00014176914,0.00006492269,0.11644433,0.000024949855,0.00014066194,0.00023542401,0.00012261738,0.00017746272],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005283183,0.000919002,0.10536984,0.0004215564,0.00038888285,0.00062015484,0.012427532,0.003254162,0.7124292,0.024116684,0.04008962,0.099435],"study_design_scores_gemma":[0.0007405755,0.0010158935,0.40144074,0.00013470264,0.00019685907,0.0000046703503,0.00026633954,0.5118957,0.07454154,0.0003465293,0.008751854,0.00066457223],"about_ca_topic_score_codex":0.0008513736,"about_ca_topic_score_gemma":0.00018178676,"teacher_disagreement_score":0.6378877,"about_ca_system_score_codex":0.000101353435,"about_ca_system_score_gemma":0.000032934397,"threshold_uncertainty_score":0.9995722},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410290080","doi":"10.1097/md.0000000000042383","title":"Corneal esthesiometry between 2000 and 2024: A bibliometric and knowledge mapping analysis","year":2025,"lang":"en","type":"article","venue":"Medicine","topic":"Ocular Surface and Contact Lens","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Scopus; Contact lens; Optometry; Ophthalmology; Corneal disease; Web of science; Bibliometrics; Cornea; MEDLINE; Pathology; Meta-analysis; Library science","score_opus":0.026979645367440942,"score_gpt":0.31725373915109784,"score_spread":0.2902740937836569,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410290080","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9445979,0.04008774,0.0012017069,0.003832992,0.00010114378,0.00020464954,0.0000039475376,0.000046137215,0.009923763],"genre_scores_gemma":[0.98792946,0.0016523785,0.00020666975,0.0006091693,0.00017170742,0.0000057291136,0.000020964555,0.000010941328,0.009392995],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9988258,0.00004831767,0.0003098048,0.00034584774,0.00022092296,0.000249317],"domain_scores_gemma":[0.9990302,0.00030254512,0.000055555152,0.00029300363,0.000102475286,0.00021622398],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.000564669,0.0001740521,0.0007428842,0.03827815,0.00008262987,0.000017525148,0.000064245134,0.00011245303,0.00023754564],"category_scores_gemma":[0.00027718215,0.00013028839,0.00007035421,0.078495055,0.00014839691,0.00005011636,0.000074433294,0.0002196238,0.000015635296],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021660519,0.000031718562,0.8872281,0.00018276276,0.0008303062,0.00004023441,0.0003034045,1.03568475e-7,0.00048806076,0.000046378045,0.0013498629,0.10947741],"study_design_scores_gemma":[0.0016766245,0.00023384254,0.9535036,0.00028560267,0.002656075,0.000005637695,0.00039537923,0.0002518954,0.000130589,0.00006178502,0.04069334,0.000105608015],"about_ca_topic_score_codex":0.00020322844,"about_ca_topic_score_gemma":0.000025317426,"teacher_disagreement_score":0.1093718,"about_ca_system_score_codex":0.00004725021,"about_ca_system_score_gemma":0.0001015072,"threshold_uncertainty_score":0.97262216},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410370362","doi":"10.1186/s41687-025-00878-1","title":"Research trends among new investigators at ISOQOL: a bibliometric analysis from 2019 to 2023","year":2025,"lang":"en","type":"article","venue":"Journal of Patient-Reported Outcomes","topic":"Health and Medical Research Impacts","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Trinity Western University; Western University; University of Alberta; University of Victoria","funders":"","keywords":"Scopus; Thematic analysis; Descriptive statistics; Bibliometrics; Library science; Psychology; Medical education; Political science; Sociology; Social science; Medicine; MEDLINE; Computer science; Qualitative research; Statistics","score_opus":0.15360137381107414,"score_gpt":0.48497300802004606,"score_spread":0.3313716342089719,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410370362","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96478546,0.001072946,0.000173746,0.030889707,0.0006273789,0.0003279769,0.000017327273,0.000022020016,0.0020834538],"genre_scores_gemma":[0.9618466,0.00032970088,0.0011224259,0.006635378,0.00024531002,0.00000851134,0.000042225634,0.000026546348,0.029743308],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9922087,0.0003383356,0.0019413762,0.0003766932,0.0038677093,0.0012671931],"domain_scores_gemma":[0.98291945,0.001829775,0.0007412951,0.00075428595,0.001595548,0.012159671],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0039219116,0.00023973447,0.0012579921,0.16191633,0.00016448791,0.00007332993,0.0003206905,0.00024176884,0.0013483027],"category_scores_gemma":[0.098322794,0.00016367948,0.0006057199,0.20821646,0.00016160413,0.00021441166,0.0002406151,0.0012452747,0.00007773779],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028607674,0.00014648109,0.62371206,0.000024437319,0.0014940533,0.0006203744,0.00019477785,0.00001353321,0.00019065088,0.000002820191,0.32807916,0.045235593],"study_design_scores_gemma":[0.0023207327,0.0010294665,0.9406982,0.00027091405,0.00083952205,0.0000194695,0.0002479333,0.000040348048,0.0006500383,0.000104708044,0.053663917,0.00011477193],"about_ca_topic_score_codex":0.0056202994,"about_ca_topic_score_gemma":0.00047738117,"teacher_disagreement_score":0.31698614,"about_ca_system_score_codex":0.0009872972,"about_ca_system_score_gemma":0.0070273327,"threshold_uncertainty_score":0.9995646},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410380897","doi":"10.3389/fnut.2025.1550142","title":"Nutritional composition and health benefits of peas—a bibliometric research","year":2025,"lang":"en","type":"review","venue":"Frontiers in Nutrition","topic":"Phytase and its Applications","field":"Agricultural and Biological Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Fundação para a Ciência e a Tecnologia","keywords":"Composition (language); Health benefits; Environmental health; Food science; Medicine; Chemistry; Traditional medicine","score_opus":0.1392033794840654,"score_gpt":0.3898396001077248,"score_spread":0.2506362206236594,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410380897","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00019202668,0.9968542,0.000019797104,0.00040954744,0.000094778814,0.0011341161,0.0011047765,0.000014806759,0.00017594773],"genre_scores_gemma":[0.0001992604,0.9956784,0.0006894895,0.00002608004,0.00017295228,0.00035447898,0.0028580506,0.0000011248914,0.000020165859],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984154,0.00024756233,0.00046335373,0.00033976923,0.0002921068,0.00024179499],"domain_scores_gemma":[0.9992985,0.00026375186,0.00016701545,0.000065670414,0.00013439846,0.00007063615],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0006613007,0.00014347483,0.00069948845,0.0046082693,0.00017095233,0.000036683356,0.0001974726,0.0002000306,0.0000137241095],"category_scores_gemma":[0.00002112182,0.00007216396,0.000119304175,0.02897377,0.000081056045,0.000093680464,0.000065662665,0.00028496925,0.0000024470314],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004527527,0.00039227365,0.00006235931,0.012686541,0.000009651152,2.429699e-7,0.0000031761651,5.2746895e-8,0.000004556692,0.00070923864,0.03574598,0.9503814],"study_design_scores_gemma":[0.00022822953,0.00017580176,0.0020109275,0.036000382,0.000033436936,0.000005821888,0.000051157112,0.000004773334,0.000006211938,0.003463617,0.9578396,0.0001800338],"about_ca_topic_score_codex":0.000092749055,"about_ca_topic_score_gemma":0.000024594168,"teacher_disagreement_score":0.9502014,"about_ca_system_score_codex":0.00013254551,"about_ca_system_score_gemma":0.00004920855,"threshold_uncertainty_score":0.99166584},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410450843","doi":"10.3389/fphar.2025.1513318","title":"Global research trends and hotspots in aspirin studies (2014–2024): a bibliometric perspective","year":2025,"lang":"en","type":"review","venue":"Frontiers in Pharmacology","topic":"Inflammatory mediators and NSAID effects","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Natural Science Foundation of Beijing Municipality","keywords":"Aspirin; China; Bibliometrics; Web of science; Medicine; Library science; Geography; Computer science; Pathology; Internal medicine; Meta-analysis","score_opus":0.09184802064632555,"score_gpt":0.5152288673926663,"score_spread":0.4233808467463407,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410450843","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009828774,0.985811,0.0000043540867,0.00024754516,0.0071582985,0.0015290614,0.00005359809,0.000035161134,0.0041781086],"genre_scores_gemma":[0.00018016735,0.99556845,0.0002341763,0.00013544693,0.0005297383,0.0007553104,0.00001808347,0.000035353536,0.0025432792],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9953488,0.001513699,0.00076620263,0.0009942129,0.00040466085,0.0009724292],"domain_scores_gemma":[0.99811965,0.0008826321,0.0001740184,0.0003055906,0.00028164766,0.00023648104],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0025750338,0.00053547806,0.0033164297,0.0918145,0.00008307402,0.000020821906,0.0003130525,0.00066596316,0.00011547215],"category_scores_gemma":[0.0007737319,0.0004529132,0.00025499286,0.08250202,0.0005501265,0.00007759473,0.00038542246,0.0016518964,0.000021566444],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022892779,0.00015649416,0.0059067686,0.016212665,0.0005827538,0.0015535371,0.00018849618,2.8667162e-7,2.4435747e-7,0.000054602304,0.30216447,0.67295074],"study_design_scores_gemma":[0.0039933575,0.000687892,0.0051637827,0.009363097,0.0013678898,0.0001305483,0.0008616629,0.000025156502,0.0000014769078,0.00041753193,0.9775199,0.00046769145],"about_ca_topic_score_codex":0.0000902643,"about_ca_topic_score_gemma":0.00006568921,"teacher_disagreement_score":0.67535543,"about_ca_system_score_codex":0.0029957935,"about_ca_system_score_gemma":0.0011636117,"threshold_uncertainty_score":0.9997923},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410464964","doi":"10.1007/s12671-025-02592-x","title":"Half a Century of Research for Mind–Body Interventions: A Scientometric Analysis","year":2025,"lang":"en","type":"article","venue":"Mindfulness","topic":"Mindfulness and Compassion Interventions","field":"Psychology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital","funders":"Université de Genève","keywords":"Psychology; Psychological intervention; Public health; Mindfulness; Psychotherapist; Applied psychology; Psychiatry; Medicine; Nursing","score_opus":0.1269358898474106,"score_gpt":0.49042849315267023,"score_spread":0.3634926033052596,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410464964","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2681663,0.001323799,0.709287,0.00069258164,0.002179139,0.0009174873,0.0002444082,0.00003491451,0.01715435],"genre_scores_gemma":[0.9599285,1.1981194e-7,0.00016196858,0.000016984957,0.000048035854,0.008416089,0.000047984962,0.000012716797,0.031367607],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9971394,0.00040843143,0.0008491351,0.00060932577,0.00044954297,0.00054415606],"domain_scores_gemma":[0.99720603,0.0007718463,0.00021608068,0.0008385821,0.0008634059,0.00010404547],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0025065355,0.00016621321,0.00050571683,0.010017308,0.00033958638,0.00007892409,0.00074266945,0.00016586114,0.019144174],"category_scores_gemma":[0.00047617976,0.00015418735,0.0012095544,0.018555835,0.00028178032,0.000076961376,0.00021334157,0.0002582428,0.000110529276],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037697123,0.0032984565,0.0030359463,0.0005894017,0.0029022885,0.00000925765,0.00043229567,0.000020317859,0.000018687566,0.59895015,0.052248422,0.33811784],"study_design_scores_gemma":[0.012085962,0.002013397,0.09842465,0.004356095,0.0077147186,0.000020070705,0.034433685,0.0037931772,0.227369,0.14984503,0.45820147,0.0017427609],"about_ca_topic_score_codex":0.00029363393,"about_ca_topic_score_gemma":0.0001342967,"teacher_disagreement_score":0.70912504,"about_ca_system_score_codex":0.000088803536,"about_ca_system_score_gemma":0.00009049932,"threshold_uncertainty_score":0.98175246},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410524024","doi":"10.3389/fneur.2025.1593987","title":"Status and trends in transcranial magnetic stimulation research: a bibliometric analysis","year":2025,"lang":"en","type":"review","venue":"Frontiers in Neurology","topic":"Transcranial Magnetic Stimulation Studies","field":"Neuroscience","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Southwest Medical University","keywords":"Transcranial magnetic stimulation; Neuroscience; Psychology; Bibliometrics; Medicine; Physical medicine and rehabilitation; Stimulation; Computer science; Library science","score_opus":0.1349897759429762,"score_gpt":0.41416819687550155,"score_spread":0.2791784209325253,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410524024","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023247208,0.9941429,0.00042671742,0.00015306499,0.00085111934,0.0009604569,0.000104184786,0.00005362144,0.0009832012],"genre_scores_gemma":[0.016700108,0.9823383,0.0002660873,0.00010276477,0.00004331195,0.00019400108,0.000022521142,0.000037257665,0.0002956634],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99305993,0.0028020015,0.0010966124,0.0014686384,0.00059222354,0.0009806205],"domain_scores_gemma":[0.9959825,0.003215017,0.0001547878,0.00047016333,0.000054196484,0.00012328385],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00074178545,0.0004774486,0.0023980206,0.31110212,0.0001209117,0.00007547286,0.00045668767,0.0005376932,0.00011068161],"category_scores_gemma":[0.001336975,0.00047303998,0.00033330132,0.34382284,0.00054079556,0.00010816415,0.00013511494,0.0013012469,0.0000055613414],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015946377,0.000110357614,0.006202079,0.0012043327,0.000036493344,0.00011007706,0.00013755145,0.00033994197,0.0000031331824,0.00006082936,0.00020827103,0.9914275],"study_design_scores_gemma":[0.0044163642,0.0015301293,0.3653053,0.00068784854,0.0023908357,0.000016702976,0.000015878077,0.0053044683,3.5824084e-7,0.0011021674,0.61831105,0.0009188855],"about_ca_topic_score_codex":0.00017607764,"about_ca_topic_score_gemma":0.00023563723,"teacher_disagreement_score":0.99050856,"about_ca_system_score_codex":0.000089890025,"about_ca_system_score_gemma":0.00012222478,"threshold_uncertainty_score":0.99977213},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410564409","doi":"10.5539/jel.v14n5p209","title":"Trends in Massive Open Online Courses (MOOCs) Research Over the Past Ten Years (2015–2024): A Bibliometric Analysis","year":2025,"lang":"en","type":"article","venue":"Journal of Education and Learning","topic":"Online Learning and Analytics","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Trend analysis; Mathematics education; Psychology; Statistics; Mathematics","score_opus":0.04626387669062402,"score_gpt":0.44944928039818255,"score_spread":0.4031854037075585,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410564409","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94899017,0.0019618873,0.0022554519,0.044192903,0.00048158263,0.000057971574,8.285022e-7,0.000008802331,0.0020503872],"genre_scores_gemma":[0.97744715,0.00026876715,0.0026882696,0.0002501538,0.00017352677,0.0000014770012,0.0000030684848,0.000003967167,0.019163603],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99844486,0.00052450213,0.0003233087,0.00016503182,0.00036501675,0.00017728895],"domain_scores_gemma":[0.99856263,0.0005279553,0.0002598474,0.00018924144,0.00038620227,0.00007410442],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0027634138,0.00006998503,0.00021019776,0.049616754,0.00018955022,0.0006570354,0.0007811871,0.00004425325,0.00006853036],"category_scores_gemma":[0.0007037227,0.00004960361,0.00008093733,0.10356398,0.00004780774,0.00034466674,0.0002410774,0.0010374783,0.0000023954992],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001176404,0.0003783772,0.43657628,0.00000814754,0.00018917845,0.000009725063,0.001729845,0.0061510704,0.000013975831,0.0011832684,0.018735468,0.5350129],"study_design_scores_gemma":[0.00023990663,0.00008922126,0.92737,0.00007227351,0.00006429686,0.000007329477,0.0031492976,0.023488889,6.760836e-7,0.00032909424,0.045131683,0.00005734254],"about_ca_topic_score_codex":0.00018826644,"about_ca_topic_score_gemma":0.000051114195,"teacher_disagreement_score":0.53495556,"about_ca_system_score_codex":0.00006794299,"about_ca_system_score_gemma":0.00044317252,"threshold_uncertainty_score":0.96115506},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410802186","doi":"10.1117/12.3053835","title":"A bibliometric review of machine learning applications in multidomain operations: a decade of progress","year":2025,"lang":"en","type":"review","venue":"","topic":"Machine Learning and Data Classification","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Computer science; Data science; Artificial intelligence","score_opus":0.04171308528816181,"score_gpt":0.38820070579400284,"score_spread":0.34648762050584103,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410802186","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[6.193722e-8,0.84772444,0.14996608,0.00009468093,0.00002102891,0.0013874703,0.00002597576,0.00006211429,0.0007181396],"genre_scores_gemma":[0.0000042937527,0.94010484,0.05824615,0.000023201497,0.000009104351,0.0007689975,0.00045777683,0.0000105617555,0.00037508985],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9970064,0.00062186667,0.0013294588,0.00053355715,0.00033701234,0.00017170458],"domain_scores_gemma":[0.9973605,0.0007561006,0.000613072,0.0010131819,0.00020778384,0.00004933339],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0014015174,0.000257607,0.0013777015,0.025023097,0.00005721281,0.000046996207,0.0014686298,0.00014362978,0.000021915372],"category_scores_gemma":[0.0012353632,0.00019523846,0.00024066973,0.09911106,0.000051614865,0.00019777218,0.00038827144,0.00047395617,0.000010096312],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[1.7210886e-7,0.00009722301,0.00012386267,0.12699251,0.000016081085,2.6396992e-7,0.000008837762,0.00000552553,1.1624047e-7,0.0108601935,0.000047772854,0.86184746],"study_design_scores_gemma":[0.000106258056,0.00003109612,0.00008247421,0.09640069,0.00010854252,0.0000042906954,0.0000024378212,0.005893141,0.0000010967487,0.00001854577,0.8971611,0.0001903163],"about_ca_topic_score_codex":0.000122010184,"about_ca_topic_score_gemma":0.000019773526,"teacher_disagreement_score":0.8971133,"about_ca_system_score_codex":0.00004834704,"about_ca_system_score_gemma":0.00046320327,"threshold_uncertainty_score":0.9860274},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410834711","doi":"10.17576/jkukm-2025-37(3)-13","title":"The Relationship between Walkability and The Built Environment: The Bibliometric Analysis","year":2025,"lang":"en","type":"article","venue":"Jurnal Kejuruteraan","topic":"Urban Transport and Accessibility","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Cancer Institute; Universiti Malaya","keywords":"Walkability; Computer science; Built environment; Civil engineering; Engineering","score_opus":0.03772266208353588,"score_gpt":0.3314921285332184,"score_spread":0.29376946644968255,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410834711","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96892464,0.001810094,0.0006956038,0.021008654,0.00011048586,0.00032213674,0.000010126274,0.00002400205,0.007094268],"genre_scores_gemma":[0.9973583,0.00023840868,0.0000189657,0.00019409001,0.00016291626,0.000021681235,0.0000036676965,0.000003287473,0.0019986783],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99792093,0.00070728245,0.00037676533,0.0002485073,0.00047169326,0.0002748464],"domain_scores_gemma":[0.99367034,0.0055727586,0.00013751531,0.000505636,0.000041348732,0.00007241226],"candidate_categories":["bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.004603775,0.000108657354,0.00019832264,0.0025203638,0.0032913224,0.00052684825,0.0007131544,0.000077223885,0.00005218227],"category_scores_gemma":[0.00070171495,0.00004785847,0.00023106512,0.038382743,0.001497115,0.00022410773,0.00007636109,0.0003570454,0.0000072897537],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000020795358,0.0000131864035,0.9730968,0.0000032742917,0.00014474987,2.4595047e-7,0.0009985036,0.0000028736458,3.183348e-7,0.011397598,0.00013670942,0.014184937],"study_design_scores_gemma":[0.00023390392,0.000004460317,0.9416322,0.000003495664,0.00042313206,3.5446924e-8,0.0002870205,0.000029664701,0.0000021596468,0.03214438,0.025182953,0.000056583813],"about_ca_topic_score_codex":0.0019355185,"about_ca_topic_score_gemma":0.0019896927,"teacher_disagreement_score":0.03586238,"about_ca_system_score_codex":0.000069866845,"about_ca_system_score_gemma":0.000075793774,"threshold_uncertainty_score":0.9980063},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4410942588","doi":"10.1016/j.rser.2025.115871","title":"Energy communities in social sciences: A bibliometric analysis and systematic literature review","year":2025,"lang":"en","type":"article","venue":"Renewable and Sustainable Energy Reviews","topic":"Sustainability and Climate Change Governance","field":"Environmental Science","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Social science; Systematic review; Sociology; Management science; Regional science; Political science; Library science; Computer science; MEDLINE; Engineering","score_opus":0.02064331229408661,"score_gpt":0.27972611802999897,"score_spread":0.25908280573591236,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410942588","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0087309405,0.97333306,0.0006998132,0.0005652938,0.000020931615,0.00042468167,0.0000035288092,0.000021567092,0.016200213],"genre_scores_gemma":[0.060807757,0.9201711,0.000068764326,0.0027274424,0.00001326569,0.00035361075,0.00001549354,0.0000073971796,0.015835168],"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","domain_scores_codex":[0.9974501,0.0006591591,0.00063050084,0.00040699207,0.00024994736,0.0006033019],"domain_scores_gemma":[0.9990749,0.00022850002,0.00022647466,0.0003578124,0.00004798036,0.00006437377],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.002873309,0.00025537788,0.0010612765,0.006968247,0.00061568106,0.00021590047,0.00033679642,0.000109828616,0.00013720448],"category_scores_gemma":[0.0002953651,0.00019821788,0.00014702068,0.11591927,0.00035230105,0.0004709843,0.00052745256,0.00010331807,9.508612e-7],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000057189707,0.00056322786,0.0715912,0.75721395,0.00044004188,0.00025207837,0.0065721883,0.001175732,0.00003780312,0.07046466,0.042949703,0.048682198],"study_design_scores_gemma":[0.00070334325,0.00014161812,0.004289345,0.05077028,0.0013161934,0.000029082514,0.022174891,0.0022099563,0.000036229605,0.01388272,0.9034649,0.0009814261],"about_ca_topic_score_codex":0.010678491,"about_ca_topic_score_gemma":0.004419819,"teacher_disagreement_score":0.86051524,"about_ca_system_score_codex":0.0002929525,"about_ca_system_score_gemma":0.000040948285,"threshold_uncertainty_score":0.9959095},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4411061502","doi":"10.21664/2238-8869.2025v14i2.p168-184","title":"Bibliometric Analysis of Institutional Theory Research","year":2025,"lang":"en","type":"article","venue":"Fronteiras Journal of Social Technological and Environmental Science","topic":"Business and Economic Development","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Regional science; Sociology","score_opus":0.02900312400427774,"score_gpt":0.2856799590391748,"score_spread":0.2566768350348971,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411061502","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9918861,0.00028133066,0.0016760259,0.00033421608,0.00006708647,0.000055318556,0.000005889966,0.0000057310494,0.005688328],"genre_scores_gemma":[0.9986871,0.00022748948,0.00084468914,0.000051555853,0.000007637891,0.0000023498712,7.476732e-7,0.0000015092808,0.00017695193],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99851805,0.00004425055,0.00037454016,0.00022049535,0.0005858833,0.00025677416],"domain_scores_gemma":[0.9995895,0.00007890055,0.00016101837,0.00009012911,0.000014598306,0.000065867665],"candidate_categories":["bibliometrics","sts","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.002977121,0.000084288855,0.00026220744,0.0131181935,0.00047728096,0.000032759533,0.000550717,0.000092039845,0.00092296506],"category_scores_gemma":[0.0001831493,0.00006109958,0.00008481431,0.049286406,0.006180253,0.00030435497,0.00065579783,0.00023169967,0.000010482107],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000065255124,0.0003827539,0.43031627,0.000005722702,0.00016865785,0.000010452379,0.00021964735,0.00028347058,0.038184155,0.018041095,0.0003413847,0.5119811],"study_design_scores_gemma":[0.00015876116,0.000046809266,0.9874077,0.0000070849387,0.00005113277,0.000004176303,0.00057709374,0.00018429398,0.0013171745,0.0063790204,0.0037985255,0.00006826609],"about_ca_topic_score_codex":0.000019393887,"about_ca_topic_score_gemma":0.0000031406425,"teacher_disagreement_score":0.5570914,"about_ca_system_score_codex":0.00039419832,"about_ca_system_score_gemma":0.00003542697,"threshold_uncertainty_score":0.99999034},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4411091216","doi":"10.51982/bagimli.1675447","title":"Mapping research on online gambling: a bibliometric analysis","year":2025,"lang":"en","type":"article","venue":"Journal of Dependence","topic":"Gambling Behavior and Treatments","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Data science; Computer science; Psychology","score_opus":0.4185691947598688,"score_gpt":0.5610414978971526,"score_spread":0.14247230313728382,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411091216","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99091125,0.0011500451,0.0012244566,0.0004116736,0.0006631961,0.000077226585,0.0000071683708,0.000012535721,0.0055424627],"genre_scores_gemma":[0.9956182,0.000112964306,0.00064592686,0.00008801616,0.00008731571,0.0000034445764,0.000002239901,0.000008113493,0.0034337484],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99771947,0.00026459724,0.0005383658,0.00024146917,0.0008894004,0.00034669272],"domain_scores_gemma":[0.9979633,0.0005960779,0.00024545303,0.00038304174,0.0006926953,0.00011941943],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0018038712,0.00011605425,0.0003467141,0.16158749,0.00012446805,0.00008472244,0.00048448128,0.00011902724,0.00037280182],"category_scores_gemma":[0.0003500668,0.00009277674,0.00028095563,0.18465361,0.00004887697,0.00010252995,0.000058518508,0.0007557169,0.00009034319],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029160312,0.002364359,0.9204344,0.000014144712,0.002308877,0.0013855587,0.0005146535,0.00018029174,0.0006711976,0.0009332568,0.0024268702,0.06847476],"study_design_scores_gemma":[0.00094649824,0.00035944252,0.9955214,0.000092141316,0.00034547274,0.000052355223,0.000854968,0.00002710936,0.00022681791,0.00043047103,0.0010691272,0.000074141346],"about_ca_topic_score_codex":0.000113587084,"about_ca_topic_score_gemma":0.000024397747,"teacher_disagreement_score":0.07508703,"about_ca_system_score_codex":0.00012933623,"about_ca_system_score_gemma":0.00011547399,"threshold_uncertainty_score":0.8479152},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4411174920","doi":"10.1007/s11192-025-05339-6","title":"Mapping the unseen in practice: comparing latent Dirichlet allocation and BERTopic for navigating topic spaces","year":2025,"lang":"en","type":"article","venue":"Scientometrics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung","keywords":"Latent Dirichlet allocation; Topic model; Computer science; Data science; Information retrieval","score_opus":0.1049383044834109,"score_gpt":0.4485186821781035,"score_spread":0.3435803776946926,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411174920","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.901986,0.0017241986,0.05515245,0.029150039,0.0005224028,0.00046123247,8.239328e-7,0.000030804626,0.0109720575],"genre_scores_gemma":[0.98817855,0.000081456885,0.010442397,0.00029005884,0.00005612604,0.000015615866,0.0000025232155,0.0000016463445,0.0009315995],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.998965,0.00018203023,0.00016934107,0.00017305724,0.0003374914,0.00017306051],"domain_scores_gemma":[0.99741244,0.0021946859,0.000085160034,0.00007353548,0.00020068146,0.000033466615],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0039688884,0.00004864371,0.000099797,0.0005892284,0.0005238798,0.0002593134,0.00017033644,0.000030310504,0.0000024543497],"category_scores_gemma":[0.0041031595,0.00003823644,0.000029575644,0.008232093,0.00009853038,0.0001724799,0.0000670994,0.00009618097,8.5875706e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000460698,0.000053738055,0.51764363,0.00003339139,0.00003704091,6.158145e-7,0.020085895,0.0012738958,0.0000446103,0.12967089,0.00014692578,0.33100474],"study_design_scores_gemma":[0.00057632884,0.000020733973,0.7455396,0.00012441327,0.000062318744,4.968405e-7,0.020281563,0.06776394,0.00004457236,0.023427872,0.14196102,0.0001971257],"about_ca_topic_score_codex":0.0010982257,"about_ca_topic_score_gemma":0.0006284598,"teacher_disagreement_score":0.33080763,"about_ca_system_score_codex":0.00010129224,"about_ca_system_score_gemma":0.00008501285,"threshold_uncertainty_score":0.49121615},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4411237045","doi":"10.1001/jamanetworkopen.2025.15160","title":"Citation Contamination by Paper Mill Articles in Systematic Reviews of the Life Sciences","year":2025,"lang":"en","type":"article","venue":"JAMA Network Open","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Citation; Mill; Contamination; Environmental science; History; Library science; Biology; Computer science; Archaeology; Ecology","score_opus":0.04692872264228141,"score_gpt":0.34333756742881316,"score_spread":0.29640884478653173,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411237045","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.34806147,0.034032505,0.001431447,0.12230541,0.0020363366,0.011568502,0.0000110972305,0.000039555667,0.48051366],"genre_scores_gemma":[0.9929544,0.0012530083,0.00017832147,0.0035165336,0.00006905095,0.0000833656,0.0000013544242,0.0000017054622,0.0019422965],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99677396,0.0022501322,0.00044270762,0.0001223887,0.0002480244,0.0001627958],"domain_scores_gemma":[0.9983222,0.0012324941,0.0002487962,0.000103218044,0.00006986894,0.00002343105],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013764938,0.00005425756,0.00024167828,0.000031233627,0.00036047498,0.00010055439,0.00074264355,0.00018857587,0.00007892936],"category_scores_gemma":[0.005715516,0.000034799185,0.000038780887,0.0010048229,0.00021072537,0.00044964804,0.000094195486,0.0002651085,0.00001051773],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003443654,0.00007124401,0.07739233,0.0017119398,0.00003596999,3.1485243e-7,0.028164612,0.00031333513,0.00015347976,0.4902994,0.39129546,0.010527474],"study_design_scores_gemma":[0.002070797,0.00011329912,0.1846403,0.05543144,0.0002710051,7.639561e-7,0.070088685,0.008295704,0.00045346707,0.19685811,0.48103952,0.00073688745],"about_ca_topic_score_codex":0.0027908331,"about_ca_topic_score_gemma":0.003497158,"teacher_disagreement_score":0.6448929,"about_ca_system_score_codex":0.000047956953,"about_ca_system_score_gemma":0.00016606718,"threshold_uncertainty_score":0.68424195},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity","bibliometrics"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W4411264090","doi":"10.2478/fon-2025-0023","title":"Trend in research and hotspot on exercise therapy at home for stroke patients: a bibliometric analysis","year":2025,"lang":"en","type":"article","venue":"Frontiers of Nursing","topic":"Stroke Rehabilitation and Recovery","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Hotspot (geology); Physical medicine and rehabilitation; Medicine; Psychology; Physical therapy; Gerontology","score_opus":0.06730920789804944,"score_gpt":0.40348765045396373,"score_spread":0.3361784425559143,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411264090","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99164546,0.006067759,0.00024068968,0.0005297769,0.00028941742,0.0005196488,0.00003477169,0.000008209942,0.0006642723],"genre_scores_gemma":[0.99501026,0.0009824972,0.0023166446,0.000028747436,0.000014957093,0.000026206533,0.000033847544,0.000008273929,0.0015785862],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9990187,0.00004366185,0.00022562235,0.0002259572,0.00028152598,0.00020453145],"domain_scores_gemma":[0.9993043,0.00029487308,0.000047231384,0.00017122246,0.00012232886,0.000060056605],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00034477087,0.00007932976,0.00037966997,0.081902005,0.000060823404,0.000013151668,0.00004959886,0.00007828612,0.0000073465753],"category_scores_gemma":[0.0001643446,0.000066406676,0.00013635028,0.042406876,0.0001487129,0.000049213842,0.000009000277,0.00010097742,2.591604e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0025909476,0.0007669359,0.65165645,0.000053298663,0.00019394173,4.830493e-7,0.0005501473,0.0000095183805,0.00029060245,0.000016121572,0.02381908,0.32005247],"study_design_scores_gemma":[0.0035534778,0.0009280663,0.9915574,0.00032800774,0.00012259083,9.7610744e-8,0.0011570762,0.0003059686,0.0007183081,0.00021651882,0.0010584346,0.000054043267],"about_ca_topic_score_codex":0.000012645238,"about_ca_topic_score_gemma":0.0000056871872,"teacher_disagreement_score":0.33990097,"about_ca_system_score_codex":0.00026881119,"about_ca_system_score_gemma":0.000028393533,"threshold_uncertainty_score":0.97794724},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4411375260","doi":"10.18280/ijsdp.200539","title":"Trend Transformation in Population Administration in Indonesia: A Bibliometric Analysis 2012-2024","year":2025,"lang":"en","type":"article","venue":"International Journal of Sustainable Development and Planning","topic":"Economic Growth and Fiscal Policies","field":"Economics, Econometrics and Finance","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Transformation (genetics); Administration (probate law); Population; Geography; Regional science; Political science; Medicine; Environmental health; Chemistry","score_opus":0.01892994335484075,"score_gpt":0.2608626863486201,"score_spread":0.24193274299377934,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411375260","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98935795,0.0012510233,0.0020868722,0.0004549346,0.00021398485,0.000062303705,0.0000047367016,0.0000026887099,0.006565517],"genre_scores_gemma":[0.99918383,0.00008267066,0.00016992183,0.00006313492,0.00003037525,0.0000044768094,0.000036439924,0.0000027410306,0.0004264208],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99881184,0.000010101276,0.00087788195,0.00010122527,0.00005125298,0.00014767765],"domain_scores_gemma":[0.99953,0.000052710184,0.00030846577,0.000029788354,0.000048652913,0.000030356994],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00070507906,0.00007563813,0.00024530033,0.05426482,0.00003206422,0.00011438607,0.00011355981,0.00006449395,0.000019310399],"category_scores_gemma":[0.00006494099,0.00008679836,0.000046332916,0.011236803,0.000009344634,0.00078119943,0.000018085879,0.00012856092,0.0000010063726],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007820824,0.00003879894,0.96190405,0.000025002422,0.0001390736,0.00006784862,0.0023459245,0.00096324057,5.3339403e-7,0.030851003,0.000060753024,0.0035255752],"study_design_scores_gemma":[0.0006514065,0.000015055874,0.98843724,0.00003974481,0.000008721989,0.0000046861287,0.0024215665,0.0008736166,0.000014866779,0.0048311865,0.0026215352,0.00008040318],"about_ca_topic_score_codex":0.00015264009,"about_ca_topic_score_gemma":0.00012120328,"teacher_disagreement_score":0.04302802,"about_ca_system_score_codex":0.0002505991,"about_ca_system_score_gemma":0.000048908725,"threshold_uncertainty_score":0.9564543},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4411436385","doi":"10.1108/bfj-09-2024-0944","title":"Technological advancements and challenges in food supply chain management: a scientometric analysis","year":2025,"lang":"en","type":"article","venue":"British Food Journal","topic":"Food Waste Reduction and Sustainability","field":"Agricultural and Biological Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University Canada West","funders":"","keywords":"Perishability; Supply chain; Supply chain management; Originality; Sustainability; Business; Food security; Big data; Analytics; Knowledge management; Marketing; Process management; Computer science; Data science; Qualitative research; Agriculture","score_opus":0.026327979270956182,"score_gpt":0.23975389113399584,"score_spread":0.21342591186303966,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411436385","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9737554,0.020390494,0.000023477827,0.0039913263,0.00008162555,0.0001949005,0.0000125566385,0.000036621732,0.0015135506],"genre_scores_gemma":[0.98523605,0.014367257,0.00012563766,0.000060305058,0.000028011853,0.000017065782,0.0000036337951,4.936712e-7,0.00016156262],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99863726,0.00010641034,0.0002931342,0.00035808858,0.00025376404,0.00035132567],"domain_scores_gemma":[0.9996591,0.000045949495,0.00006884306,0.000052838608,0.00007582127,0.000097454154],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00072835834,0.00010627206,0.00023442684,0.00051053497,0.00024799485,0.00022254727,0.0002430942,0.000091394555,0.00007053105],"category_scores_gemma":[0.00011289009,0.000058585236,0.00011326498,0.005657504,0.000160085,0.0001553991,0.00015427858,0.00022034971,6.1707266e-7],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013830501,0.00019819714,0.01709887,0.000023639135,0.0001419106,0.000029095552,0.00001356597,0.0000074718077,0.00005297975,0.001943613,0.000034750912,0.98044205],"study_design_scores_gemma":[0.00058768666,0.0006758571,0.9623515,0.000074062016,0.00006302549,0.00014859681,0.011293874,0.00004422786,0.00003887504,0.00805605,0.016481752,0.00018448911],"about_ca_topic_score_codex":0.000013458475,"about_ca_topic_score_gemma":0.0007238141,"teacher_disagreement_score":0.9802576,"about_ca_system_score_codex":0.00008620102,"about_ca_system_score_gemma":0.0000074433115,"threshold_uncertainty_score":0.2718243},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4411459434","doi":"10.18280/ijsse.150401","title":"Security of Personal Information in Electronic Payments: A Bibliometric Analysis and Model Extension","year":2025,"lang":"en","type":"article","venue":"International Journal of Safety and Security Engineering","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Extension (predicate logic); Payment; Computer science; Computer security; Data science; World Wide Web","score_opus":0.006958704925649267,"score_gpt":0.2487574131937168,"score_spread":0.24179870826806754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411459434","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.40192968,0.000833755,0.59486926,0.0021853498,0.00009522899,0.0000317758,0.000011898991,0.000016701239,0.00002635989],"genre_scores_gemma":[0.98670167,0.0027417033,0.010500272,0.000043176817,0.0000068639315,6.3007275e-7,0.000003719918,0.0000013735657,5.9606424e-7],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99899226,0.000014172125,0.0004379047,0.00010224327,0.00033030676,0.00012311776],"domain_scores_gemma":[0.99914646,0.00011660736,0.00017968209,0.00028761756,0.00024172499,0.000027921278],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0006376698,0.000079652316,0.00019375997,0.020088518,0.000017550285,0.000068432346,0.0020123217,0.00006195934,0.0000014358413],"category_scores_gemma":[0.0022422264,0.00007726945,0.000054503493,0.01083764,0.0000226406,0.0013923662,0.0033162648,0.0002745661,8.489245e-8],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013814342,0.0010622696,0.101875655,0.0010500577,0.007897388,0.00016019514,0.011474037,0.25986832,0.0062258695,0.22962227,0.0026867867,0.3766957],"study_design_scores_gemma":[0.00040848163,0.000024586385,0.018958231,0.00008549352,0.000022217357,0.000019418972,0.000029110488,0.95369625,0.00031069128,0.026252717,0.0001337405,0.00005903877],"about_ca_topic_score_codex":0.000024350364,"about_ca_topic_score_gemma":0.000007209769,"teacher_disagreement_score":0.6938279,"about_ca_system_score_codex":0.000113284405,"about_ca_system_score_gemma":0.000057420402,"threshold_uncertainty_score":0.99101794},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4411614654","doi":"10.46843/jiecr.v6i3.2193","title":"Global Research Dynamics: A Bibliometric Exploration of Child Education in Artisanal Mining","year":2025,"lang":"en","type":"article","venue":"Journal of Innovation in Educational and Cultural Research","topic":"Higher Education Learning Practices","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Dynamics (music); Earth science; Data science; Computer science; Sociology; Pedagogy; Geology","score_opus":0.1903704054087698,"score_gpt":0.5694404386588247,"score_spread":0.37907003325005484,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411614654","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.912493,0.0008397582,0.000043671254,0.06540798,0.00047836924,0.00019921751,0.000001530011,0.0000022611207,0.0205342],"genre_scores_gemma":[0.9952556,0.00054240634,0.0011646507,0.00008614656,0.0003034852,0.000029906105,0.000014761423,0.0000031297604,0.0025999327],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99617666,0.0012069297,0.00096516707,0.00016760794,0.001207464,0.000276162],"domain_scores_gemma":[0.99015754,0.0012915519,0.00042971925,0.000087440385,0.007971982,0.00006176104],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.01288828,0.00006701221,0.00015548844,0.01991219,0.00037705372,0.00021131481,0.00028668926,0.00010361564,0.00011506311],"category_scores_gemma":[0.017523924,0.00006172124,0.00002083728,0.10429047,0.00038601502,0.0018816101,0.0000440621,0.0006331157,0.0000041301037],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007648793,0.00058675936,0.50724006,0.000044945715,0.0000124566,4.1834105e-7,0.010569234,0.00006947119,0.000068588495,0.41498086,0.010564908,0.0557858],"study_design_scores_gemma":[0.00032238834,0.00007895349,0.8077681,0.0004988502,0.0000032062512,0.0000051610627,0.1255413,0.000057095116,0.000022084707,0.03783893,0.027791891,0.00007205056],"about_ca_topic_score_codex":0.0018786177,"about_ca_topic_score_gemma":0.0023329263,"teacher_disagreement_score":0.37714192,"about_ca_system_score_codex":0.001124902,"about_ca_system_score_gemma":0.006501736,"threshold_uncertainty_score":0.9991305},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4411615288","doi":"10.1177/21925682251353137","title":"An Assessment of the World’s Contribution to Osteoporotic Vertebral Compression Fractures: A Bibliometric Analysis of Surgical Management; An AO Spine Knowledge Forum Trauma and Infection Initiative","year":2025,"lang":"en","type":"review","venue":"Global Spine Journal","topic":"Spinal Fractures and Fixation Techniques","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Osteoporosis; China; Modalities; Web of science; General surgery; Meta-analysis; Social science; Pathology","score_opus":0.02700419891752204,"score_gpt":0.45210681703572514,"score_spread":0.4251026181182031,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411615288","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01263812,0.9510389,0.029200783,0.0002919338,0.00056496105,0.0032171423,0.00029245237,0.000070871494,0.0026848256],"genre_scores_gemma":[0.12579499,0.87120825,0.0021359294,0.00014314675,0.00020000637,0.00004448049,0.00038644427,0.000017813978,0.00006893963],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9971126,0.0004196312,0.0012190957,0.00035536435,0.00060562015,0.00028771293],"domain_scores_gemma":[0.99744934,0.000037513673,0.0012093377,0.00039890836,0.0006202982,0.00028458243],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00072633155,0.0003818892,0.002036192,0.021668345,0.00016619459,0.00007349326,0.000253659,0.00026460085,0.00013627189],"category_scores_gemma":[0.000108813285,0.00023264553,0.00077614374,0.058437243,0.00007097628,0.00021075411,0.00015511236,0.00058941636,4.0403424e-7],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007018416,0.00080075127,0.008066552,0.0034784493,0.0020618297,0.000024829167,0.00001217782,0.000028286202,0.0000030401832,0.00089097075,0.0008118531,0.98375106],"study_design_scores_gemma":[0.0015137987,0.0019396658,0.7724925,0.014529285,0.013184655,0.00016031404,0.000023855044,0.00036072364,0.000024794377,0.00012221628,0.19538367,0.00026450865],"about_ca_topic_score_codex":0.00005171474,"about_ca_topic_score_gemma":0.00017831802,"teacher_disagreement_score":0.9834866,"about_ca_system_score_codex":0.00049851905,"about_ca_system_score_gemma":0.000246975,"threshold_uncertainty_score":0.98942024},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4411626115","doi":"10.55980/ebasr.v4i1.184","title":"The Evolution of Behavioral Accounting Research: A 25-Year Bibliometric Analysis","year":2025,"lang":"en","type":"article","venue":"Economics Business Accounting & Society Review","topic":"Auditing, Earnings Management, Governance","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Accounting; Psychology; Business","score_opus":0.03376998740809429,"score_gpt":0.30202050664693325,"score_spread":0.26825051923883897,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411626115","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8089905,0.12154485,0.035037484,0.007990382,0.0019499839,0.0028905973,0.00002810907,0.0004661589,0.021101972],"genre_scores_gemma":[0.92431885,0.07105499,0.00051483436,0.0016298265,0.0009210401,0.00016585323,0.00005639444,0.00008260946,0.0012555919],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9960387,0.000059577316,0.0014563202,0.00085507997,0.0006450552,0.0009452634],"domain_scores_gemma":[0.9830914,0.00048049685,0.012915496,0.0011070443,0.0023866282,0.000018911776],"candidate_categories":["metaepi_narrow","bibliometrics","sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.008523735,0.0004196289,0.0009846588,0.0067036636,0.0014093457,0.0013519734,0.0016182405,0.00016340715,0.00014532026],"category_scores_gemma":[0.007391338,0.00037796528,0.00083141983,0.16658466,0.0003861496,0.0022431728,0.0015319462,0.00063334266,0.00019673075],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000022689774,0.00024376962,0.60442454,0.010253552,0.0014076006,0.0000017492403,0.000055663524,0.0008264599,0.000051936266,0.05573544,0.04749926,0.27947736],"study_design_scores_gemma":[0.0004637909,0.000004509337,0.41187152,0.0021766955,0.0018306582,5.041404e-7,0.0006776438,0.0033751319,0.0000035296512,0.0015896306,0.5774995,0.0005068481],"about_ca_topic_score_codex":0.002959718,"about_ca_topic_score_gemma":0.00031770876,"teacher_disagreement_score":0.53000027,"about_ca_system_score_codex":0.0006526505,"about_ca_system_score_gemma":0.00025231295,"threshold_uncertainty_score":0.9998907},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4411659928","doi":"10.32479/ijeep.19320","title":"Adopting Indirect Carbon Pricing Strategies for Indonesia: Insights from Global Practices Using a Bibliometrics and Systematic Literature Review","year":2025,"lang":"en","type":"article","venue":"International Journal of Energy Economics and Policy","topic":"Energy, Environment, and Transportation Policies","field":"Energy","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Economics; Computer science; Library science","score_opus":0.02050142215235948,"score_gpt":0.3080028115993441,"score_spread":0.2875013894469846,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411659928","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8832496,0.112119004,0.0010753997,0.0009362076,0.00044460024,0.00008475125,0.00004621779,0.000007892496,0.002036372],"genre_scores_gemma":[0.8226854,0.1747783,0.0011770886,0.0009383086,0.00036577682,0.0000061444794,0.000017663277,0.0000096721415,0.000021651276],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9989226,0.000054630054,0.00066063454,0.0001460714,0.00010423726,0.0001118697],"domain_scores_gemma":[0.9981271,0.00030876498,0.0012555651,0.000081635975,0.0001637095,0.00006322131],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00017803836,0.00015221637,0.0003890429,0.0015513576,0.0000603609,0.00028229764,0.00017207648,0.00008698313,9.750158e-7],"category_scores_gemma":[0.00017922992,0.00013288186,0.0000843513,0.0006986351,0.000035832403,0.00046653848,0.000030798634,0.000081351995,3.7218815e-8],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010835958,0.0000830854,0.006896593,0.004650269,0.0017842659,0.00003075744,0.0006552078,0.014378034,0.00019162768,0.9664473,0.00002656594,0.004747947],"study_design_scores_gemma":[0.017706951,0.00095345976,0.08615814,0.24426456,0.006084383,0.0019717312,0.0065182424,0.06764596,0.0030804435,0.37495354,0.18688717,0.0037754204],"about_ca_topic_score_codex":0.0017736377,"about_ca_topic_score_gemma":0.0006122147,"teacher_disagreement_score":0.5914938,"about_ca_system_score_codex":0.00013503636,"about_ca_system_score_gemma":0.00016072002,"threshold_uncertainty_score":0.5418765},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4411661421","doi":"10.3389/fpsyt.2025.1621830","title":"Research trends in esketamine for depression over the past decade: a bibliometric analysis","year":2025,"lang":"en","type":"review","venue":"Frontiers in Psychiatry","topic":"Treatment of Major Depression","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Hwa Mei Hospital, University of Chinese Academy of Sciences; Medical Scientific Research Foundation of Zhejiang Province, China","keywords":"Depression (economics); Psychology; Major depressive disorder; Psychiatry; Bibliometrics; Medicine; Library science; Mood; Computer science","score_opus":0.05292631482484921,"score_gpt":0.44345495257790724,"score_spread":0.390528637753058,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411661421","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00015422374,0.9906773,0.00078035827,0.00044979586,0.004077799,0.0020289754,0.00010574633,0.000044014483,0.0016817901],"genre_scores_gemma":[0.00008738278,0.98077875,0.011001801,0.000059650785,0.00047730727,0.0013208155,0.00071826775,0.000077775614,0.005478272],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99517703,0.0006713589,0.0011449123,0.0011522267,0.0010376147,0.00081686],"domain_scores_gemma":[0.99709535,0.0007807576,0.00033667305,0.0015193471,0.00010535111,0.00016250915],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0019216635,0.00057833194,0.002788384,0.29799905,0.00015117666,0.00006956913,0.00075096154,0.00085634657,0.00008391667],"category_scores_gemma":[0.0003152354,0.0003513805,0.0013975499,0.34050345,0.00011973714,0.00009378433,0.00024775433,0.0015411576,0.0000059068134],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001073817,0.00039967205,0.17999847,0.005331648,0.00081210374,0.000010748953,0.000011516536,0.0000022351335,1.2562409e-8,0.000007981851,0.2222304,0.5910878],"study_design_scores_gemma":[0.0029203983,0.00015682072,0.22592577,0.018124495,0.0074029732,0.0000074612226,0.00007024444,0.00011403836,1.2318553e-7,0.000181831,0.7447788,0.00031703484],"about_ca_topic_score_codex":0.000073270254,"about_ca_topic_score_gemma":0.00014304627,"teacher_disagreement_score":0.5907708,"about_ca_system_score_codex":0.0006104942,"about_ca_system_score_gemma":0.00052037055,"threshold_uncertainty_score":0.99989384},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4411693809","doi":"10.3389/fpsyg.2025.1541092","title":"Global research trends on artificial intelligence in psychological interventions for stroke survivors: a bibliometric and visualized analysis (2000–2024)","year":2025,"lang":"en","type":"review","venue":"Frontiers in Psychology","topic":"Stroke Rehabilitation and Recovery","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Psychological intervention; Psychology; Applied psychology; Bibliometrics; Medical education; Medicine; Library science; Computer science; Psychiatry","score_opus":0.2203204411111077,"score_gpt":0.5813381772429104,"score_spread":0.36101773613180266,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411693809","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00023352374,0.95534503,0.034017954,0.0003520194,0.004022924,0.0015056522,0.00036720076,0.000032414486,0.004123263],"genre_scores_gemma":[0.00017622235,0.9880362,0.008490731,0.000112570924,0.0001328475,0.0007679138,0.00023524862,0.000022464143,0.0020257772],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9946111,0.001403308,0.0016228198,0.0013392254,0.0003400863,0.0006835003],"domain_scores_gemma":[0.9971546,0.0016173288,0.00023575746,0.0006584908,0.00016047253,0.00017335458],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0028492108,0.00040542282,0.0030333598,0.14378619,0.00006358376,0.00004981603,0.0003654514,0.0010078219,0.0002776757],"category_scores_gemma":[0.0021783323,0.00033401285,0.0015724153,0.13958123,0.00042601133,0.000042068736,0.00008158202,0.0009863607,0.000011543052],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005925938,0.0012989178,0.0071958844,0.0041332957,0.0006805716,0.000027258413,0.000028329889,0.0000011991323,2.9689257e-8,0.00036842373,0.012267176,0.9734063],"study_design_scores_gemma":[0.0031327088,0.006751617,0.06577138,0.02300563,0.004461975,0.00006746933,0.00083125185,0.0004261553,1.2387886e-7,0.014392183,0.88012314,0.0010363923],"about_ca_topic_score_codex":0.00005358854,"about_ca_topic_score_gemma":0.0003157261,"teacher_disagreement_score":0.9723699,"about_ca_system_score_codex":0.00053685415,"about_ca_system_score_gemma":0.00016677588,"threshold_uncertainty_score":0.9999112},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4411781025","doi":"10.29278/azd.1652297","title":"Reflections of Automation in the Dairy Industry: A Bibliometric Analysis Approach on Robotic Milking Systems","year":2025,"lang":"en","type":"article","venue":"Akademik Ziraat Dergisi","topic":"Food Supply Chain Traceability","field":"Agricultural and Biological Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Automation; Milking; Dairy industry; Manufacturing engineering; Engineering; Automatic milking; Computer science; Engineering management; Operations management; Mechanical engineering; Animal science; Biology","score_opus":0.048842837446509864,"score_gpt":0.3172947989434994,"score_spread":0.2684519614969895,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411781025","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9897009,0.00044697325,0.00025253854,0.001756779,0.00012187586,0.00043154682,0.0000109901275,0.00006242379,0.007215946],"genre_scores_gemma":[0.9994022,0.00001901142,0.00010947147,0.00016350987,0.000050393523,0.00007144179,0.000043155393,7.4765217e-7,0.00014006764],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99806005,0.000498671,0.00045734903,0.00035609552,0.00038191496,0.00024591075],"domain_scores_gemma":[0.998448,0.0011057756,0.00014953694,0.00018503027,0.00007672619,0.000034921708],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0012955179,0.00013883498,0.00031086936,0.0055335173,0.00016199023,0.00007942423,0.00044255832,0.00032531287,0.000024541754],"category_scores_gemma":[0.00043973228,0.000052022337,0.00015088142,0.16756946,0.000061529645,0.000120558594,0.00004636591,0.00049867807,0.0000033366555],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009456031,0.0020048837,0.7334742,0.00025708022,0.00082154205,0.000007116365,0.0014317286,0.1926046,0.017154211,0.007936359,0.003872058,0.040341657],"study_design_scores_gemma":[0.000119711534,0.00012130113,0.9809504,0.00004360465,0.00015422422,0.0000022219388,0.0023809553,0.0148246875,0.0002412415,0.00028432786,0.0007527317,0.00012456175],"about_ca_topic_score_codex":0.0016055617,"about_ca_topic_score_gemma":0.0005525264,"teacher_disagreement_score":0.24747624,"about_ca_system_score_codex":0.00008890928,"about_ca_system_score_gemma":0.000018132348,"threshold_uncertainty_score":0.85012424},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4411789574","doi":"10.53287/viyt6531ns47t","title":"Analisis bibliometrico de politicas publicas para adultos mayores en Scopus","year":2024,"lang":"es","type":"article","venue":"Temas sociales","topic":"Health and Medical Education","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Humanities; Political science; Philosophy; MEDLINE","score_opus":0.23255724576632003,"score_gpt":0.571072411270942,"score_spread":0.33851516550462196,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411789574","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6815075,0.030358046,0.00041121585,0.25186,0.005549939,0.0018343162,0.00046036433,0.0008366341,0.027181946],"genre_scores_gemma":[0.9724583,0.0063882796,0.00046875642,0.009190054,0.007957421,0.0002842326,0.0001503156,0.00009353189,0.003009085],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9934438,0.0020803786,0.0011212549,0.0006129046,0.0008362766,0.0019054268],"domain_scores_gemma":[0.9919578,0.0053180386,0.00021030469,0.0004271826,0.00036786636,0.0017188191],"candidate_categories":["metaresearch","metaepi_narrow","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0038276983,0.00035461533,0.000630365,0.0054681296,0.0010171965,0.00028880688,0.00038310862,0.0010833199,0.0048178476],"category_scores_gemma":[0.009821063,0.00031977877,0.00024853958,0.012924015,0.00043300173,0.00035302487,0.00012908691,0.0015960138,0.002998513],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000037805243,0.00039763254,0.08139633,0.013960656,0.00019725935,0.00006508385,0.02111313,1.9315769e-7,0.00007935,0.25724226,0.59563065,0.029879674],"study_design_scores_gemma":[0.0005700348,0.00018405906,0.34652054,0.004476181,0.00028670227,0.0000143794805,0.024553727,0.00021108057,0.0000640279,0.00650978,0.61609125,0.0005182453],"about_ca_topic_score_codex":0.011043192,"about_ca_topic_score_gemma":0.0001738246,"teacher_disagreement_score":0.29095078,"about_ca_system_score_codex":0.0015418116,"about_ca_system_score_gemma":0.0151165845,"threshold_uncertainty_score":0.99992543},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4411854991","doi":"10.18196/jiwp.v9i1.178","title":"Exploring Gender Dynamics in Hajj Research: A Bibliometric Review","year":2025,"lang":"en","type":"review","venue":"Journal of Islamic World and Politics","topic":"Education and Islamic Studies","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Hajj; Dynamics (music); Data science; Computer science; Psychology; Geography; Islam; Pedagogy","score_opus":0.5340330161375859,"score_gpt":0.5225555583529298,"score_spread":0.011477457784656142,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411854991","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000014267062,0.91461414,0.0000021131302,0.0019161286,0.0010851236,0.00031051046,0.000012777898,0.0000062301538,0.0820387],"genre_scores_gemma":[0.00008540608,0.9865415,0.000090846064,0.00024433777,0.0008998582,0.000019436608,0.000003212683,0.000015750116,0.012099663],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9963005,0.0010872479,0.0011398409,0.00018745851,0.00072115223,0.0005637775],"domain_scores_gemma":[0.99613607,0.0024927969,0.000529135,0.00018480352,0.0005563202,0.000100899444],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0048816022,0.00021709838,0.0014399579,0.037069775,0.0002952667,0.000112418034,0.00045266096,0.00010292572,0.00004935598],"category_scores_gemma":[0.0032380628,0.00017472736,0.00030810645,0.045963343,0.00029715349,0.00023123623,0.00012432146,0.0011545549,0.000008648979],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[5.602542e-7,0.00006157474,0.000094817355,0.03413996,0.00009517416,0.000020358935,0.0019388057,3.1003506e-8,1.698081e-9,0.314221,0.02037312,0.6290546],"study_design_scores_gemma":[0.00007275064,0.000013777106,0.000040641346,0.07196705,0.00021601841,0.000011882867,0.0020999399,4.1035858e-7,7.8487465e-9,0.005749755,0.91968596,0.00014178485],"about_ca_topic_score_codex":0.00035910713,"about_ca_topic_score_gemma":0.0012366228,"teacher_disagreement_score":0.89931285,"about_ca_system_score_codex":0.0011325874,"about_ca_system_score_gemma":0.0023728113,"threshold_uncertainty_score":0.97431517},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4411898912","doi":"10.25236/fmsr.2025.070316","title":"Research on Traumatic Knee Osteoarthritis: A Bibliometric and Visualized Study","year":2025,"lang":"en","type":"article","venue":"Frontiers in Medical Science Research","topic":"Musculoskeletal Disorders and Rehabilitation","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Osteoarthritis; Medicine; Physical medicine and rehabilitation; Alternative medicine; Pathology","score_opus":0.0763898573796808,"score_gpt":0.5199698948473029,"score_spread":0.44358003746762215,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411898912","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9746415,0.002842857,0.00021171685,0.0042816126,0.00031680978,0.0020781783,6.0712694e-7,0.000022081702,0.015604645],"genre_scores_gemma":[0.9976285,0.00054841116,0.00057532085,0.00011475207,0.00003709907,0.00025567375,0.0000010562034,0.000008204523,0.0008309758],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9896275,0.0011564207,0.00040221453,0.0007169739,0.007110009,0.0009868834],"domain_scores_gemma":[0.996907,0.0014658143,0.000018471232,0.00041314473,0.00062675786,0.0005688134],"candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.033909675,0.000110054985,0.00036019518,0.06127398,0.00054440217,0.00013931969,0.00045419898,0.00016314832,0.00017552623],"category_scores_gemma":[0.021270301,0.00008012807,0.000043772296,0.140358,0.0039007158,0.0001580596,0.0003011217,0.0015253931,0.000018697909],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033062475,0.0018537713,0.12082947,0.00041015935,0.000015167923,0.000085120184,0.0012599407,6.466848e-7,0.000103044455,0.0010853902,0.005685569,0.8683411],"study_design_scores_gemma":[0.0056396807,0.0069063064,0.935035,0.001276681,0.0000069822636,0.0000041159847,0.04249054,0.0027562443,0.000021135931,0.0045591793,0.0011929175,0.00011123517],"about_ca_topic_score_codex":0.0002847073,"about_ca_topic_score_gemma":0.0000844504,"teacher_disagreement_score":0.86822987,"about_ca_system_score_codex":0.00034291815,"about_ca_system_score_gemma":0.0016552982,"threshold_uncertainty_score":0.9988101},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4411940044","doi":"10.21432/cjlt28788","title":"Intelligence artificielle et formation universitaire : analyse bibliométrique des tendances et perspectives de recherche","year":2025,"lang":"fr","type":"article","venue":"Canadian Journal of Learning and Technology","topic":"Competitive and Knowledge Intelligence","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Psychology; Sociology; Political science","score_opus":0.05889015404330378,"score_gpt":0.3324018800154343,"score_spread":0.2735117259721305,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411940044","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.51694393,0.19253089,0.14996143,0.08632812,0.00057391135,0.0001510786,0.0000036491235,0.00006620509,0.053440794],"genre_scores_gemma":[0.98446125,0.0071582864,0.0012720284,0.00012172547,0.00006573617,9.3911217e-7,0.000001221061,0.00000866234,0.0069101253],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.998981,0.00019815793,0.00028534018,0.00015734289,0.000057340792,0.00032079156],"domain_scores_gemma":[0.998289,0.00025867947,0.00026244557,0.00007970121,0.0010541285,0.000056032615],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017632255,0.00014193662,0.00022210953,0.0053467755,0.00038357434,0.00026254728,0.0002526709,0.00028985777,0.00019034518],"category_scores_gemma":[0.0024669557,0.00014888978,0.000071509974,0.004948921,0.0006009361,0.00090392167,0.000058701335,0.0012438404,0.000020423393],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000027916325,0.000045749308,0.19091913,0.0003595527,0.00015350142,0.0002627188,0.0038803525,0.0007493336,0.0003219142,0.34869558,0.001522437,0.45306182],"study_design_scores_gemma":[0.00019455941,0.00020978664,0.0016126729,0.0030450625,0.00022235446,0.00023914714,0.31844735,0.007285145,0.0021228264,0.10745854,0.5588377,0.00032484773],"about_ca_topic_score_codex":0.002457297,"about_ca_topic_score_gemma":0.051570855,"teacher_disagreement_score":0.5573153,"about_ca_system_score_codex":0.00037318195,"about_ca_system_score_gemma":0.0010274729,"threshold_uncertainty_score":0.9657355},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4412141613","doi":"10.3389/fonc.2025.1594848","title":"A bibliometric analysis of liver cancer bone metastases: advances in mechanisms of occurrence and treatment options","year":2025,"lang":"en","type":"review","venue":"Frontiers in Oncology","topic":"Bone health and treatments","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Medicine; Cancer; Liver cancer; Oncology; Internal medicine; Cancer research","score_opus":0.05348551065851828,"score_gpt":0.4308797235302952,"score_spread":0.3773942128717769,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412141613","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0001069045,0.99656034,0.0005465044,0.000023877821,0.00031342596,0.0011316403,0.0011779751,0.000005232681,0.00013408248],"genre_scores_gemma":[0.0001523044,0.9893582,0.0094419485,0.000022357874,0.0000050180975,0.00061072403,0.0002878474,0.000006900301,0.0001147193],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980344,0.00026299513,0.0009058456,0.00039331015,0.00014423138,0.0002591764],"domain_scores_gemma":[0.99866325,0.00036942895,0.0005435058,0.000249717,0.000068313064,0.00010575926],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00026246413,0.00026119387,0.0042266995,0.09629349,0.000016408756,0.0000017864829,0.000073293304,0.0002809471,0.000045898385],"category_scores_gemma":[0.00012237088,0.00019730427,0.00032332703,0.087040514,0.00010957588,0.000048936894,0.00003929165,0.00017014383,4.7366862e-7],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008291984,0.00081936887,0.0070472904,0.0075631705,0.0016844913,0.000092685594,0.000046622758,0.000005580883,4.7615302e-8,0.00005566403,0.00016120676,0.98244095],"study_design_scores_gemma":[0.0044373465,0.0026950808,0.0066425153,0.010043614,0.048856944,0.000021649676,0.00012753949,0.0001307924,0.0000030474653,0.00015290901,0.9266009,0.0002877063],"about_ca_topic_score_codex":0.001512599,"about_ca_topic_score_gemma":0.0026052769,"teacher_disagreement_score":0.98215324,"about_ca_system_score_codex":0.00094302854,"about_ca_system_score_gemma":0.0014687417,"threshold_uncertainty_score":0.9323649},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4412143621","doi":"10.4103/bc.bc_107_24","title":"Exploring the link trend in the field of coronaviruses and cognitive impairment: A bibliometric analysis based on bibliometrix","year":2025,"lang":"en","type":"article","venue":"Brain Circulation","topic":"Long-Term Effects of COVID-19","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Field (mathematics); Coronavirus disease 2019 (COVID-19); Link (geometry); Cognition; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Cognitive impairment; 2019-20 coronavirus outbreak; Geography; Data science; Psychology; Computer science; Medicine; Virology; Neuroscience; Infectious disease (medical specialty); Mathematics; Disease; Outbreak","score_opus":0.060630298458533766,"score_gpt":0.373198650353623,"score_spread":0.31256835189508925,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412143621","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9814855,0.00029346807,0.0027006222,0.014278691,0.00004367123,0.0007268773,0.000009085823,0.000020374287,0.0004417112],"genre_scores_gemma":[0.99421227,0.00003802674,0.000034763394,0.005531701,0.000047196987,0.00008811722,0.00003222499,0.000006662846,0.0000090082385],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99871504,0.00023760933,0.00026772218,0.00022107497,0.00041666918,0.0001418835],"domain_scores_gemma":[0.9840119,0.015494294,0.00010741172,0.0002824086,0.0000686381,0.000035356643],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0012002822,0.000114484545,0.00024673165,0.14118019,0.00007162972,0.000050852716,0.000100569094,0.000054406297,0.000023575909],"category_scores_gemma":[0.0050693653,0.00007165554,0.00012510044,0.2848229,0.000053325155,0.00014555764,0.000024116236,0.00015482909,0.0000012624464],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030597395,0.00014514939,0.90213287,0.0002228264,0.00029214207,0.000013759826,0.0005379601,0.00083345396,0.00039741528,0.00017099103,0.00019472587,0.09475274],"study_design_scores_gemma":[0.0017027927,0.0002664421,0.98700607,0.0002460636,0.00057610736,0.0000012830945,0.00014514195,0.009275531,0.00050421286,0.0000951458,0.000117864474,0.00006334605],"about_ca_topic_score_codex":0.00021207158,"about_ca_topic_score_gemma":0.00009319929,"teacher_disagreement_score":0.14364271,"about_ca_system_score_codex":0.000043747197,"about_ca_system_score_gemma":0.000062217034,"threshold_uncertainty_score":0.8685538},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4412148307","doi":"10.26900/hsq.2709","title":"Trends and current topics in child and family-centered care research: A bibliometric analysis and visual mapping","year":2025,"lang":"en","type":"article","venue":"Health Sciences Quarterly","topic":"Delphi Technique in Research","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Current (fluid); Data science; Library science; Psychology; Regional science; Geography; Computer science; Engineering","score_opus":0.2510049583489637,"score_gpt":0.5600990201313116,"score_spread":0.30909406178234794,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412148307","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9536949,0.022206632,0.00031103837,0.015994303,0.000148362,0.00044787308,0.000008601796,0.000052978492,0.007135302],"genre_scores_gemma":[0.99472696,0.004627717,0.00034942455,0.00014014782,0.000043787248,0.000034334513,0.0000018333295,0.0000028171191,0.000072989395],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99656534,0.0007727488,0.0003500024,0.0005731718,0.0009691168,0.0007696162],"domain_scores_gemma":[0.99903005,0.00035328703,0.000068468784,0.00012593843,0.00015144823,0.00027078087],"candidate_categories":["bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0065528294,0.00010220351,0.00027418302,0.08175646,0.0014213914,0.0005194714,0.00030083445,0.00007573709,0.000007636656],"category_scores_gemma":[0.00020442648,0.000095553805,0.000031607313,0.17959532,0.0013608722,0.0002997761,0.00011007919,0.0003742685,6.8797965e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000037761167,0.000030211106,0.17090918,0.000093023424,0.00000606056,0.0000018213393,0.015457706,1.2223649e-7,0.0000025467657,0.0012033269,0.00021063448,0.8120816],"study_design_scores_gemma":[0.00024860876,0.00044583186,0.9314946,0.00021053654,0.000006114304,8.1335656e-7,0.053952504,0.0007091071,0.0000013550848,0.0008306705,0.011988619,0.00011122478],"about_ca_topic_score_codex":0.011719089,"about_ca_topic_score_gemma":0.01526336,"teacher_disagreement_score":0.81197035,"about_ca_system_score_codex":0.00017627722,"about_ca_system_score_gemma":0.00042707223,"threshold_uncertainty_score":0.99987864},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4412173650","doi":"10.1097/as9.0000000000000594","title":"Bibliometric Analysis of Surgical Articles Using Bayesian Statistics","year":2025,"lang":"en","type":"review","venue":"Annals of Surgery Open","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Health Sciences Centre; University of Ottawa; Sunnybrook Health Science Centre","funders":"","keywords":"Statistics; Bayesian probability; Bayesian statistics; Data science; Computer science; Bayesian inference; Mathematics","score_opus":0.9381523087201591,"score_gpt":0.6551801492884783,"score_spread":0.28297215943168075,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412173650","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00016398623,0.98756254,0.008411389,0.000048396236,0.00015412782,0.00093263044,0.0012101029,0.0000017876808,0.001515058],"genre_scores_gemma":[0.0002172752,0.9949372,0.0028916239,0.000032626933,0.000023790328,0.000025103735,0.00012776697,0.000017209371,0.0017274374],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.95282036,0.014086766,0.025531076,0.001256851,0.0058954074,0.00040956467],"domain_scores_gemma":[0.90478694,0.06580403,0.020601807,0.004912715,0.0036537133,0.00024078465],"candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","metaepi_broad","bibliometrics"],"category_scores_codex":[0.1777411,0.0005544459,0.031940665,0.20618649,0.00008623905,0.0010809762,0.004300006,0.00024482724,0.013307393],"category_scores_gemma":[0.05326555,0.00028451538,0.011543458,0.5004396,0.00012755806,0.00033390548,0.0008957793,0.00017716335,0.00014392243],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000033347624,0.000089278896,0.004773016,0.0049073384,0.0070420066,0.000019899635,0.000008117058,0.00003287035,2.4748621e-8,0.00040719716,0.046873674,0.9358432],"study_design_scores_gemma":[0.000029631603,0.0000099512245,0.0010712963,0.0043644393,0.018111631,0.0000046199243,0.000021353037,0.0018230573,0.0000010357585,0.00092510675,0.9733244,0.00031344633],"about_ca_topic_score_codex":0.00025276787,"about_ca_topic_score_gemma":0.000041738385,"teacher_disagreement_score":0.93552977,"about_ca_system_score_codex":0.000015368563,"about_ca_system_score_gemma":0.0008670817,"threshold_uncertainty_score":0.9999607},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4412371521","doi":"10.55863/ijees.2025.0680","title":"Responses of Pine and Oak Species to Drought: A Bibliometric Analysis","year":2025,"lang":"en","type":"article","venue":"International Journal of Ecology and Environmental Sciences","topic":"Forest ecology and management","field":"Environmental Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Indian Council of Forestry Research and Education; Indian National Science Academy","keywords":"Forestry; Geography; Environmental science; Agroforestry","score_opus":0.010466353043900624,"score_gpt":0.2672959707106232,"score_spread":0.2568296176667226,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412371521","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9933122,0.00013719677,0.0004139405,0.002743015,0.00022312753,0.000050165945,0.000008390775,0.0000015424498,0.00311047],"genre_scores_gemma":[0.9967789,0.0005584761,0.00094559096,0.00067046855,0.000012211472,0.0000017823763,9.485808e-7,0.0000010149442,0.0010306274],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9991912,0.00004860281,0.0002637005,0.0001504101,0.00023471586,0.000111333204],"domain_scores_gemma":[0.9995682,0.0001743504,0.00015382298,0.00004588976,0.0000058146875,0.00005187645],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0006373819,0.00006562935,0.00015905524,0.0065924046,0.00009591863,0.000020101676,0.00029917492,0.000036623194,0.001209311],"category_scores_gemma":[0.00009993495,0.00005223942,0.00004740883,0.0055043944,0.00086573133,0.00019285915,0.00036384523,0.000055598204,0.00000893157],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008882145,0.00010451329,0.9904488,0.0000014304891,0.00021486453,0.000014398964,0.00008897084,0.001300928,0.0020809097,0.0018033085,0.0008713777,0.0029816786],"study_design_scores_gemma":[0.00021079674,0.0002945401,0.99096197,0.000004924468,0.000077493096,0.00001655511,0.00022758327,0.00013084503,0.00042913633,0.002548907,0.0050513623,0.000045882036],"about_ca_topic_score_codex":0.00003093254,"about_ca_topic_score_gemma":0.00015980768,"teacher_disagreement_score":0.004179985,"about_ca_system_score_codex":0.00006535087,"about_ca_system_score_gemma":0.00000840563,"threshold_uncertainty_score":0.9997037},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4412487386","doi":"10.1007/s00345-025-05805-z","title":"AI-driven scientometric analysis with recency insights: lower urinary tract symptoms in older and frail populations","year":2025,"lang":"en","type":"article","venue":"World Journal of Urology","topic":"Urinary Bladder and Prostate Research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Terminology; Medicine; Metric (unit); Lower urinary tract symptoms; Mainstream; Data science; Computer science; Internal medicine; Political science","score_opus":0.020449178022128735,"score_gpt":0.33904191511381526,"score_spread":0.3185927370916865,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412487386","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97580963,0.009564894,0.000576112,0.010085651,0.00013315026,0.0001770149,0.0000023183525,0.0000071179848,0.0036440832],"genre_scores_gemma":[0.9968502,0.0006452961,0.00035649817,0.0007333523,0.000043486925,0.0000030928095,0.0000046627915,0.0000072603225,0.0013561745],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99874395,0.00013721238,0.0004011314,0.0001854818,0.00028560686,0.00024659449],"domain_scores_gemma":[0.99918765,0.00015704529,0.000117319774,0.00016660779,0.00022655315,0.00014480914],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003521104,0.00010704218,0.0004166614,0.009121692,0.000069513095,0.000021471735,0.00010161671,0.00006189774,0.00026590304],"category_scores_gemma":[0.00007364323,0.00007521967,0.0000911733,0.0090461485,0.00014206392,0.00014539203,0.000044387125,0.0006032725,0.0000029444072],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00092756696,0.0010823953,0.98886925,0.00002830712,0.00046159935,0.001458264,0.0002397377,0.00036456605,0.00005417917,0.0002841941,0.00054279726,0.0056871586],"study_design_scores_gemma":[0.002155518,0.0022256563,0.9786932,0.00004740219,0.0006367507,0.00067958736,0.000029972849,0.0005876559,0.0000043249097,0.00043978228,0.014430696,0.0000694415],"about_ca_topic_score_codex":0.000023350729,"about_ca_topic_score_gemma":0.00009613844,"teacher_disagreement_score":0.021040518,"about_ca_system_score_codex":0.00007230505,"about_ca_system_score_gemma":0.00021651053,"threshold_uncertainty_score":0.81391907},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4412489960","doi":"10.4328/acam.22740","title":"Transcranial magnetic stimulation in neurological rehabilitation: A bibliometric mapping of 34 years of literature","year":2025,"lang":"en","type":"article","venue":"The Annals of Clinical and Analytical Medicine","topic":"Transcranial Magnetic Stimulation Studies","field":"Neuroscience","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Transcranial magnetic stimulation; Rehabilitation; Physical medicine and rehabilitation; Stimulation; Medicine; Neuroscience; Psychology","score_opus":0.19207425882590778,"score_gpt":0.4497261185659509,"score_spread":0.2576518597400431,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412489960","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9751866,0.0011702462,0.00030077266,0.022356128,0.00009404777,0.00022670494,0.000008141809,0.000007829697,0.0006495428],"genre_scores_gemma":[0.9979544,0.00090593944,0.00015356341,0.0008899794,0.000052030802,0.000002656354,4.912554e-7,0.0000032026878,0.000037686783],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9974853,0.00040328284,0.001348325,0.0002617884,0.00036555468,0.0001357498],"domain_scores_gemma":[0.9862205,0.01319989,0.00016631163,0.00018780625,0.00017274142,0.000052750205],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0014300003,0.00009804568,0.00067651144,0.0025071695,0.000021332684,0.0000041978024,0.00017934684,0.000099977464,0.000059295246],"category_scores_gemma":[0.017480286,0.0000615862,0.00012899996,0.014365258,0.0016256344,0.000043407133,0.00003794597,0.00023740368,4.9063766e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005421895,0.001707129,0.4025831,0.001107001,0.000066337925,0.000028961462,0.0026981763,0.00065025454,0.019651454,0.06502961,0.0005354883,0.5005206],"study_design_scores_gemma":[0.0019966213,0.0019120744,0.9720585,0.00033667128,0.000046346904,8.4572366e-7,0.00004476732,0.008262517,0.000049524457,0.014993663,0.00025237442,0.000046142846],"about_ca_topic_score_codex":0.000016834714,"about_ca_topic_score_gemma":0.0000016751073,"teacher_disagreement_score":0.56947535,"about_ca_system_score_codex":6.692542e-7,"about_ca_system_score_gemma":0.000020274421,"threshold_uncertainty_score":0.9907959},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4412539073","doi":"10.1080/21622965.2025.2520462","title":"Knowledge mapping of Specific Language Impairment in children: A bibliometric analysis (2010–2024)","year":2025,"lang":"en","type":"article","venue":"Applied Neuropsychology Child","topic":"Language Development and Disorders","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Psychology; Specific language impairment; Cognitive psychology; Developmental psychology","score_opus":0.012333097952128542,"score_gpt":0.292347786448701,"score_spread":0.28001468849657246,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412539073","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.75180817,0.0024653191,0.00040109883,0.00027352755,0.0008380551,0.00051805505,0.00001263945,0.000080507714,0.24360263],"genre_scores_gemma":[0.99582434,0.00015029537,0.00018187205,0.0010192355,0.000060915394,0.00010572164,0.00007390234,0.000030226558,0.0025535207],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99747676,0.00016288304,0.00071751536,0.00089060003,0.00016135562,0.0005909146],"domain_scores_gemma":[0.99865365,0.0001543635,0.00019396438,0.0008859297,0.000033574877,0.00007849942],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00042600397,0.00032242038,0.0006301822,0.07127305,0.00007592471,0.000021835864,0.00056325947,0.00023089586,0.0024267023],"category_scores_gemma":[0.00002350828,0.0003164382,0.00021249025,0.13539924,0.00015620516,0.000036905272,0.00015452765,0.00047752244,0.00035679046],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025863125,0.0024399455,0.89183325,0.0000328403,0.0018547522,0.0000760126,0.007549128,0.000020070625,0.0013040474,0.007917606,0.025359426,0.061354287],"study_design_scores_gemma":[0.0025269806,0.0000466642,0.9939889,0.000012505222,0.00011650592,0.000014024997,0.00094387133,0.0000026097982,0.00011616379,0.00013050121,0.0018482438,0.0002530375],"about_ca_topic_score_codex":0.000082989754,"about_ca_topic_score_gemma":0.000039991195,"teacher_disagreement_score":0.24401614,"about_ca_system_score_codex":0.000036475696,"about_ca_system_score_gemma":0.000026198904,"threshold_uncertainty_score":0.9999288},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4413024604","doi":"10.1002/asi.70013","title":"Understanding discrepancies in the coverage of <scp>OpenAlex</scp> : The case of China","year":2025,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Bureau de Coopération Interuniversitaire; Université du Québec à Montréal","funders":"Fonds de recherche du Québec; China Scholarship Council; Social Sciences and Humanities Research Council of Canada; National Natural Science Foundation of China","keywords":"China; Computer science; Data science; Information retrieval; Geography","score_opus":0.25345515263717894,"score_gpt":0.4762531204382082,"score_spread":0.22279796780102928,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413024604","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96058667,0.00013301033,0.0070142406,0.02428059,0.00037008032,0.0003364198,0.000020582684,0.000002671698,0.007255763],"genre_scores_gemma":[0.9995866,0.00009821795,0.00006469374,0.00011460066,0.000005227492,0.0000025574632,1.0152261e-7,6.0148267e-7,0.00012741426],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9961826,0.00009019184,0.00083907304,0.00007276501,0.0026251285,0.00019023582],"domain_scores_gemma":[0.9898437,0.004641227,0.001816756,0.00027786783,0.0033990087,0.00002144136],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.044418216,0.00004749222,0.00017289931,0.01591107,0.00046077397,0.0005297453,0.0019416392,0.00007252412,0.000001019791],"category_scores_gemma":[0.11444524,0.00001957084,0.00007553293,0.08355657,0.00045443207,0.0016197497,0.00032038754,0.00021986796,5.167143e-7],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013651867,0.00006677442,0.19276787,0.000020317082,0.000032917993,0.0000018055181,0.004090281,0.00033162168,0.0006047405,0.7562651,0.008748167,0.03705675],"study_design_scores_gemma":[0.0016081018,0.00035709076,0.2638918,0.00006737998,0.000038738614,0.00020483886,0.06739923,0.009288346,0.00680065,0.59974754,0.050530568,0.00006569522],"about_ca_topic_score_codex":0.000030275129,"about_ca_topic_score_gemma":0.000040513787,"teacher_disagreement_score":0.15651755,"about_ca_system_score_codex":0.00025861914,"about_ca_system_score_gemma":0.00040157323,"threshold_uncertainty_score":0.9952427},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4413053204","doi":"10.1016/j.ecofro.2025.07.008","title":"The effect of vegetation restoration on soil organic carbon mineralization in bibliometric analysis and a review","year":2025,"lang":"en","type":"review","venue":"Ecological Frontiers","topic":"Soil Carbon and Nitrogen Dynamics","field":"Agricultural and Biological Sciences","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Key Research and Development Program of China; National Natural Science Foundation of China; Ministry of Natural Resources","keywords":"Mineralization (soil science); Soil carbon; Environmental science; Vegetation (pathology); Soil science; Earth science; Environmental chemistry; Physical geography; Geology; Geography; Chemistry; Soil water","score_opus":0.013136954391484792,"score_gpt":0.26457936388523834,"score_spread":0.25144240949375357,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413053204","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08218032,0.9167082,0.0000024870917,0.00009943997,0.00013240625,0.00069568615,0.000006264915,0.000016816017,0.00015841163],"genre_scores_gemma":[0.0041900496,0.99533045,0.000007318715,0.00004869908,0.000027312562,0.00010490236,0.00019489694,0.0000011212047,0.000095234165],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9979,0.00090201595,0.00047982606,0.00032711588,0.00021484727,0.0001762024],"domain_scores_gemma":[0.99809974,0.0013757847,0.00035025788,0.00008200192,0.000048486072,0.00004370972],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0009797631,0.00022941812,0.0011806332,0.0026976452,0.00007642951,0.000035559435,0.00019839637,0.00027629605,0.000008728989],"category_scores_gemma":[0.0010403492,0.00007018835,0.00032284486,0.06795101,0.000055928314,0.00002733162,0.000047780435,0.0001906044,0.0000010152211],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000015323849,0.000026678403,0.003907372,0.0035723546,0.00015624937,0.0000026737118,0.0000019436422,0.000007120444,0.0000050963263,0.00001569026,0.00034542638,0.9919441],"study_design_scores_gemma":[0.0014599799,0.009882127,0.5846532,0.039806,0.029612266,0.000012012268,0.000056580488,0.0114633525,0.000038322505,0.00034115874,0.31981465,0.0028603447],"about_ca_topic_score_codex":0.000076782024,"about_ca_topic_score_gemma":0.0020272685,"teacher_disagreement_score":0.9890837,"about_ca_system_score_codex":0.0001264854,"about_ca_system_score_gemma":0.000020465937,"threshold_uncertainty_score":0.95186013},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4413192159","doi":"10.3389/fpsyt.2025.1553883","title":"Research trends of music in children with autism: a bibliometric analysis","year":2025,"lang":"en","type":"review","venue":"Frontiers in Psychiatry","topic":"Autism Spectrum Disorder Research","field":"Neuroscience","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Autism; Psychology; Music therapy; Population; China; Medical education; Developmental psychology; Medicine; Psychiatry; Political science","score_opus":0.0506786845579532,"score_gpt":0.3868159187011866,"score_spread":0.3361372341432334,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413192159","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0003734321,0.98979443,0.0003325252,0.0002678947,0.001106084,0.0011880004,0.00016674484,0.0000438247,0.0067270338],"genre_scores_gemma":[0.0017480071,0.99409246,0.0022922566,0.000021849963,0.00003845729,0.0003239753,0.000055260796,0.00007621357,0.0013515048],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9918184,0.0019535564,0.0012634506,0.0017455586,0.0018820262,0.0013369906],"domain_scores_gemma":[0.99738425,0.00040132197,0.00041934254,0.0016121761,0.000021651622,0.00016123669],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0018493977,0.0005789648,0.0030041856,0.54176503,0.00012463325,0.00010065413,0.002451755,0.0005534559,0.00020448906],"category_scores_gemma":[0.00030296124,0.00049127697,0.00068003614,0.73964524,0.0005494867,0.00018476945,0.0005009099,0.0026361574,0.000016761694],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004236046,0.00070576905,0.17636822,0.004758585,0.00051981525,0.000019874353,0.000060052113,0.000019862184,5.0897327e-8,0.00417376,0.011972247,0.8013594],"study_design_scores_gemma":[0.0068817553,0.0015944485,0.7017911,0.04006439,0.006252031,0.00010597996,0.00034313434,0.0005367464,0.0000058195687,0.026105395,0.21204574,0.004273442],"about_ca_topic_score_codex":0.00047693422,"about_ca_topic_score_gemma":0.00054616964,"teacher_disagreement_score":0.79708594,"about_ca_system_score_codex":0.00042486255,"about_ca_system_score_gemma":0.0019219803,"threshold_uncertainty_score":0.9997539},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4413210381","doi":"10.2196/76378","title":"Bibliometric Insights Into the Infodemic: Global Research Trends and Policy Responses: Quantitative Research","year":2025,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"Misinformation and Its Impacts","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Preprint; Data science; MEDLINE; Computer science; Political science; World Wide Web","score_opus":0.1804711852942205,"score_gpt":0.5712346276353928,"score_spread":0.39076344234117233,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413210381","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.41806102,0.00062721886,0.00022466143,0.036121987,0.00016720904,0.00046078995,0.000005971039,0.0000808987,0.54425025],"genre_scores_gemma":[0.98288417,0.004258621,0.00050822814,0.0043784943,0.00019996708,0.00004081904,0.000009272308,0.000006086817,0.0077143367],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99317884,0.0014082759,0.0007625452,0.000113461676,0.0037247944,0.00081209536],"domain_scores_gemma":[0.9938668,0.0039621796,0.00011419259,0.00036861497,0.0010461503,0.00064204156],"candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":["bibliometrics","sts"],"category_scores_codex":[0.014505954,0.00012814095,0.00021113078,0.030906087,0.002070962,0.00064664346,0.0010089903,0.00036893136,0.00022978606],"category_scores_gemma":[0.030096706,0.00008290025,0.000048774913,0.15659957,0.003024876,0.0011868287,0.0005496408,0.001144832,0.0002014417],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000053582324,0.000028409591,0.00021403495,0.000043444474,0.000015219107,0.0000012166219,0.25783285,3.7654775e-7,4.718242e-7,0.42520925,0.07845458,0.23814653],"study_design_scores_gemma":[0.00054910366,0.00015376735,0.013385562,0.00015477293,0.0000035137084,0.0000022614126,0.22738902,0.0018037519,0.0000064375395,0.024259463,0.7321762,0.00011615369],"about_ca_topic_score_codex":0.0026809422,"about_ca_topic_score_gemma":0.0014508361,"teacher_disagreement_score":0.65372163,"about_ca_system_score_codex":0.0005448825,"about_ca_system_score_gemma":0.004030381,"threshold_uncertainty_score":0.9996883},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4413282742","doi":"10.1002/pne2.70013","title":"A Bibliometric Analysis of Publications on the Prevalence of Chronic Pain in Children and Adolescents From 2009 to 2023","year":2025,"lang":"en","type":"review","venue":"Paediatric and Neonatal Pain","topic":"Musculoskeletal pain and rehabilitation","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Dalhousie University; Izaak Walton Killam Health Centre","funders":"Canadian Institutes of Health Research; Dalhousie University; Dalhousie Medical Research Foundation","keywords":"Medicine; Chronic pain; Environmental health; Psychology; Physical therapy","score_opus":0.011739234255811292,"score_gpt":0.292598013031535,"score_spread":0.2808587787757237,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413282742","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010064641,0.98700035,0.00027086455,0.00017651015,0.000022788246,0.0017321296,0.0006939632,0.000009838668,0.000028938492],"genre_scores_gemma":[0.022366766,0.97680503,0.00006987118,0.00007725566,0.00007086766,0.00010287922,0.00030848043,0.000011232905,0.00018763916],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9973877,0.0009384808,0.00070119876,0.00045270848,0.00034096517,0.00017895292],"domain_scores_gemma":[0.9954674,0.0036064177,0.00029534893,0.00043445913,0.00008209885,0.00011423695],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.003325385,0.00023758621,0.0011161196,0.04284486,0.000033175438,0.0000121719695,0.00018732756,0.00019502605,0.000064021115],"category_scores_gemma":[0.0043565016,0.00015244317,0.0003557727,0.10310827,0.000087801534,0.000034049768,0.00009467064,0.00027165,0.0000026949183],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009587846,0.00011910327,0.053514455,0.028457727,0.00033477464,1.9602136e-7,0.000026911144,0.0000013238167,3.8307053e-7,0.000021030168,0.0005067506,0.91700774],"study_design_scores_gemma":[0.00037097398,0.00044229184,0.97351015,0.011233603,0.0034342867,3.556619e-7,0.000019109451,0.00027018497,2.1108063e-7,0.000026452217,0.010500789,0.00019157091],"about_ca_topic_score_codex":0.00025962802,"about_ca_topic_score_gemma":0.000030228271,"teacher_disagreement_score":0.9199957,"about_ca_system_score_codex":0.00006861085,"about_ca_system_score_gemma":0.00020863728,"threshold_uncertainty_score":0.9680037},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4413388963","doi":"10.1016/j.dwt.2025.101393","title":"A bibliometric review of smart technologies: IoT and AI for sustainable and efficient desalination","year":2025,"lang":"en","type":"article","venue":"Desalination and Water Treatment","topic":"Water Quality Monitoring Technologies","field":"Environmental Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Francophone University Association","funders":"","keywords":"Desalination; Internet of Things; Environmental science; Computer science; Process engineering; Engineering; Environmental engineering; Embedded system; Chemistry","score_opus":0.019775319147611534,"score_gpt":0.2973163254922319,"score_spread":0.27754100634462037,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413388963","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9814534,0.00669392,0.0018200597,0.008281093,0.00003603538,0.0013865052,0.000006466846,0.00017803058,0.00014449183],"genre_scores_gemma":[0.99174505,0.0050747097,0.0020790175,0.00009114218,0.0000022911354,0.00029875644,0.000010147614,0.000005877998,0.0006929795],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9992048,0.000027942817,0.00021376747,0.00026155275,0.00011092282,0.00018101271],"domain_scores_gemma":[0.9996545,0.00006762983,0.000050326787,0.00015913349,0.000048680227,0.000019768904],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00039782058,0.00012144918,0.00018264019,0.0021953257,0.00011523422,0.00003930896,0.000066975736,0.00007794097,0.0000037222744],"category_scores_gemma":[0.00015053463,0.0000803584,0.000022918748,0.0027865842,0.00017491817,0.00008931818,0.00017631213,0.000030819887,0.0000013376851],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014245897,0.0010804824,0.12450358,0.010510674,0.00021220837,0.000012956834,0.0012158309,0.00010933569,0.009763021,0.036934383,0.0018326242,0.81368244],"study_design_scores_gemma":[0.0032664626,0.0015739789,0.052832395,0.0013504328,0.00030127537,0.000011753075,0.0013455618,0.002324183,0.83588433,0.025154037,0.075457886,0.000497716],"about_ca_topic_score_codex":0.00011452209,"about_ca_topic_score_gemma":0.000002590768,"teacher_disagreement_score":0.8261213,"about_ca_system_score_codex":0.00017629942,"about_ca_system_score_gemma":0.0000055706323,"threshold_uncertainty_score":0.32769203},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4413430814","doi":"10.3390/healthcare13172087","title":"Exercise Interventions for Metabolic Diseases: An Analysis of the Evolution of Aerobic Exercise Bibliometrics in the Field of Type 2 Diabetes Mellitus","year":2025,"lang":"en","type":"review","venue":"Healthcare","topic":"Cardiac Health and Mental Health","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Natural Science Foundation of Henan Province; National Natural Science Foundation of China","keywords":"Bibliometrics; Aerobic exercise; Psychological intervention; Diabetes mellitus; Medicine; Physical therapy; Type 2 Diabetes Mellitus; Gerontology; Endocrinology; Computer science; Library science; Nursing","score_opus":0.06635818779983202,"score_gpt":0.45742930883847976,"score_spread":0.3910711210386477,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413430814","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0020828259,0.99124855,0.000020301932,0.00028876268,0.00039182114,0.0047572902,0.0011802281,0.000008948477,0.000021246407],"genre_scores_gemma":[0.032435086,0.96617675,0.000077908895,0.00015990643,0.000045183668,0.00046969161,0.00058284943,0.000018262283,0.000034337147],"study_design_codex":"design_other","study_design_gemma":"systematic_review","domain_scores_codex":[0.9966304,0.00053464045,0.0016797759,0.0003310072,0.00044898188,0.00037521197],"domain_scores_gemma":[0.9962599,0.0011395054,0.000886536,0.0010241696,0.0004770142,0.00021285468],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0012462819,0.0002457689,0.0032686335,0.007965813,0.000077045035,0.000005266968,0.0003499749,0.00028545747,0.000015314405],"category_scores_gemma":[0.00063466316,0.00014868753,0.0021771332,0.030490628,0.000066482935,0.000046001147,0.000070884824,0.0003439737,4.398228e-7],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000080756006,0.0003834272,0.0060674017,0.3301127,0.00016740678,3.3798096e-7,0.00015508481,0.0000017041355,8.614793e-8,0.00027801076,0.00026127486,0.6624918],"study_design_scores_gemma":[0.0025310337,0.0030621744,0.15130538,0.5556762,0.117002785,0.0000041991034,0.0041251127,0.00018878751,0.000030512716,0.0003841623,0.16498558,0.00070407765],"about_ca_topic_score_codex":0.0033653798,"about_ca_topic_score_gemma":0.0007954668,"teacher_disagreement_score":0.66178775,"about_ca_system_score_codex":0.00026066502,"about_ca_system_score_gemma":0.0022728196,"threshold_uncertainty_score":0.9901168},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4413480749","doi":"10.3390/healthcare13172106","title":"Inclusive Healthcare System for Children with Disabilities: A Bibliometric Analysis and Visualization","year":2025,"lang":"en","type":"review","venue":"Healthcare","topic":"Down syndrome and intellectual disability research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Education, Audiovisual and Culture Executive Agency","keywords":"Visualization; Health care; Healthcare system; Inclusion (mineral); Computer science; Psychology; Data science; Political science; Artificial intelligence; Social psychology","score_opus":0.08176043162340335,"score_gpt":0.46433866427992837,"score_spread":0.382578232656525,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413480749","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00037894087,0.9882719,0.0004756642,0.00049972016,0.00006234202,0.00797353,0.0019857343,0.00014477734,0.00020742661],"genre_scores_gemma":[0.007051747,0.9851523,0.00009683824,0.00015022421,0.00014079305,0.0012128962,0.005440779,0.000077496414,0.00067692134],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9957984,0.0005552109,0.0010918927,0.0011263334,0.0007414111,0.0006867632],"domain_scores_gemma":[0.9962264,0.0010311388,0.00025518352,0.00087905215,0.0010591825,0.0005490582],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0007390562,0.00056176475,0.0036687672,0.051167503,0.00036278373,0.00009914771,0.00022065046,0.0005745813,0.000056504505],"category_scores_gemma":[0.0015062958,0.00039702922,0.0007182308,0.1399594,0.00019779694,0.00008725051,0.00019011727,0.00050767674,0.00001881196],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000091061906,0.00010089542,0.025952911,0.40609655,0.0018631797,9.45887e-7,0.00061326206,7.3148485e-8,4.440345e-9,0.0003650179,0.00032752674,0.56458855],"study_design_scores_gemma":[0.0053944713,0.017789913,0.027714467,0.35486358,0.06837141,0.0010723118,0.024330013,0.0002156095,0.0000034763332,0.00015031036,0.4965346,0.0035598427],"about_ca_topic_score_codex":0.0022342657,"about_ca_topic_score_gemma":0.0004971579,"teacher_disagreement_score":0.5610287,"about_ca_system_score_codex":0.0012329405,"about_ca_system_score_gemma":0.0024367126,"threshold_uncertainty_score":0.9998482},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4413558550","doi":"10.64628/aap.ujg7ru4pt","title":"Why the h-index is a bogus measure of academic impact","year":2020,"lang":"en","type":"preprint","venue":"","topic":"Higher Education Governance and Development","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Measure (data warehouse); Index (typography); Psychology; Statistics; Mathematics; Computer science; World Wide Web; Data mining","score_opus":0.05528254239620098,"score_gpt":0.3821717528764561,"score_spread":0.3268892104802551,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413558550","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12374793,0.0035859307,0.0018241024,0.5763553,0.00350341,0.0017978183,0.00010723571,0.00023276272,0.28884554],"genre_scores_gemma":[0.9843443,0.00067230884,0.00015348865,0.010358053,0.00047224722,0.000035566885,0.0000041425174,0.000009889358,0.0039499886],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99823207,0.00010182362,0.00027784158,0.0002693337,0.0009015198,0.00021742316],"domain_scores_gemma":[0.9990806,0.0000697389,0.00027547538,0.00025910098,0.00018472444,0.00013034136],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0006003869,0.00014281312,0.00022349613,0.000029413844,0.000164379,0.000052983436,0.0007262212,0.0003063352,0.0024184291],"category_scores_gemma":[0.00016232148,0.000090823785,0.00017201045,0.00026706932,0.0001381498,0.00006118129,0.00020539926,0.0006479406,0.000089820554],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000117186355,0.00002064592,0.048433684,0.000022601387,0.000098676355,4.7411646e-7,0.09367041,0.000013611888,0.00001619292,0.0036548784,0.8520591,0.0019979537],"study_design_scores_gemma":[0.00012529918,0.0000122983465,0.15714847,0.00011294218,0.00003202008,2.7786015e-7,0.0052252016,0.000015067338,0.00021910208,0.017095244,0.81973416,0.00027993944],"about_ca_topic_score_codex":0.024688007,"about_ca_topic_score_gemma":0.000752802,"teacher_disagreement_score":0.86059636,"about_ca_system_score_codex":0.00021683967,"about_ca_system_score_gemma":0.002989353,"threshold_uncertainty_score":0.9984935},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4413751223","doi":"10.2196/77214","title":"Social Media–Based Cancer Education: Bibliometric and Thematic Analysis","year":2025,"lang":"en","type":"article","venue":"JMIR Cancer","topic":"Social Media in Health Education","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Preprint; Thematic map; Thematic analysis; Sociology; Qualitative research; Social science; Geography; Computer science; World Wide Web; Cartography","score_opus":0.12037816072798274,"score_gpt":0.5308964582902044,"score_spread":0.41051829756222163,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413751223","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.90234476,0.012189373,0.000070431786,0.0647958,0.008320825,0.0017734437,0.000021955897,0.00013416022,0.010349277],"genre_scores_gemma":[0.98703444,0.0012336599,0.00014298093,0.0014624031,0.0020020457,0.007608547,0.000007515383,0.000009076418,0.000499338],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9986266,0.00018473965,0.00023200226,0.00024427343,0.00042335235,0.00028903515],"domain_scores_gemma":[0.99774075,0.0015591596,0.00014707897,0.00012052716,0.0003012337,0.00013126686],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00046552345,0.00009825689,0.00024665735,0.009634664,0.000639282,0.00009387305,0.00017913322,0.00014045765,0.0010404545],"category_scores_gemma":[0.0011744576,0.00010049951,0.00007898827,0.076838754,0.00024004445,0.00013376748,0.000017432501,0.00012453599,0.0000067610717],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011841458,0.00015508142,0.40829542,0.00020789176,0.00017922749,1.4370428e-7,0.042208187,0.000006064494,0.000006879719,0.008692408,0.026451178,0.51378566],"study_design_scores_gemma":[0.00022202947,0.0000043971154,0.88121325,0.000088578825,0.000433817,1.2838378e-8,0.0075248796,0.000059031023,0.000022184773,0.0020511781,0.108206004,0.00017463462],"about_ca_topic_score_codex":0.014630656,"about_ca_topic_score_gemma":0.022018535,"teacher_disagreement_score":0.513611,"about_ca_system_score_codex":0.0006598455,"about_ca_system_score_gemma":0.003968198,"threshold_uncertainty_score":0.99987274},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4413840551","doi":"10.1055/a-2690-9595","title":"Impact and Citation Trends of Surveys Endorsed by the EUPSA Network Office in Pediatric Surgery: A Bibliometric Analysis","year":2025,"lang":"en","type":"article","venue":"European Journal of Pediatric Surgery","topic":"Congenital Diaphragmatic Hernia Studies","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hospital for Sick Children","funders":"","keywords":"Medicine; Citation; Bibliometrics; Family medicine; Demography; Pediatrics; Library science","score_opus":0.029228402733353573,"score_gpt":0.2876931143181936,"score_spread":0.25846471158484,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413840551","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9676437,0.029834302,0.00042453405,0.00028194446,0.00045517954,0.000103301325,0.00002512074,0.000014118767,0.0012178074],"genre_scores_gemma":[0.98786175,0.010934391,0.000053891243,0.000050660215,0.000818983,0.0000018798698,0.000018754057,0.000028273751,0.0002313954],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9939525,0.002807401,0.0018577988,0.0002447981,0.00071698235,0.000420508],"domain_scores_gemma":[0.9856628,0.012095376,0.0012552743,0.0002876881,0.00051699154,0.0001818523],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.016461827,0.00027250432,0.0013082928,0.07453494,0.000077388184,0.000050814335,0.0001642502,0.000049250975,0.00011190568],"category_scores_gemma":[0.0063972143,0.00018700452,0.00082289544,0.18207024,0.000068297864,0.00015707966,0.00008538743,0.00038555468,0.0000065446247],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013791623,0.00016360001,0.8781399,0.000100563004,0.0008461732,0.00016557648,0.00013704252,0.00005935857,0.000007959628,4.4276862e-7,0.06537896,0.054862518],"study_design_scores_gemma":[0.0006788829,0.00009021748,0.9957317,0.000029613138,0.002978309,0.00001942129,0.00007781449,0.000043507214,0.0000021324765,0.000005075134,0.00020801397,0.00013528057],"about_ca_topic_score_codex":0.00005323704,"about_ca_topic_score_gemma":0.000018532357,"teacher_disagreement_score":0.117591836,"about_ca_system_score_codex":0.00008069111,"about_ca_system_score_gemma":0.00034025582,"threshold_uncertainty_score":0.9359544},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4413931068","doi":"10.47191/ijsshr/v8-i8-103","title":"Socioeducational Observation in Breathwork: Bibliometrics for Rochrematic","year":2025,"lang":"en","type":"article","venue":"International Journal of Social Science and Human Research","topic":"Resilience and Mental Health","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Computer science; Library science","score_opus":0.2511047470736848,"score_gpt":0.6324138290138817,"score_spread":0.38130908194019686,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413931068","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9768354,0.0001319252,0.00012971347,0.006262229,0.0006456791,0.00017675608,0.0000037805705,0.0000019981621,0.01581248],"genre_scores_gemma":[0.99737173,0.000026023547,0.00016398758,0.00028656246,0.0002779892,0.000018117673,0.0000024502633,0.0000023145126,0.0018507923],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9980656,0.000059410566,0.00032544584,0.00013299492,0.0011690385,0.0002475201],"domain_scores_gemma":[0.9974949,0.0002799227,0.00011624277,0.00004309427,0.002008949,0.000056920893],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005308268,0.00004149803,0.00009808705,0.0052909334,0.00041949068,0.00015577617,0.0005441962,0.00004521523,0.00008046705],"category_scores_gemma":[0.0004070355,0.000037202033,0.000033333883,0.0038111971,0.0004722258,0.00034975674,0.000062081686,0.00021387468,0.000004177775],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013614717,0.00046301133,0.080160275,0.000035244964,0.000030449919,0.000007980761,0.0076277307,0.000001163085,0.001881937,0.79554707,0.014791975,0.099317014],"study_design_scores_gemma":[0.0007643381,0.00013105012,0.93382263,0.00007665342,0.0000026927444,0.0000091088295,0.006030651,0.000017696235,0.00010615649,0.053879626,0.0051161554,0.000043238488],"about_ca_topic_score_codex":0.00011666391,"about_ca_topic_score_gemma":0.000024171557,"teacher_disagreement_score":0.8536624,"about_ca_system_score_codex":0.00048881513,"about_ca_system_score_gemma":0.00081134547,"threshold_uncertainty_score":0.4721045},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4414039474","doi":"10.1016/j.rineng.2025.107123","title":"Global insights into biomass pyrolysis mechanisms: A scientometric and mechanistic approach","year":2025,"lang":"en","type":"article","venue":"Results in Engineering","topic":"Thermochemical Biomass Conversion Processes","field":"Engineering","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thornhill Medical (Canada); University of Saskatchewan","funders":"Natural Sciences and Engineering Research Council of Canada; Mitacs","keywords":"Biomass (ecology); Pyrolysis; Biochemical engineering; Environmental science; Data science; Computer science; Biology; Engineering; Ecology; Chemical engineering","score_opus":0.005303110259795996,"score_gpt":0.2079178775032309,"score_spread":0.2026147672434349,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414039474","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6001764,0.005088215,0.3882539,0.000050917595,0.00068981946,0.0003324168,0.000021875281,0.0010844385,0.0043020356],"genre_scores_gemma":[0.9895331,0.00008602113,0.010279149,0.000011446439,0.000013284474,0.000027248592,0.000010151652,0.000018044673,0.000021542977],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9987848,0.000009232346,0.00032181406,0.0003672292,0.00020247273,0.00031445833],"domain_scores_gemma":[0.99950665,0.000094971365,0.00002192924,0.00024199196,0.000033977027,0.00010045794],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00017365224,0.00022758567,0.0002467386,0.0011589447,0.00003711983,0.00006304966,0.0002634024,0.00015167103,0.0000023865732],"category_scores_gemma":[0.00031355498,0.00023398119,0.000037783167,0.0053034197,0.000026216832,0.00012651717,0.00013391262,0.00011872071,0.0000045790666],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000089908455,0.000112982176,0.000116680436,0.0029673001,0.00022747241,0.000044508623,0.00067700114,0.039373543,0.9117947,0.029410243,0.00019067351,0.014994963],"study_design_scores_gemma":[0.0021815016,0.00003451237,0.00079993444,0.0003402207,0.000060658134,0.000009548299,0.0002332407,0.49101374,0.49591765,0.007939425,0.0007065328,0.0007630278],"about_ca_topic_score_codex":0.000019705418,"about_ca_topic_score_gemma":0.0000043090554,"teacher_disagreement_score":0.45164022,"about_ca_system_score_codex":0.00029740797,"about_ca_system_score_gemma":0.000020393061,"threshold_uncertainty_score":0.9541476},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4414067311","doi":"10.1111/jep.70257","title":"Current Trends and Future Directions of Statistical Methods in Medical Research: A Scientometric Analysis","year":2025,"lang":"en","type":"article","venue":"Journal of Evaluation in Clinical Practice","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Biostatistics; Scientometrics; Medical research; Thematic analysis; MEDLINE; Focus (optics); Statistical analysis","score_opus":0.6668623805292179,"score_gpt":0.7728677648608374,"score_spread":0.10600538433161955,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414067311","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7030899,0.043202303,0.042841747,0.19471839,0.0071257204,0.000783411,0.0000062189624,0.00000903346,0.008223294],"genre_scores_gemma":[0.9464912,0.017540814,0.034820165,0.00037171802,0.0007130415,0.000014213654,0.0000056171116,0.0000047302847,0.000038496208],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9892755,0.0059925206,0.002540829,0.00021611908,0.0018010713,0.00017393247],"domain_scores_gemma":[0.95241475,0.042319212,0.00075316784,0.00021917862,0.0040204907,0.00027320426],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.097637504,0.000058436868,0.00046749212,0.0063877054,0.000051082585,0.000019429615,0.00008099569,0.00016891093,0.0005475662],"category_scores_gemma":[0.24029367,0.000046985744,0.0001106429,0.01441553,0.00021459209,0.00021830028,0.000032940847,0.0014959284,0.0000015716935],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003850167,0.0010809627,0.10484878,0.000036697507,0.000099312376,0.00000707409,0.0003978412,0.000021294174,0.0000027955507,0.00077062723,0.00058935897,0.89176023],"study_design_scores_gemma":[0.00044285288,0.0005553226,0.9174512,0.00027789414,0.0010702088,0.000028745915,0.0026955544,0.02452004,0.000018978682,0.0045498856,0.04834408,0.000045218436],"about_ca_topic_score_codex":0.00024288222,"about_ca_topic_score_gemma":0.00021458766,"teacher_disagreement_score":0.89171505,"about_ca_system_score_codex":0.00025492813,"about_ca_system_score_gemma":0.0019214097,"threshold_uncertainty_score":0.9291721},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4414086455","doi":"10.3389/fspor.2025.1609141","title":"Trends and hotspots in running shoe research: a bibliometric study from 2005 to 2024","year":2025,"lang":"en","type":"review","venue":"Frontiers in Sports and Active Living","topic":"Lower Extremity Biomechanics and Pathologies","field":"Engineering","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Key (lock); Bibliometrics; Trend analysis; Work (physics)","score_opus":0.05677332567464193,"score_gpt":0.3435534554716822,"score_spread":0.28678012979704026,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414086455","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0055079213,0.9910027,0.000108097694,0.0000062845947,0.0011091164,0.00058038416,0.0002471701,0.000054117307,0.0013842237],"genre_scores_gemma":[0.005952643,0.9918933,0.0009818196,0.000005149898,0.000088419416,0.00015478296,0.000027536058,0.000046503414,0.00084983144],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978437,0.000088685134,0.00048088288,0.00070606684,0.0003143687,0.0005663169],"domain_scores_gemma":[0.9990862,0.00040760738,0.000062840154,0.00029358888,0.0000248589,0.00012488935],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0010911409,0.00040452837,0.001373529,0.058189627,0.00007184378,0.00016668682,0.00024205564,0.0002895199,0.00010656848],"category_scores_gemma":[0.00015884303,0.00037195356,0.000083151295,0.032328676,0.000034577555,0.0001713338,0.0003928743,0.0009167517,9.287786e-7],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004598655,0.000053628875,0.012161202,0.0011031234,0.00005234548,0.00019849061,0.0010275266,0.000005070729,1.8441823e-7,0.0000010133083,0.0072327214,0.9781601],"study_design_scores_gemma":[0.00033728636,0.00014372636,0.15523908,0.04403119,0.00022285811,0.000008018999,0.0062117204,0.0005109517,3.7837708e-7,0.00018888558,0.7918902,0.0012156846],"about_ca_topic_score_codex":0.00016789579,"about_ca_topic_score_gemma":0.0001711017,"teacher_disagreement_score":0.9769444,"about_ca_system_score_codex":0.00023237473,"about_ca_system_score_gemma":0.00005485741,"threshold_uncertainty_score":0.9998732},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4414092465","doi":"10.47115/bsagriculture.1692279","title":"Bibliometric Analysis of Poppy Alkaloids and Morphine Biosynthesis","year":2025,"lang":"en","type":"article","venue":"Black Sea Journal of Agriculture","topic":"Berberine and alkaloids research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Poppy; Thematic analysis; Citation; Bibliometrics; Web of science; Opium Poppy","score_opus":0.012924003731983762,"score_gpt":0.29806488057051983,"score_spread":0.28514087683853606,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414092465","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98789924,0.005233605,0.000049675506,0.0041100536,0.000038790837,0.00011173077,0.000021158676,0.000006987,0.0025287664],"genre_scores_gemma":[0.99347186,0.0033159389,0.0006189209,0.000179385,0.0001235249,7.735709e-7,0.00000904593,0.000006219496,0.002274348],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9984857,0.00005376631,0.0005359808,0.00015186607,0.00056761113,0.00020507273],"domain_scores_gemma":[0.9982527,0.00018129977,0.00025685565,0.00017495456,0.00093257445,0.00020160206],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00043341768,0.00014940918,0.00075270643,0.04512343,0.00004015022,0.000032238713,0.0001588138,0.0001322207,0.00017827279],"category_scores_gemma":[0.00042207856,0.000082798106,0.00038158393,0.13448387,0.00010347162,0.00008892467,0.000095445204,0.0003744644,0.0000029293267],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006639803,0.0010356723,0.4486929,0.00069272885,0.013080111,0.00028317087,0.00024622018,0.00013691103,0.37370557,0.0003346402,0.111490496,0.0496376],"study_design_scores_gemma":[0.0010672674,0.00038361596,0.95710415,0.00021007349,0.003277833,0.000096638025,0.00028054533,0.000107894164,0.031154253,0.00007080761,0.006148881,0.00009806845],"about_ca_topic_score_codex":0.000053277923,"about_ca_topic_score_gemma":0.000014370411,"teacher_disagreement_score":0.5084112,"about_ca_system_score_codex":0.000040631956,"about_ca_system_score_gemma":0.00010547217,"threshold_uncertainty_score":0.96569926},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4414150891","doi":"10.3389/frobt.2025.1605682","title":"A comprehensive review and bibliometric analysis on collaborative robotics for industry: safety emerging as a core focus","year":2025,"lang":"en","type":"review","venue":"Frontiers in Robotics and AI","topic":"Robot Manipulation and Learning","field":"Engineering","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"École de Technologie Supérieure; Institut de recherche Robert-Sauvé en santé et en sécurité du travail; Toronto Metropolitan University","funders":"Institut de Recherche Robert-Sauvé en Santé et en Sécurité du Travail; Toronto Metropolitan University","keywords":"Categorization; Bibliometrics; Process (computing); Focus (optics); Selection (genetic algorithm); Perspective (graphical); Descriptive statistics; Key (lock)","score_opus":0.036124274408642056,"score_gpt":0.3364154105429346,"score_spread":0.30029113613429254,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414150891","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[2.457589e-7,0.9017086,0.09616635,0.00018909092,0.00037230935,0.0011852576,0.000036048878,0.000051273302,0.00029083458],"genre_scores_gemma":[0.000015425867,0.9881356,0.011003872,0.00021539228,0.000046152436,0.000069228365,0.00018108287,0.000051679206,0.00028157167],"study_design_codex":"simulation_or_modeling","study_design_gemma":"not_applicable","domain_scores_codex":[0.99821436,0.000100936595,0.0007318336,0.0004496579,0.00017420159,0.00032897937],"domain_scores_gemma":[0.9989101,0.00033112068,0.00021000874,0.0002654316,0.00015654413,0.00012678404],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.000222939,0.00049553363,0.0025274705,0.024202915,0.00012629364,0.00010715098,0.00015574534,0.0005082284,0.000009316719],"category_scores_gemma":[0.0002053374,0.0004695041,0.00024559288,0.06283093,0.000046408775,0.00007382016,0.0000689805,0.00094011915,0.0000013297673],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000047153376,0.000014083294,0.00017225956,0.037124027,0.001179022,0.000006612027,0.00003706566,0.54684085,6.5266104e-9,0.0005400268,0.007994787,0.40608656],"study_design_scores_gemma":[0.0004931361,0.00008445435,0.00015496765,0.025245028,0.007882333,0.00000408003,0.00011509207,0.12440282,3.415765e-8,0.00015192121,0.8406989,0.0007672499],"about_ca_topic_score_codex":0.0000054363236,"about_ca_topic_score_gemma":0.0000038696644,"teacher_disagreement_score":0.8327041,"about_ca_system_score_codex":0.00014422576,"about_ca_system_score_gemma":0.000103504935,"threshold_uncertainty_score":0.99977565},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4414218259","doi":"10.31235/osf.io/q6fpw_v1","title":"University journals in the global academic publishing landscape: Mapping over 19,000 diverse titles","year":2025,"lang":"en","type":"article","venue":"","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Publishing; Quarter (Canadian coin); Discipline; Identification (biology); Scholarly communication; Web of science; Electronic publishing; Bibliometrics","score_opus":0.5837607990015768,"score_gpt":0.5647070565081341,"score_spread":0.019053742493442716,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414218259","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5554465,0.0011342462,0.003950924,0.011589526,0.0004393417,0.00016047277,0.0000459561,0.000021038784,0.427212],"genre_scores_gemma":[0.98836285,0.00045619122,0.00015833256,0.0012652768,0.000045878212,3.7584374e-7,0.0000012003367,0.0000013742132,0.009708535],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9941576,0.00033894536,0.00033509007,0.00037365608,0.0043896353,0.00040503318],"domain_scores_gemma":[0.99549323,0.0031922942,0.00011343745,0.00039983503,0.0006341915,0.0001669848],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.022779569,0.000081521706,0.00015584107,0.030712442,0.00029148304,0.0049103587,0.003731572,0.00014429503,0.00223156],"category_scores_gemma":[0.021882504,0.000046660163,0.000092768765,0.21890557,0.00007754541,0.0017724135,0.001048568,0.0005408898,0.00012371388],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000006300517,0.000017621212,0.693204,0.0000011432255,0.000004998056,0.00002074836,0.00009106597,0.00001856672,0.000008424889,0.0077793957,0.27242893,0.026418852],"study_design_scores_gemma":[0.0003462648,0.000005941982,0.6951081,0.00001064719,0.0000021474395,0.0000028751938,0.0067037046,0.0016110721,0.0000030333756,0.01176576,0.2843808,0.000059646496],"about_ca_topic_score_codex":0.0008331632,"about_ca_topic_score_gemma":0.0002425889,"teacher_disagreement_score":0.43291634,"about_ca_system_score_codex":0.00020945922,"about_ca_system_score_gemma":0.00021449092,"threshold_uncertainty_score":0.99868053},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4414249228","doi":"10.5539/hes.v15n4p203","title":"Research Analysis and Trend Development of Grit Among College Students: A Bibliometric Study Using CiteSpace","year":2025,"lang":"en","type":"article","venue":"Higher Education Studies","topic":"Grit, Self-Efficacy, and Motivation","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Grit; Field (mathematics); Psychological resilience; Bibliometrics; Realm; Trend analysis; Web of science; Citation analysis","score_opus":0.18703557309886779,"score_gpt":0.5313995904779323,"score_spread":0.3443640173790645,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414249228","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9904393,0.0064398986,0.000041151983,0.00016302532,0.00090288714,0.0005445767,0.000004219924,0.000024416106,0.001440556],"genre_scores_gemma":[0.9833705,0.000072021176,0.0005467304,0.000029661984,0.000057253532,0.00018911746,0.0000067078677,0.000009370208,0.015718678],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99797624,0.00037611756,0.00046378735,0.00043100867,0.00050898537,0.00024383808],"domain_scores_gemma":[0.998118,0.00054280943,0.00015492208,0.00030833838,0.00082423905,0.000051688017],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0012131919,0.00014663028,0.00038795374,0.03326464,0.00036784745,0.000046913796,0.0001776852,0.00006393125,0.00009350336],"category_scores_gemma":[0.000076273434,0.00013433823,0.000046104375,0.076976605,0.00014295192,0.00009108868,0.00017744988,0.00013859176,0.000005438785],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000029793691,0.0030483436,0.96940196,0.000043106174,0.003486955,7.057272e-7,0.015890962,0.0000021218525,0.000039976734,0.0006791972,0.0053711412,0.0020057252],"study_design_scores_gemma":[0.0006508944,0.00005796781,0.904287,0.00003821953,0.0004552155,1.3415548e-7,0.09111338,5.7472147e-7,0.000031341224,0.00003838303,0.0032335517,0.000093350885],"about_ca_topic_score_codex":0.00029386356,"about_ca_topic_score_gemma":0.0003529943,"teacher_disagreement_score":0.07522242,"about_ca_system_score_codex":0.00017251304,"about_ca_system_score_gemma":0.00013870858,"threshold_uncertainty_score":0.9776925},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4414394245","doi":"10.3390/ph18091422","title":"Tracing Five Decades of Psoriasis Pharmacotherapy: A Large-Scale Bibliometric Investigation with AI-Guided Terminology Normalization","year":2025,"lang":"en","type":"article","venue":"Pharmaceuticals","topic":"Psoriasis: Treatment and Pathogenesis","field":"Immunology and Microbiology","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Terminology; Citation; Normalization (sociology); Citation impact; Bibliometrics; Citation analysis; Scopus; Alliance; Psoriasis","score_opus":0.040435010012030065,"score_gpt":0.3493872740884505,"score_spread":0.3089522640764204,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414394245","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9869169,0.006016493,0.0039449707,0.0008161453,0.00055668363,0.0005273625,0.00009177522,0.00012400927,0.0010056491],"genre_scores_gemma":[0.99584913,0.0009333629,0.00050114613,0.0011648464,0.000036775593,0.00010033145,0.00013300397,0.000020660062,0.0012607368],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9982434,0.00037873082,0.00048547256,0.00037739566,0.00007368464,0.00044131585],"domain_scores_gemma":[0.9990114,0.0003335543,0.00018702121,0.00021784929,0.00020271115,0.000047470578],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00040778457,0.00025165046,0.00043405744,0.006081792,0.00017885632,0.00001915539,0.00021198603,0.0002404482,0.0005118175],"category_scores_gemma":[0.00007109551,0.00020741332,0.00010261677,0.010455007,0.0002157847,0.00021379186,0.00004775575,0.00019092682,0.00006186198],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011208678,0.00065454864,0.21592893,0.00024404319,0.001713398,0.0000110393885,0.0017809797,0.000105457555,0.6580085,0.0009836347,0.0050302767,0.114418305],"study_design_scores_gemma":[0.00633603,0.0001266351,0.021627046,0.000084050225,0.0007226526,0.00002790446,0.00015169095,0.0002523944,0.95770687,0.00025760697,0.012488334,0.00021879267],"about_ca_topic_score_codex":0.000029793015,"about_ca_topic_score_gemma":0.000011974172,"teacher_disagreement_score":0.29969832,"about_ca_system_score_codex":0.000049434315,"about_ca_system_score_gemma":0.000114017595,"threshold_uncertainty_score":0.845807},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4414443776","doi":"10.3389/frwa.2025.1650870","title":"Mapping the research landscape: energy storage of bio-nanoparticle-enhanced phase change materials for solar desalination—a scientometric framework","year":2025,"lang":"en","type":"article","venue":"Frontiers in Water","topic":"Solar-Powered Water Purification Methods","field":"Energy","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Department of Science and Technology, Ministry of Science and Technology, India","keywords":"Phase change; Phase-change material; Desalination; Solar energy; Concentrator; Solar still; Renewable energy; Water desalination","score_opus":0.1037674600492488,"score_gpt":0.38441914883725864,"score_spread":0.28065168878800983,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414443776","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.55401605,0.0010264263,0.4395912,0.0012937387,0.002784872,0.0007172062,0.000028985018,0.000058710084,0.00048281255],"genre_scores_gemma":[0.95950544,0.000060436352,0.038198605,0.0001862481,0.00017134294,0.00075930817,0.000050724313,0.000031954038,0.0010359673],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99696636,0.0008483686,0.0005743796,0.00044136043,0.00050817424,0.00066134793],"domain_scores_gemma":[0.998341,0.00036144842,0.00010010328,0.0006589153,0.00048121254,0.000057302314],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0057018553,0.00017123448,0.00034680832,0.00247019,0.00025463637,0.00013233554,0.000642919,0.0002113852,0.00011365046],"category_scores_gemma":[0.0009859222,0.0001156318,0.0000730602,0.0034418064,0.00020678225,0.00022416476,0.00015300167,0.00018699662,0.000009613022],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007237375,0.0006199926,0.0036196066,0.00032590362,0.00020718967,0.0000048688107,0.014602119,0.00034454264,0.8692279,0.021097206,0.013607414,0.07561954],"study_design_scores_gemma":[0.0012277908,0.000062074694,0.0007922625,0.00008876805,0.000013553118,3.1340966e-7,0.000833732,0.0017177792,0.9270453,0.027220754,0.040851664,0.00014600085],"about_ca_topic_score_codex":0.00016076905,"about_ca_topic_score_gemma":0.000009593054,"teacher_disagreement_score":0.40548936,"about_ca_system_score_codex":0.00018084866,"about_ca_system_score_gemma":0.00005140745,"threshold_uncertainty_score":0.47153282},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4414461396","doi":"10.15517/1qke4780","title":"Aprendizaje motor en las ciencias del movimiento humano: un análisis bibliométrico","year":2025,"lang":"en","type":"article","venue":"Pensar en Movimiento Revista de Ciencias del Ejercicio y la Salud","topic":"Physical Education and Pedagogy","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Subject (documents); Bibliometrics; Thematic analysis; Thematic map; Scientific literature","score_opus":0.049569558079019894,"score_gpt":0.42280752462847204,"score_spread":0.3732379665494521,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414461396","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.883111,0.0023205965,0.016354883,0.0035093583,0.0019622343,0.0027043077,0.00026521337,0.00076314626,0.08900927],"genre_scores_gemma":[0.9576822,0.0008624353,0.002900726,0.014969119,0.0008798989,0.00065098907,0.00015649929,0.00013736512,0.02176075],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9905397,0.002833012,0.0017214809,0.0015393498,0.0012526285,0.0021137807],"domain_scores_gemma":[0.99262464,0.0032386375,0.00064006366,0.0016875782,0.0008803314,0.00092875375],"candidate_categories":["metaepi_narrow","sts","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0034399326,0.00080882525,0.0011370021,0.0038414875,0.0017694057,0.00028308193,0.0019937085,0.0007688362,0.0023508028],"category_scores_gemma":[0.0040638233,0.0007719964,0.0005287083,0.00876109,0.00036619953,0.00066601206,0.0009149636,0.001758469,0.0015021099],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007541541,0.0047220583,0.17693168,0.0027863206,0.0005725575,0.00020572527,0.06735177,0.00063517055,0.034055818,0.29401463,0.3559921,0.061978012],"study_design_scores_gemma":[0.0027479338,0.000260586,0.3687399,0.0008492421,0.0004607589,0.000022507234,0.008975622,0.0061625144,0.0010348018,0.005907262,0.6032623,0.0015766008],"about_ca_topic_score_codex":0.0012213854,"about_ca_topic_score_gemma":0.00014015396,"teacher_disagreement_score":0.28810737,"about_ca_system_score_codex":0.0015942209,"about_ca_system_score_gemma":0.0026708678,"threshold_uncertainty_score":0.99953014},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4414488125","doi":"10.18280/ijsdp.200801","title":"Digital Twin Integration in Field Service Management: A Hybrid Systematic Literature Review and Bibliometric Analysis (2017-2024)","year":2025,"lang":"en","type":"article","venue":"International Journal of Sustainable Development and Planning","topic":"Digital Transformation in Industry","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Field (mathematics); Systematic review; Service (business); Bibliometrics; Scientific literature","score_opus":0.009056725453668492,"score_gpt":0.24799094140601682,"score_spread":0.23893421595234834,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414488125","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5440962,0.17627399,0.07232379,0.001978387,0.0012881517,0.0013512841,0.000028463062,0.000112700545,0.20254709],"genre_scores_gemma":[0.98937416,0.007769881,0.00065252994,0.00035116373,0.000022045091,0.000017400025,0.000036244594,0.00000665477,0.0017699338],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.9990112,0.000010952401,0.0005574189,0.00007558999,0.00022400984,0.00012080859],"domain_scores_gemma":[0.9993483,0.00010054378,0.000099238176,0.000044256503,0.00037049392,0.000037156457],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00035100812,0.00011008553,0.00024977245,0.021399733,0.000026402911,0.0004826306,0.00016027948,0.00003931479,0.000007817023],"category_scores_gemma":[0.00009427885,0.00009741752,0.00003222528,0.014836255,0.000004944802,0.0011011625,0.000040174928,0.00019040666,9.525786e-7],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027287012,0.0002554776,0.06403317,0.477849,0.017287165,0.015252534,0.01915253,0.03023855,0.000009392403,0.008557604,0.024488272,0.34260342],"study_design_scores_gemma":[0.00638388,0.00016097841,0.14060192,0.63390696,0.003691432,0.002074501,0.10927616,0.025772298,0.0009578999,0.0028061585,0.071828865,0.0025389672],"about_ca_topic_score_codex":6.5482016e-7,"about_ca_topic_score_gemma":5.260177e-7,"teacher_disagreement_score":0.445278,"about_ca_system_score_codex":0.00010696739,"about_ca_system_score_gemma":0.00002410856,"threshold_uncertainty_score":0.98969185},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4414556258","doi":"10.1111/joes.70022","title":"Exploring Non‐Fungible Tokens: A Bibliometric Analysis and Future Research Opportunities","year":2025,"lang":"en","type":"article","venue":"Journal of Economic Surveys","topic":"Blockchain Technology Applications and Security","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University Canada West","funders":"","keywords":"Scopus; Bibliometrics; Patent analysis; Web of science; Bibliographic coupling","score_opus":0.23212056030678627,"score_gpt":0.3548915132959937,"score_spread":0.12277095298920745,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414556258","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92724794,0.0022451025,0.06385105,0.0045860386,0.0003865026,0.00006939139,0.000003365187,0.000023768267,0.0015868653],"genre_scores_gemma":[0.98841643,0.007982743,0.0032797544,0.00004364046,0.00011276693,0.000011084614,5.631496e-7,0.0000031509169,0.00014985989],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99879974,0.0002747035,0.00041677052,0.00019021351,0.00011263771,0.0002059573],"domain_scores_gemma":[0.99859065,0.00049678556,0.00017967675,0.0004095308,0.00024673247,0.00007660322],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.007945399,0.00007888495,0.0002955131,0.067192905,0.0001722628,0.00017990805,0.000758557,0.000075553035,0.000018173001],"category_scores_gemma":[0.000056885186,0.00007241399,0.000101658436,0.039895963,0.00008055988,0.00057173555,0.00026941928,0.0003456352,0.0000049023415],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009552657,0.00011544614,0.102831185,0.000029389132,0.0010981881,0.000032449538,0.00048021338,0.00012248964,0.000082389364,0.11557515,0.003413016,0.77621055],"study_design_scores_gemma":[0.00059808954,0.00016570382,0.9231838,0.000026243082,0.000106223364,0.000046278128,0.0012808685,0.014362847,0.001245812,0.018706858,0.04004304,0.00023421542],"about_ca_topic_score_codex":0.00009305019,"about_ca_topic_score_gemma":0.00006808182,"teacher_disagreement_score":0.8203526,"about_ca_system_score_codex":0.00008625087,"about_ca_system_score_gemma":0.00018405075,"threshold_uncertainty_score":0.9805115},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4414612712","doi":"10.56975/jetir.v12i9.569690","title":"Yoga's Evolution in Sports Science:A Bibliometric Assessment of Research Trends","year":2025,"lang":"en","type":"article","venue":"Journal of Emerging Technologies and Innovative Research","topic":"Martial Arts: Techniques, Psychology, and Education","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Thematic analysis; Convergence (economics); Bibliometrics; Web of science; Thematic map","score_opus":0.1524954193811528,"score_gpt":0.5876061246323365,"score_spread":0.4351107052511837,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414612712","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92351294,0.0009909285,0.0006136748,0.016596092,0.00037053972,0.0001848046,8.522162e-7,0.000049852973,0.057680316],"genre_scores_gemma":[0.9941977,0.0032814161,0.002071271,0.0000072561584,0.000050503913,0.00001932117,2.385682e-7,0.0000035360476,0.0003687425],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.996408,0.00039945857,0.0006063337,0.00023947563,0.0017467195,0.0006000074],"domain_scores_gemma":[0.99402565,0.00041856227,0.0002446233,0.00022285884,0.005047383,0.000040950155],"candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.03583732,0.000073850584,0.00023705846,0.115457706,0.0006972555,0.0000889902,0.0007525509,0.0001708548,0.000046937002],"category_scores_gemma":[0.006307669,0.00006247701,0.000029275747,0.2287328,0.004084654,0.0005026869,0.00033696942,0.001223079,2.7856575e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000037422127,0.0002466057,0.28658223,0.000036804562,0.00001102867,0.000006618356,0.0014666584,0.0000013409855,0.0037629544,0.25895405,0.009452862,0.4394414],"study_design_scores_gemma":[0.0002900236,0.00043632372,0.7736614,0.0003570627,0.0000021344235,0.000002584502,0.09128042,0.000043650827,0.0019356498,0.109791905,0.02210853,0.000090354355],"about_ca_topic_score_codex":0.0008890715,"about_ca_topic_score_gemma":0.00010151937,"teacher_disagreement_score":0.48707914,"about_ca_system_score_codex":0.0007304737,"about_ca_system_score_gemma":0.00155734,"threshold_uncertainty_score":0.99862564},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4414652547","doi":"10.1097/md.0000000000043936","title":"Global trends and hotspots of randomized controlled trials in chronic pain therapy (1989–2024): A 35-year bibliometric analysis","year":2025,"lang":"en","type":"article","venue":"Medicine","topic":"Musculoskeletal pain and rehabilitation","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Chronic pain; Randomized controlled trial; Psychological intervention; Clinical trial; Alternative medicine; MEDLINE; Multimodal therapy","score_opus":0.017806518837868972,"score_gpt":0.3619643420340218,"score_spread":0.3441578231961528,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414652547","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9182446,0.05513364,0.0026428252,0.013267097,0.00032841958,0.0026946333,0.000007642391,0.000039437433,0.007641685],"genre_scores_gemma":[0.99227303,0.0051945667,0.00013244206,0.0004107658,0.00014610947,0.00015360516,0.000028215843,0.0000068236955,0.0016544168],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9937264,0.0032692323,0.0019692765,0.00031825795,0.000501545,0.0002153357],"domain_scores_gemma":[0.98998076,0.008987436,0.000461742,0.000305724,0.00015833038,0.0001060096],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.033252332,0.00021891357,0.006547284,0.06434538,0.000024542966,0.000008202479,0.00007531415,0.00014222748,0.00085437135],"category_scores_gemma":[0.026570627,0.00011505832,0.0010160271,0.12215239,0.00036844684,0.00003284392,0.000016371245,0.00012552258,7.726023e-7],"study_design_candidate":"randomized_trial","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.2783373,0.00062036485,0.22154437,0.0016163668,0.014696259,0.000014824402,0.0005988388,0.0000783616,0.0068494715,0.001415641,0.0055936393,0.46863455],"study_design_scores_gemma":[0.47480285,0.00072071806,0.5170832,0.0005568392,0.002444575,3.924707e-7,0.00015031791,0.00352579,0.000012722345,0.0003810568,0.00024543414,0.00007606912],"about_ca_topic_score_codex":0.00051461166,"about_ca_topic_score_gemma":0.000106197556,"teacher_disagreement_score":0.4685585,"about_ca_system_score_codex":0.00016005422,"about_ca_system_score_gemma":0.00015229198,"threshold_uncertainty_score":0.99547017},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4414772345","doi":"10.1007/978-3-031-98880-6_23","title":"Experiential Learning Using AR-VR: A Scientometric Analysis","year":2025,"lang":"en","type":"book-chapter","venue":"Lecture notes in networks and systems","topic":"Virtual Reality Applications and Impacts","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Experiential learning; Bridging (networking); Scopus; Virtual reality; Educational technology; Experiential education","score_opus":0.02680655730499053,"score_gpt":0.2839094408407314,"score_spread":0.25710288353574084,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414772345","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00024076483,0.00864646,0.9749683,0.000037803813,0.00049455976,0.00030169656,0.0000041415374,0.00006031348,0.015245921],"genre_scores_gemma":[0.9924918,0.0003201717,0.001176357,0.00007696411,0.0002909165,0.000016759795,0.000027148295,0.000017644983,0.005582259],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9979658,0.00007421901,0.00048575638,0.000725819,0.00037389767,0.0003744838],"domain_scores_gemma":[0.99854064,0.00033982628,0.00028362384,0.0006019481,0.000105166044,0.00012879015],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0005575786,0.00031440466,0.0006658647,0.0022118874,0.00024460637,0.0007420455,0.0004967169,0.00044089122,0.000009553652],"category_scores_gemma":[0.00010402851,0.00028072245,0.00017794485,0.002997107,0.00007016563,0.00013426141,0.00029762092,0.0005729386,0.0000013880125],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000005164523,0.000014112139,0.0003413126,0.000060888236,0.00027210964,0.00001153786,0.0005224395,0.8680062,0.000010392569,0.05087152,0.000043117998,0.0798412],"study_design_scores_gemma":[0.00011186479,0.00003436603,0.000056528927,0.00019990133,0.00012588856,0.000009057171,0.0000056445197,0.9844691,0.000001912509,0.0007791613,0.013920451,0.00028610125],"about_ca_topic_score_codex":0.00029644943,"about_ca_topic_score_gemma":0.00008331205,"teacher_disagreement_score":0.99225104,"about_ca_system_score_codex":0.00014127421,"about_ca_system_score_gemma":0.000088486624,"threshold_uncertainty_score":0.9999645},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4414845510","doi":"10.24106/kefdergi.1795795","title":"Investigation of Studies on Micro-credentials by Bibliometric Analysis Method","year":2025,"lang":"en","type":"article","venue":"Kastamonu Eğitim Dergisi","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Higher education; Web of science; Bibliometrics; Quality (philosophy); Lifelong learning; Trend analysis","score_opus":0.5019674545803363,"score_gpt":0.6055027287630623,"score_spread":0.10353527418272601,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414845510","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92824155,0.011195901,0.05153052,0.0018622269,0.0012057385,0.0004086111,0.00016493667,0.00004997346,0.0053405142],"genre_scores_gemma":[0.97899395,0.0010257825,0.008183179,0.00047408068,0.000034441506,0.000026203415,0.000021839574,0.000013361381,0.01122717],"study_design_codex":"not_applicable","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.987456,0.0010320401,0.0018205907,0.0011825422,0.007884017,0.0006247993],"domain_scores_gemma":[0.9733805,0.017878167,0.00090036646,0.0013797773,0.0061048064,0.00035637812],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.028333465,0.00028072207,0.0010605515,0.7972701,0.0003012258,0.0013359486,0.0022716487,0.00018439742,0.00033847708],"category_scores_gemma":[0.055380043,0.00020868397,0.00053961284,0.9449048,0.00035326218,0.000538807,0.0007760555,0.00027717397,0.00013348593],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010475155,0.00031444547,0.3335643,0.00008462988,0.0028382074,0.000016330863,0.00045569075,0.0022047455,0.06225613,0.004121067,0.48751828,0.106521405],"study_design_scores_gemma":[0.001576747,0.00051544764,0.3935974,0.00009506459,0.0007637748,0.0000023620287,0.0026083814,0.010184634,0.52845174,0.042198386,0.019319307,0.0006867495],"about_ca_topic_score_codex":0.00019967853,"about_ca_topic_score_gemma":0.000026545786,"teacher_disagreement_score":0.46819898,"about_ca_system_score_codex":0.00016804031,"about_ca_system_score_gemma":0.00023383881,"threshold_uncertainty_score":0.9997008},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4414870645","doi":"10.1097/md.0000000000045001","title":"A global perspective: Trends and insights in premenstrual disorder comorbidity research by bibliometric analysis (1999–2023)","year":2025,"lang":"en","type":"review","venue":"Medicine","topic":"Menstrual Health and Disorders","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Comorbidity; Bibliometrics; MEDLINE; Mental health; Psychiatric comorbidity; National Comorbidity Survey","score_opus":0.14027991564547568,"score_gpt":0.5257244308848328,"score_spread":0.3854445152393571,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414870645","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0010026984,0.96973336,0.000019624416,0.0016189488,0.00024557472,0.0011578294,0.00017307012,0.000046413345,0.026002487],"genre_scores_gemma":[0.0023506493,0.9930565,0.000035775574,0.00018820491,0.00016464666,0.00017266038,0.0008999046,0.000026862854,0.0031048409],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9942474,0.00081209396,0.0012376677,0.0012336974,0.0015454141,0.00092373695],"domain_scores_gemma":[0.9970159,0.00078232447,0.00025127834,0.0007894114,0.00043219392,0.00072889955],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0018177861,0.0006216855,0.0040321825,0.13049471,0.00017984165,0.000026447366,0.00030342795,0.0005892684,0.00062654546],"category_scores_gemma":[0.002146487,0.00043421972,0.00031776936,0.37980315,0.0006593157,0.000067711815,0.00021748293,0.0012495182,0.000016024977],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008972489,0.00033588932,0.0056526503,0.017631138,0.0006229176,0.00009316514,0.0002196998,1.3173965e-7,3.837477e-8,0.00025128637,0.041045655,0.9340577],"study_design_scores_gemma":[0.0031989594,0.00096664374,0.017412836,0.010179007,0.0041944413,0.000037906888,0.0011376335,0.00003470035,2.1498998e-8,0.00019394369,0.96230066,0.00034325648],"about_ca_topic_score_codex":0.012680534,"about_ca_topic_score_gemma":0.00342658,"teacher_disagreement_score":0.93371445,"about_ca_system_score_codex":0.0010820655,"about_ca_system_score_gemma":0.0010503129,"threshold_uncertainty_score":0.99981093},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4414947210","doi":"10.51176/1997-9967-2025-3-68-84","title":"Exploring Women’s Migration through Bibliometrics: Trends and Research Networks","year":2025,"lang":"en","type":"article","venue":"The economy strategy and practice","topic":"Migration and Labor Dynamics","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Latin Americans; Globalization; Thematic map; Thematic analysis; Sample (material); Scientific literature; Bibliometrics","score_opus":0.23765454257507446,"score_gpt":0.4349279863369911,"score_spread":0.19727344376191663,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414947210","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.28873694,0.0020366984,0.0010323536,0.039616115,0.0002003372,0.0001555909,0.0000020070831,0.000039172046,0.66818076],"genre_scores_gemma":[0.9699168,0.021212284,0.00008736986,0.00067936647,0.00015333296,0.0001015471,0.0000021968867,0.0000034856046,0.007843619],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99862903,0.00074189436,0.00013355573,0.00016103039,0.00009712686,0.0002373319],"domain_scores_gemma":[0.9971081,0.002532513,0.000060121376,0.00011293903,0.00013256283,0.000053782194],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0044982415,0.000060276547,0.000077743345,0.00068255165,0.0010148183,0.0006024139,0.00010911849,0.000060484705,0.00012183596],"category_scores_gemma":[0.0006055532,0.00004883853,0.000010756873,0.0044224593,0.00023589046,0.0018965625,0.00003436316,0.00027347263,0.0000045045185],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000073680065,0.00002693545,0.0003983524,0.000007891752,0.000037359645,0.0000011715873,0.037340872,0.00024314997,0.0000019887043,0.86825836,0.0025941315,0.091016136],"study_design_scores_gemma":[0.00012434389,0.00004003123,0.0017414438,0.00000574959,0.000011080077,0.000001308127,0.08347614,0.0017183501,0.0000022276429,0.0099365655,0.902878,0.00006475362],"about_ca_topic_score_codex":0.0017699159,"about_ca_topic_score_gemma":0.00418892,"teacher_disagreement_score":0.9002839,"about_ca_system_score_codex":0.000061171886,"about_ca_system_score_gemma":0.00010888257,"threshold_uncertainty_score":0.78052646},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4415055368","doi":"10.31083/ijp44192","title":"Understanding the Opioid Overdose Crisis: A Comprehensive Bibliometric Analysis","year":2025,"lang":"en","type":"article","venue":"International Journal of Pharmacology","topic":"Opioid Use Disorder Treatment","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Opioid overdose; Thematic analysis; Citation; Drug overdose; Medical prescription; Bibliometrics; Citation analysis; Opioid","score_opus":0.0628334564196168,"score_gpt":0.4044182605392684,"score_spread":0.3415848041196516,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415055368","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8935096,0.0069716643,0.036956266,0.053955384,0.005684272,0.00035005144,0.000025319263,0.000025436399,0.002521965],"genre_scores_gemma":[0.9905266,0.0014314633,0.000285825,0.0073486054,0.00023879576,0.0000058261603,0.0000071234317,0.0000064255155,0.0001493189],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9987569,0.000084192354,0.0004405944,0.000120007055,0.00045655496,0.00014173784],"domain_scores_gemma":[0.99844545,0.00034749962,0.00028752405,0.00008462797,0.0007770964,0.00005781214],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00018536126,0.0001105397,0.0003174986,0.03526739,0.00005485058,0.0000385479,0.00031851666,0.000048258695,0.0005034357],"category_scores_gemma":[0.000056319324,0.00007354202,0.00034496255,0.023511488,0.00006157819,0.00009021436,0.00008202897,0.00025655862,0.0000100883],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0057785106,0.004130161,0.29739633,0.0001365507,0.161278,0.0030626839,0.0034576182,0.0064353924,0.035473023,0.009890538,0.46097133,0.0119898515],"study_design_scores_gemma":[0.026131727,0.0012112921,0.8021139,0.00020886512,0.031708688,0.0009166233,0.010941222,0.0054608416,0.016462805,0.010498193,0.09394697,0.0003988672],"about_ca_topic_score_codex":0.00002630204,"about_ca_topic_score_gemma":0.0000025310344,"teacher_disagreement_score":0.5047176,"about_ca_system_score_codex":0.00060653203,"about_ca_system_score_gemma":0.0002767381,"threshold_uncertainty_score":0.99724424},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4415209608","doi":"10.3390/jrfm18100578","title":"Exploring the Research Landscape of Impact Investing and Sustainable Finance: A Bibliometric Review","year":2025,"lang":"en","type":"review","venue":"Journal of risk and financial management","topic":"Community Development and Social Impact","field":"Economics, Econometrics and Finance","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Sustainability; Scopus; Corporate governance; Sustainable development; Bibliometrics; Investment (military); China; Sustainable business; Corporate social responsibility","score_opus":0.21844903260375226,"score_gpt":0.3708091702664023,"score_spread":0.15236013766265005,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415209608","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012740885,0.994915,0.00011341806,0.000053938264,0.00018030008,0.00059404556,0.000031860625,0.0000034810364,0.0028338626],"genre_scores_gemma":[0.001461325,0.99774057,0.0001831697,0.000023276085,0.00010949584,0.00003314183,0.0000030911924,0.000016085423,0.0004298457],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975344,0.00024415323,0.0014232649,0.00020449741,0.00015731677,0.0004363482],"domain_scores_gemma":[0.9970382,0.0008868738,0.0014765722,0.00030492476,0.00021834287,0.0000751028],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0095533095,0.00025061914,0.0018482694,0.015032842,0.0004574788,0.0001499237,0.0005213909,0.000094052884,0.000013112064],"category_scores_gemma":[0.0027033396,0.00018176467,0.0003919954,0.02224137,0.00011110153,0.00034459986,0.0006649001,0.000947415,0.000003143687],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000007560444,0.000036045618,0.0010082221,0.03111107,0.00009985961,0.000021706575,0.00034465548,6.1459997e-7,6.454217e-10,0.012423576,0.0031367613,0.95180994],"study_design_scores_gemma":[0.00024107752,0.00012658804,0.011396687,0.025019402,0.00024433853,0.00001132206,0.00037621986,0.0000011880662,1.59509e-8,0.0070833964,0.95530945,0.00019034182],"about_ca_topic_score_codex":0.0002590901,"about_ca_topic_score_gemma":0.000007677134,"teacher_disagreement_score":0.95217264,"about_ca_system_score_codex":0.00014477037,"about_ca_system_score_gemma":0.00021676719,"threshold_uncertainty_score":0.99854136},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4415213069","doi":"10.1108/jts-04-2025-0016","title":"Digital technologies in the circular economy: a bibliometric analysis","year":2025,"lang":"en","type":"article","venue":"Journal of Trade Science","topic":"Sustainable Supply Chain Management","field":"Business, Management and Accounting","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Trois-Rivières","funders":"","keywords":"Scopus; Transparency (behavior); Thematic analysis; Thematic map; Key (lock); Bibliometrics; Resource (disambiguation); Product (mathematics); Emerging technologies","score_opus":0.011660300928890015,"score_gpt":0.2430567101728237,"score_spread":0.2313964092439337,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415213069","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7674837,0.0018532218,0.01134485,0.10458316,0.00058482046,0.0005472663,8.544918e-7,0.00010945067,0.113492675],"genre_scores_gemma":[0.9990421,0.000015393713,0.00009262176,0.000740221,0.000052661348,0.0000032256987,2.3616424e-7,0.0000023929495,0.00005118017],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9987653,0.000006148519,0.0003531096,0.00016784832,0.0004326745,0.0002749345],"domain_scores_gemma":[0.9992349,0.000084975334,0.0002808202,0.00026260864,0.00013133485,0.000005371977],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0023834675,0.00008885378,0.00019171438,0.13448787,0.00014500387,0.0016299391,0.0016223004,0.000025575197,0.000009813808],"category_scores_gemma":[0.0009014448,0.00005871179,0.0001404604,0.31580472,0.000262105,0.003032904,0.00021158901,0.00017885398,0.000004522377],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000025188563,0.00040316526,0.75407076,0.00034100498,0.0003853006,0.0004865349,0.00018372214,0.00986577,0.0001324578,0.07100721,0.0077191372,0.15537976],"study_design_scores_gemma":[0.00078787404,0.000030416295,0.6470403,0.00006397871,0.0005314665,0.0000149918,0.027677894,0.014698607,0.000067238594,0.068408735,0.24038006,0.00029843417],"about_ca_topic_score_codex":0.000012143085,"about_ca_topic_score_gemma":0.0000028830164,"teacher_disagreement_score":0.23266093,"about_ca_system_score_codex":0.00013043328,"about_ca_system_score_gemma":0.00009390923,"threshold_uncertainty_score":0.99940646},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4415279421","doi":"10.1136/bmjebm-2024-113391","title":"Over 1000 terms have been used to describe evidence synthesis: a scoping review","year":2025,"lang":"en","type":"review","venue":"BMJ evidence-based medicine","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; Public Health Ontario; Impact; McGill University; McMaster University; Queen's University; St. Michael's Hospital","funders":"National Health and Medical Research Council","keywords":"Identification (biology); Scope (computer science); Context (archaeology); Key (lock); Perspective (graphical); Systematic review","score_opus":0.8775033471675885,"score_gpt":0.6368221377890774,"score_spread":0.24068120937851112,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415279421","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[8.759518e-7,0.9527014,0.008381511,0.013847043,0.0010144606,0.022646582,0.000044808723,0.00004861874,0.0013147134],"genre_scores_gemma":[0.000007998706,0.9704414,0.001847973,0.009013877,0.0010499612,0.006707755,0.000024336901,0.00009310302,0.010813584],"study_design_codex":"design_other","study_design_gemma":"systematic_review","domain_scores_codex":[0.88622254,0.04753166,0.039871857,0.005409221,0.019610357,0.0013543525],"domain_scores_gemma":[0.74084526,0.20902778,0.024465704,0.020533154,0.0035096635,0.0016184441],"candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","open_science","insufficient_payload"],"consensus_categories":["metaresearch","metaepi_narrow","insufficient_payload"],"category_scores_codex":[0.3014744,0.0023899167,0.04077239,0.0031505285,0.00036539513,0.0009912113,0.009891462,0.00063581776,0.035333283],"category_scores_gemma":[0.669393,0.0010985795,0.0088324305,0.008785521,0.00037554116,0.0009702796,0.00081817375,0.0010060173,0.006871364],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010215979,0.000025684618,0.0000939474,0.27074444,0.00029141066,0.000082081926,0.000055181783,0.0000033242286,6.2057137e-7,0.000036922338,0.19921944,0.52943677],"study_design_scores_gemma":[0.00008993633,0.000102957165,0.000012688538,0.58312535,0.009521466,0.000026730228,0.000026125486,0.000029561545,0.00000140388,0.000041581472,0.40643367,0.0005885561],"about_ca_topic_score_codex":0.0001247146,"about_ca_topic_score_gemma":0.0001829948,"teacher_disagreement_score":0.5288482,"about_ca_system_score_codex":0.0006311312,"about_ca_system_score_gemma":0.005275949,"threshold_uncertainty_score":0.99914646},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"methods","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4415297929","doi":"10.33545/26647559.2025.v7.i2e.266","title":"Global research trends in physical literacy and their efficacy on motor competence: A bibliometric study","year":2025,"lang":"","type":"article","venue":"International Journal of Sports Health and Physical Education","topic":"Children's Physical and Motor Development","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Psychosocial; Pace; Psychological intervention; Publishing; Literacy; Grassroots; Cognition; Physical fitness","score_opus":0.036545626951065426,"score_gpt":0.4757765406260185,"score_spread":0.43923091367495304,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415297929","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9842214,0.0018529518,0.000008552102,0.008989965,0.0028352914,0.00055443327,0.000038521426,0.000006834526,0.0014920601],"genre_scores_gemma":[0.9958963,0.00068558584,0.000032688604,0.0010170302,0.0019678874,0.000025766309,0.000026629185,0.00001346186,0.00033469297],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99630225,0.00046373036,0.0010759684,0.0006139853,0.0010364092,0.00050767825],"domain_scores_gemma":[0.99714583,0.00067380327,0.0005865064,0.00024138059,0.000756907,0.00059556105],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0011194131,0.00034561337,0.00075825735,0.015006116,0.00017954687,0.00023038051,0.0004516346,0.000085357926,0.00003893302],"category_scores_gemma":[0.00015554685,0.00027577896,0.000173289,0.015801098,0.00021257398,0.00021537047,0.00015628747,0.00096852897,0.000005283689],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019159125,0.024064668,0.07441683,0.00005002138,0.00019974919,0.000017806437,0.021180877,0.0000052584896,0.000010993827,0.009700914,0.0017187938,0.8667182],"study_design_scores_gemma":[0.0028937382,0.0025952503,0.9705569,0.00091173744,0.000019343157,0.00003241511,0.002851669,0.00014024998,0.0000059077606,0.015116307,0.0046918145,0.00018467374],"about_ca_topic_score_codex":0.00039298096,"about_ca_topic_score_gemma":0.000008472749,"teacher_disagreement_score":0.89614004,"about_ca_system_score_codex":0.000617019,"about_ca_system_score_gemma":0.0019021531,"threshold_uncertainty_score":0.9999694},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4415444643","doi":"10.22329/jtl.v19i4.10096","title":"AI and Transformative Learning in Higher Education: A Systematic Literature Review and Bibliometric Insights","year":2025,"lang":"en","type":"article","venue":"Journal of Teaching and Learning","topic":"Engineering Education and Technology","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Transformative learning; Metacognition; Systematic review; Higher education; Formative assessment; Critical thinking; Theme (computing)","score_opus":0.008535725542472406,"score_gpt":0.28247284606610373,"score_spread":0.27393712052363134,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415444643","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0878545,0.85665417,0.03014119,0.023128942,0.00044164012,0.00019887088,4.7585182e-8,0.00007938399,0.0015012568],"genre_scores_gemma":[0.97590095,0.018245762,0.003913653,0.0008828224,0.000020787518,0.000004031048,1.9523384e-7,0.0000033281965,0.0010284844],"study_design_codex":"design_other","study_design_gemma":"systematic_review","domain_scores_codex":[0.9991722,0.00030576994,0.00027884543,0.00009137285,0.000076154756,0.000075672884],"domain_scores_gemma":[0.99949133,0.00020342988,0.00014299923,0.000057946923,0.0000644071,0.000039898405],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0010572354,0.00007585193,0.00025427842,0.006348785,0.0001338442,0.00016885917,0.00011116696,0.00005142171,0.0000010639865],"category_scores_gemma":[0.00046663597,0.000057453835,0.00001923198,0.004432554,0.000013039322,0.00043446769,0.000030882857,0.0014262942,1.8988926e-7],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000108664835,0.00027924788,0.0146480305,0.16789791,0.0002588512,0.000047957892,0.054042052,0.0005322039,0.00035340132,0.18937415,0.0021965022,0.5703588],"study_design_scores_gemma":[0.003251644,0.0012798343,0.16455628,0.59282255,0.00055779493,0.004475136,0.005410016,0.026294632,0.00006056797,0.007069835,0.19290003,0.0013216728],"about_ca_topic_score_codex":0.0000019043571,"about_ca_topic_score_gemma":2.0718984e-7,"teacher_disagreement_score":0.88804644,"about_ca_system_score_codex":0.000020701318,"about_ca_system_score_gemma":0.000048307316,"threshold_uncertainty_score":0.6196618},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4415488869","doi":"10.5195/jmla.2025.2024","title":"Analyzing the citation impact of predatory journals in the health sciences","year":2025,"lang":"en","type":"article","venue":"Journal of the Medical Library Association JMLA","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Citation; Citation impact; Bibliometrics; Citation analysis; Biomedical sciences; Human science; Impact factor","score_opus":0.3545117205238558,"score_gpt":0.5962054485031559,"score_spread":0.2416937279793001,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415488869","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.77637255,0.0050648097,0.00089246687,0.21322457,0.0010470191,0.0001995833,0.000014741493,0.000002949535,0.003181332],"genre_scores_gemma":[0.99660707,0.00092140574,0.000088784924,0.0015568734,0.00015652677,0.0000010131406,2.5907661e-7,0.0000023815073,0.0006657107],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9814297,0.0037022152,0.0016204872,0.00015201837,0.012767392,0.00032818152],"domain_scores_gemma":[0.9682972,0.026510986,0.0033802495,0.00039101334,0.0012013172,0.00021923218],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.12823614,0.000075442054,0.00031640526,0.012088595,0.0004255431,0.0014260847,0.004608341,0.000098857265,0.00048412755],"category_scores_gemma":[0.13513643,0.000025825635,0.00041684837,0.10345039,0.00020098251,0.0013079507,0.00034565016,0.000849087,0.0000033811275],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017169305,0.000090230766,0.83257526,0.0000030682447,0.000041326257,0.0000017247424,0.00039648567,0.0002219643,0.000013665094,0.0007813813,0.13865915,0.027198585],"study_design_scores_gemma":[0.0003295137,0.00016129823,0.9680702,0.00010175011,0.0000063290886,0.0000072644007,0.0008250229,0.0021516075,0.00003432058,0.024019709,0.004262577,0.000030401934],"about_ca_topic_score_codex":0.000028941824,"about_ca_topic_score_gemma":0.0000051395195,"teacher_disagreement_score":0.22023451,"about_ca_system_score_codex":0.00024092697,"about_ca_system_score_gemma":0.0029893566,"threshold_uncertainty_score":0.99961054},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W4415585632","doi":"10.1177/20552076251389341","title":"Science mapping of COVID-19 contributions in primary health care by OECD countries: A machine learning approach","year":2025,"lang":"en","type":"article","venue":"Digital Health","topic":"COVID-19 epidemiological studies","field":"Mathematics","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"York University; University of Waterloo; University of Toronto","funders":"Türkiye Bilimsel ve Teknolojik Araştırma Kurumu","keywords":"Transparency (behavior); Health care; Field (mathematics); Primary care; Primary health care; Global health; Health science","score_opus":0.09723233866229378,"score_gpt":0.43239594124446357,"score_spread":0.3351636025821698,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415585632","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12495574,0.077000044,0.5825259,0.1514458,0.0004416555,0.008287717,0.0068656267,0.0013401886,0.0471373],"genre_scores_gemma":[0.98272955,0.0004967417,0.0015005653,0.014778059,0.000014627728,0.00005670512,0.0001944695,0.000011909588,0.0002173524],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9971696,0.00022429305,0.0009778478,0.0005083941,0.00037701742,0.00074280537],"domain_scores_gemma":[0.99625653,0.0025843014,0.000441531,0.0002777542,0.00014150336,0.00029835638],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0028168424,0.00020268354,0.000883293,0.00034677653,0.000654074,0.000053618787,0.00033120924,0.00006846501,0.0000052393775],"category_scores_gemma":[0.00896396,0.00017481414,0.00007947956,0.0011456462,0.00055122905,0.00019126182,0.00034490085,0.00037762825,0.0000037723698],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029943662,0.0018475184,0.33039764,0.03416231,0.00013164345,0.000008628694,0.0278517,0.0009609934,0.000032663727,0.507737,0.06765641,0.028914101],"study_design_scores_gemma":[0.0067581637,0.001898615,0.024275538,0.0013685597,0.000024820767,0.000014792164,0.0163928,0.0039107804,0.000031488646,0.10360784,0.84043133,0.001285287],"about_ca_topic_score_codex":0.0013908574,"about_ca_topic_score_gemma":0.000119820914,"teacher_disagreement_score":0.85777384,"about_ca_system_score_codex":0.0045082956,"about_ca_system_score_gemma":0.0053834748,"threshold_uncertainty_score":0.999384},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4415963683","doi":"10.1007/s11192-025-05454-4","title":"Exploring citation behavior of national self-references","year":2025,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Université du Québec à Montréal","funders":"National Social Science Fund of China; China Association for Science and Technology","keywords":"Scarcity; Diversification (marketing strategy); Citation; Norm (philosophy); Scientific communication; Scientific literature; Process (computing); Citation analysis","score_opus":0.9021341018802934,"score_gpt":0.6353747233350729,"score_spread":0.26675937854522047,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415963683","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9445034,0.00048809196,0.0051457994,0.00029339036,0.0021146338,0.00029332392,0.000045143996,0.000053872544,0.04706232],"genre_scores_gemma":[0.9933284,0.00022687599,0.004123298,0.000062128136,0.00003659438,0.000052040272,0.0000046964137,0.0000052344535,0.0021607112],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.977176,0.00017827883,0.0009974019,0.0007611617,0.020352291,0.00053485157],"domain_scores_gemma":[0.97846484,0.006223836,0.00038878236,0.0005898213,0.014047294,0.00028543413],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.035325885,0.00014516068,0.00031364223,0.30197138,0.00028975127,0.0014558379,0.0027081121,0.00008786057,0.00022962842],"category_scores_gemma":[0.09542064,0.000110957604,0.00016684853,0.6710516,0.00020678963,0.0014334747,0.0006702067,0.0002096313,0.00016715277],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013803305,0.0007082309,0.60989475,0.000022860902,0.000031537693,0.0000032128714,0.00046674686,0.00012362843,0.0018990248,0.036331072,0.008715275,0.34178984],"study_design_scores_gemma":[0.0004929491,0.00012400521,0.9564057,0.000015417325,0.000013739445,0.0000012996403,0.0007169114,0.0026593816,0.0069878306,0.0112402495,0.021148512,0.00019397971],"about_ca_topic_score_codex":0.000039603172,"about_ca_topic_score_gemma":0.0000060099815,"teacher_disagreement_score":0.36908025,"about_ca_system_score_codex":0.0002837151,"about_ca_system_score_gemma":0.0007206229,"threshold_uncertainty_score":0.99958074},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4416063431","doi":"10.35631/ijepc.1059032","title":"THE EFFECTIVENESS OF FASTING IN MANAGING SEXUAL DESIRE: A BIBLIOMETRIC ANALYSIS FROM AN ISLAMIC PERSPECTIVE (2015–2024)","year":2025,"lang":"","type":"article","venue":"International Journal of Education Psychology and Counseling","topic":"Dietary Effects on Health","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Islam; Perspective (graphical); Inclusion (mineral); Subject (documents); English language","score_opus":0.018766555342522296,"score_gpt":0.43034815255703013,"score_spread":0.4115815972145078,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416063431","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9236212,0.04705047,0.015219024,0.0039737923,0.008762865,0.00030426835,0.000012434223,0.000004781573,0.0010511898],"genre_scores_gemma":[0.9884492,0.009730234,0.0004678581,0.0007688385,0.00046133887,0.000010819376,0.000014033604,0.000013457159,0.00008418738],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9958763,0.001526829,0.0012937062,0.00044237252,0.0005856202,0.000275178],"domain_scores_gemma":[0.9909701,0.0049323468,0.0010532868,0.0002925837,0.0026516034,0.00010008677],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.007053264,0.00022827463,0.0006640603,0.040451765,0.00014493303,0.000108588254,0.0005297113,0.00017230283,0.00005658101],"category_scores_gemma":[0.001829707,0.00019597162,0.00015435684,0.02696654,0.00023950143,0.00038826282,0.00003884455,0.0007961564,0.0000019904674],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.010581854,0.0029493533,0.4717083,0.00032503207,0.010070899,0.000076564924,0.006007464,0.0018431519,0.0040271264,0.007391556,0.00019540479,0.4848233],"study_design_scores_gemma":[0.0024731217,0.00071484665,0.93815726,0.0014742924,0.0016309171,0.00008385376,0.014014806,0.004432115,0.000085122476,0.03669368,0.00008534195,0.00015466807],"about_ca_topic_score_codex":0.0009540123,"about_ca_topic_score_gemma":0.00016393988,"teacher_disagreement_score":0.4846686,"about_ca_system_score_codex":0.00044474326,"about_ca_system_score_gemma":0.0012986818,"threshold_uncertainty_score":0.99371576},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4416088174","doi":"10.2196/75909","title":"User Experience in mHealth Research: Bibliometric Analysis of Trends and Developments (2007–2023)","year":2025,"lang":"en","type":"review","venue":"JMIR mhealth and uhealth","topic":"Mobile Health and mHealth Applications","field":"Health Professions","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"mHealth; User experience design; Scope (computer science); Comparability; eHealth; Health informatics; User engagement; Health care; Usability","score_opus":0.39070597494134596,"score_gpt":0.6210898968786436,"score_spread":0.23038392193729762,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416088174","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002160331,0.9831939,0.000025187386,0.00069803203,0.00069744565,0.007659172,0.0006741499,0.00009586614,0.004795921],"genre_scores_gemma":[0.00087527075,0.9741199,0.00061379425,0.0012393792,0.00023158893,0.017969893,0.00081373914,0.000084751766,0.004051656],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9824665,0.004508349,0.005894152,0.0021325953,0.0012837073,0.0037147065],"domain_scores_gemma":[0.98693484,0.0055650044,0.002478788,0.0014953971,0.0006919033,0.002834081],"candidate_categories":["metaepi_narrow","bibliometrics","sts","research_integrity"],"consensus_categories":["bibliometrics","research_integrity"],"category_scores_codex":[0.010326237,0.00091197534,0.005761052,0.18874145,0.002066851,0.000035521778,0.0008173826,0.0016254333,0.0004951458],"category_scores_gemma":[0.00089427596,0.0008112652,0.0002970038,0.27498958,0.00044946733,0.00021619756,0.0006482046,0.0042185863,0.00005220076],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000065307824,0.00021699943,0.012772106,0.14787818,0.00009141956,0.000004082742,0.0010647335,2.3494428e-7,1.1317614e-8,0.0019931393,0.010799986,0.8251138],"study_design_scores_gemma":[0.00083541335,0.00016502611,0.13778967,0.008638555,0.00075477647,0.000003194045,0.00054730347,0.000019169303,1.0260719e-8,0.000058197315,0.8508218,0.0003668965],"about_ca_topic_score_codex":0.004650209,"about_ca_topic_score_gemma":0.0027524063,"teacher_disagreement_score":0.8400218,"about_ca_system_score_codex":0.0015883878,"about_ca_system_score_gemma":0.016399983,"threshold_uncertainty_score":0.9996707},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4416382208","doi":"10.1186/s41065-025-00544-y","title":"Mapping the global research landscape of mitophagy in Parkinson's disease: a bibliometric and visualization analysis","year":2025,"lang":"en","type":"article","venue":"Hereditas","topic":"Parkinson's Disease Mechanisms and Treatments","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Shanghai Municipal Health Commission","keywords":"Mitophagy; Translational research; Visualization; Mechanism (biology); Web of science; Metadata; Autophagy","score_opus":0.04732019749832104,"score_gpt":0.3680146712625875,"score_spread":0.32069447376426646,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416382208","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97853076,0.018497478,0.00045555367,0.00044343466,0.00008016327,0.00033228184,0.0000646802,0.00001775643,0.001577881],"genre_scores_gemma":[0.99890494,0.0006747275,0.00007341609,0.00006952104,0.000043753782,0.000045655455,0.000054995955,0.000005127479,0.00012783229],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99882334,0.00013924889,0.00019925596,0.0002640361,0.00035969494,0.00021442452],"domain_scores_gemma":[0.9992961,0.00013470954,0.0000457842,0.00026220953,0.00014110649,0.00012004004],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00044309808,0.00009393608,0.00024561433,0.011538656,0.00007653115,0.000030373865,0.000078948055,0.000051395473,0.00007385891],"category_scores_gemma":[0.0002502605,0.000066156135,0.000083014726,0.07277128,0.000059961854,0.000047693142,0.000073364645,0.00007648322,0.000004078737],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016155491,0.00027571633,0.9890551,0.00012481313,0.00049728487,0.00008887277,0.000049427086,0.000008676989,0.00002089348,0.0012701466,0.0011949664,0.007252558],"study_design_scores_gemma":[0.0011079024,0.000047037982,0.98600566,0.00009698875,0.00034752348,0.0000014681486,0.00029714959,0.0022573534,0.000055424385,0.0015259257,0.00821222,0.000045368703],"about_ca_topic_score_codex":0.00015326183,"about_ca_topic_score_gemma":0.00006364441,"teacher_disagreement_score":0.061232626,"about_ca_system_score_codex":0.000079975216,"about_ca_system_score_gemma":0.00014745572,"threshold_uncertainty_score":0.9996647},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4416444389","doi":"10.1163/21659214-bja10144","title":"Mapping the Landscape of Digital Religion Studies: A Bibliometric Analysis","year":2025,"lang":"","type":"article","venue":"Journal of Religion Media and Digital Culture","topic":"Media, Religion, Digital Communication","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Quarter (Canadian coin); Bibliometrics; Citation analysis; TRACE (psycholinguistics); Cover (algebra); Web of science; Perspective (graphical)","score_opus":0.04124338590917121,"score_gpt":0.28361444018209897,"score_spread":0.24237105427292777,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416444389","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3714776,0.48552895,0.0003702759,0.004733446,0.002130938,0.00061195996,0.00032059549,0.000047156147,0.13477908],"genre_scores_gemma":[0.83222264,0.16487972,0.000027448496,0.00026617554,0.00050424325,0.0000056234126,0.000070210845,0.000023042405,0.0020008835],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9951841,0.00014045191,0.002499021,0.00040943615,0.0013044255,0.00046253015],"domain_scores_gemma":[0.99180275,0.001892769,0.002408455,0.0007542902,0.0028523533,0.00028936582],"candidate_categories":["metaepi_narrow","bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0008864358,0.0005865588,0.0015649027,0.030575536,0.00059577107,0.0034433338,0.0010163356,0.00031579696,0.000014218292],"category_scores_gemma":[0.0037752201,0.0003510387,0.0011610021,0.041016195,0.0011695945,0.0025295955,0.00047746196,0.0008909985,0.000010083016],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013554826,0.0015325674,0.051412378,0.0025468632,0.021586524,0.00012893745,0.14874578,0.00025447804,0.00008786984,0.11744397,0.5808664,0.07403874],"study_design_scores_gemma":[0.006320108,0.0031727194,0.007401942,0.00710032,0.009841349,0.00038381352,0.05799823,0.0005319576,0.00019031725,0.36439407,0.5407043,0.001960831],"about_ca_topic_score_codex":0.000022318984,"about_ca_topic_score_gemma":0.000038386257,"teacher_disagreement_score":0.46074504,"about_ca_system_score_codex":0.00010535895,"about_ca_system_score_gemma":0.0003076718,"threshold_uncertainty_score":0.99989414},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4416455603","doi":"10.1051/bioconf/202519602003","title":"Bibliometric insights and content analysis of diatom paleolimnology in lakes: Global perspectives and Indonesian contributions over the last decade","year":2025,"lang":"fr","type":"article","venue":"BIO Web of Conferences","topic":"Geology and Paleoclimatology Research","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Universitas Diponegoro","keywords":"Paleolimnology; Diatom; China; Biogeosciences; Scopus; Trend analysis; Limnology; Bibliometrics","score_opus":0.033025617585553564,"score_gpt":0.29547609471679104,"score_spread":0.2624504771312375,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416455603","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.85443723,0.13818924,0.000019517058,0.0030600484,0.00015526237,0.00018149985,0.0002707437,0.000004066493,0.0036824052],"genre_scores_gemma":[0.97823715,0.021591948,0.000008302571,0.000048262704,0.000009893791,0.0000036030476,0.00002869222,6.416776e-7,0.000071511866],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9981286,0.00063156046,0.00041702104,0.00031887152,0.0001630046,0.00034094922],"domain_scores_gemma":[0.99769133,0.0016497907,0.00017348486,0.00016189294,0.0002400753,0.00008340095],"candidate_categories":["bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.00053036225,0.00015808652,0.0006348068,0.009280163,0.0001445783,0.000036874426,0.00026193823,0.00033095872,0.00036104803],"category_scores_gemma":[0.0004219673,0.00010942037,0.00008141296,0.02650377,0.0033946154,0.00010672659,0.000059862134,0.00023980183,0.000002743791],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000076379365,0.00007120859,0.9271483,0.000050317307,0.0007729899,0.000007786002,0.00041347413,0.0000128385145,0.000004697317,0.0691948,0.000021198748,0.0022260111],"study_design_scores_gemma":[0.0006650583,0.00016560267,0.989273,0.000056254798,0.00044037262,0.0000070748197,0.0054861307,0.0017978655,0.000026099366,0.0013326549,0.0006576133,0.00009227926],"about_ca_topic_score_codex":0.002631244,"about_ca_topic_score_gemma":0.1632316,"teacher_disagreement_score":0.16060036,"about_ca_system_score_codex":0.0000034414093,"about_ca_system_score_gemma":0.0007097364,"threshold_uncertainty_score":0.9993176},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4416717765","doi":"10.1097/md.0000000000046028","title":"Research trends in cardiac rehabilitation following COVID-19: A cross-sectional bibliometric study","year":2025,"lang":"en","type":"article","venue":"Medicine","topic":"Cardiac Health and Mental Health","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Rehabilitation; Observational study; Thematic analysis; Bibliometrics; Thematic map; MEDLINE","score_opus":0.12354214659899412,"score_gpt":0.5683219313532956,"score_spread":0.44477978475430147,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416717765","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9644651,0.0018682708,0.000027872424,0.015649347,0.0017125919,0.0014091483,0.0000070288975,0.00006622317,0.014794459],"genre_scores_gemma":[0.99386156,0.000081895705,0.000070781534,0.0022107253,0.00042542195,0.00028232398,0.00007576189,0.000014488216,0.0029770667],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9961376,0.00065294193,0.00074922503,0.0005349581,0.0013270045,0.00059827155],"domain_scores_gemma":[0.99663347,0.0019720756,0.000049434395,0.00041831713,0.00020667186,0.00072001683],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.012956473,0.00014319137,0.0006265586,0.06676966,0.00027942623,0.000015551268,0.00009042224,0.00012522422,0.00020716198],"category_scores_gemma":[0.0061981967,0.000119186865,0.00015623352,0.09942711,0.00017614354,0.00007760382,0.00006340897,0.00065712334,0.000019497513],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00057129195,0.00045224087,0.9668355,0.00044824518,0.000030354888,0.000065417975,0.0011518303,0.0000041447242,0.00007006772,0.00006210926,0.013207445,0.017101336],"study_design_scores_gemma":[0.00726258,0.0025003543,0.9596041,0.00015511124,0.00003462181,0.0000071186423,0.0038815797,0.000018667693,0.000005037672,0.00012920592,0.026336275,0.000065338885],"about_ca_topic_score_codex":0.005731513,"about_ca_topic_score_gemma":0.0002260589,"teacher_disagreement_score":0.032657452,"about_ca_system_score_codex":0.0019191086,"about_ca_system_score_gemma":0.0011305086,"threshold_uncertainty_score":0.9438077},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4416738373","doi":"10.2147/dhps.s557043","title":"Current Status and Research Trends in Deprescribing: A Bibliometric Review","year":2025,"lang":"en","type":"article","venue":"Drug Healthcare and Patient Safety","topic":"Global Public Health Policies and Epidemiology","field":"Business, Management and Accounting","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Medical and Health Research Project of Zhejiang Province; National Key Research and Development Program of China","keywords":"Scope (computer science); Deprescribing; Field (mathematics); Bibliometrics; Set (abstract data type)","score_opus":0.10674283052773612,"score_gpt":0.4237422660224196,"score_spread":0.31699943549468346,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416738373","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.22831431,0.42625213,0.000010000688,0.328387,0.0009148272,0.0010146253,0.000031281288,0.00007723719,0.014998616],"genre_scores_gemma":[0.7481875,0.19470806,0.0000208122,0.056686234,0.00016364324,0.000072173614,0.00009595023,0.0000122924075,0.000053359145],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9974518,0.00022952461,0.0007051094,0.00043148245,0.00026246905,0.0009196359],"domain_scores_gemma":[0.99896055,0.00030507988,0.00013614127,0.00024096202,0.00026929163,0.00008800036],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0023235828,0.0001602766,0.00042771743,0.0156048965,0.00032665647,0.00006715185,0.000119627926,0.00007625425,0.00003083257],"category_scores_gemma":[0.00073278585,0.00013717778,0.000038829898,0.034758653,0.00008647065,0.00031422338,0.00038595457,0.00047753972,0.0000128358],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000029744295,0.000039505067,0.12143888,0.0036315767,0.000003558055,0.0000014993107,0.00006559053,0.0000027094795,6.108883e-8,0.02228603,0.20145261,0.65104824],"study_design_scores_gemma":[0.000267051,0.000016491282,0.21994789,0.0010676391,0.0000051879097,4.6119519e-7,0.00016365369,0.00027329064,4.654449e-8,0.0012985035,0.77687204,0.00008776742],"about_ca_topic_score_codex":0.044216305,"about_ca_topic_score_gemma":0.0018427755,"teacher_disagreement_score":0.65096045,"about_ca_system_score_codex":0.00010600681,"about_ca_system_score_gemma":0.00009204784,"threshold_uncertainty_score":0.99555236},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4416756659","doi":"10.15517/3g5zkb09","title":"A bibliometric analysis of the effects of aerobic exercise on overweight (1978–2025)","year":2025,"lang":"pt","type":"article","venue":"Pensar en Movimiento Revista de Ciencias del Ejercicio y la Salud","topic":"Physical Activity and Health","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Cardiorespiratory fitness; Overweight; Aerobic exercise; Citation; Intervention (counseling); Public health; Multidisciplinary approach","score_opus":0.021489249915775995,"score_gpt":0.3232939357752875,"score_spread":0.3018046858595115,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416756659","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9840958,0.0077021224,0.0011117618,0.00041856966,0.00047729758,0.0016799219,0.000088421606,0.000045574034,0.004380535],"genre_scores_gemma":[0.99248344,0.003176352,0.00010549208,0.0010025926,0.00008724223,0.000052844174,0.000021994489,0.000038740403,0.003031308],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9937901,0.0011624802,0.0013556493,0.0010092526,0.001660357,0.0010221585],"domain_scores_gemma":[0.9916683,0.0045142686,0.0009414227,0.001910067,0.0005359304,0.00043003453],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00221259,0.0006349389,0.0026668173,0.046304043,0.00027141714,0.0000933541,0.0011751421,0.00046273659,0.00014444682],"category_scores_gemma":[0.0041996157,0.00048062418,0.0015128809,0.21344467,0.0005374478,0.0001987212,0.00050568464,0.0010510102,0.000042986412],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004430667,0.036943387,0.54304314,0.05304269,0.011937013,0.00035248674,0.015296158,0.0026623297,0.12899928,0.0690496,0.02488471,0.10935854],"study_design_scores_gemma":[0.00209706,0.0011733046,0.93946236,0.007615463,0.015185126,0.0000038216144,0.00024653203,0.007695755,0.023430908,0.00022861733,0.0023559886,0.000505065],"about_ca_topic_score_codex":0.00067953893,"about_ca_topic_score_gemma":0.00004361513,"teacher_disagreement_score":0.39641923,"about_ca_system_score_codex":0.000852708,"about_ca_system_score_gemma":0.0012638019,"threshold_uncertainty_score":0.99976456},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4416769881","doi":"10.1111/gcbb.70095","title":"Trends and Advancements in Utilization of Biomass Waste for Gasification: A Bibliometric Review","year":2025,"lang":"en","type":"review","venue":"GCB Bioenergy","topic":"Thermochemical Biomass Conversion Processes","field":"Engineering","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Regina","funders":"Universiti Teknologi Petronas","keywords":"Biomass (ecology); Scopus; Thematic map; Bibliometrics; Web of science; Thematic analysis; Citation","score_opus":0.05409953636138511,"score_gpt":0.3354553301976378,"score_spread":0.2813557938362527,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416769881","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000003912538,0.9986139,0.00042607542,0.00001982712,0.00014886608,0.0003057283,0.00011664359,0.00005441752,0.0003106504],"genre_scores_gemma":[0.00007661015,0.99860823,0.00028276144,0.000013813102,0.00001913985,0.0002730751,0.000462394,0.000029057906,0.0002349388],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987115,0.00002764799,0.00066299964,0.0002926002,0.0001267089,0.00017852872],"domain_scores_gemma":[0.9992685,0.00015726674,0.00018600508,0.00026625858,0.000075910444,0.000046071076],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00013310008,0.00026736435,0.0009714683,0.014643092,0.0000128227675,0.000010663504,0.00023030934,0.00018400417,0.000040701787],"category_scores_gemma":[0.00013325145,0.00024155427,0.00014833086,0.04054928,0.000028842212,0.00007397938,0.000053479773,0.00005614516,0.0000014457726],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000013502477,0.000012562046,8.439776e-7,0.17542839,0.000036175767,2.700105e-7,0.00000110612,4.791867e-7,0.00010832473,0.000060993625,0.00072155125,0.82362795],"study_design_scores_gemma":[0.00019239564,0.000020193142,0.0000011475079,0.04423153,0.00023057224,0.0000016433985,0.0000020358239,0.000027343154,0.0027549495,0.000019059522,0.9522971,0.000222038],"about_ca_topic_score_codex":0.0000025976988,"about_ca_topic_score_gemma":0.0000017266411,"teacher_disagreement_score":0.9515755,"about_ca_system_score_codex":0.000067044,"about_ca_system_score_gemma":0.000046020854,"threshold_uncertainty_score":0.9965251},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4416775316","doi":"10.28931/riiad.2025.373","title":"Producción científica sobre adicción a la pornografía: revisión bibliométrica de los últimos 30 años","year":2025,"lang":"es","type":"article","venue":"Revista Internacional de Investigación en Adicciones","topic":"Sexuality, Behavior, and Technology","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Thematic analysis; Pornography; Addiction; Inclusion (mineral); Bibliometrics; Latin Americans; The Internet; Scientific literature","score_opus":0.022667309260028037,"score_gpt":0.3549803540918076,"score_spread":0.3323130448317796,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416775316","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8458152,0.049422927,0.0036663224,0.03663557,0.0040370775,0.003954533,0.001336687,0.0021732866,0.052958433],"genre_scores_gemma":[0.9447345,0.008404813,0.0027500428,0.007361344,0.0017330189,0.0009901506,0.00034214903,0.00034935185,0.033334605],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.98869264,0.002939593,0.0024746463,0.002621779,0.0010231856,0.0022481491],"domain_scores_gemma":[0.9932133,0.0014910083,0.0010496906,0.0026485962,0.00061938964,0.0009780057],"candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"category_scores_codex":[0.0036139411,0.0015271901,0.0019395772,0.0069994116,0.0005338812,0.0007479002,0.0030608498,0.0021386626,0.0036892775],"category_scores_gemma":[0.0032002016,0.0015723405,0.00090685755,0.006762027,0.0019401383,0.000370015,0.001064757,0.0029142264,0.0013741805],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00041707107,0.0017372934,0.1911093,0.0019533888,0.0010542665,0.0010198989,0.0010597033,0.000009064772,0.003869085,0.6663539,0.09239366,0.039023362],"study_design_scores_gemma":[0.002198085,0.00058276026,0.09045222,0.005152078,0.0010431609,0.0012321413,0.0005126644,0.00023417464,0.0012937273,0.008022783,0.88788426,0.0013919804],"about_ca_topic_score_codex":0.00042403507,"about_ca_topic_score_gemma":0.000022126325,"teacher_disagreement_score":0.79549056,"about_ca_system_score_codex":0.00094394694,"about_ca_system_score_gemma":0.0014830761,"threshold_uncertainty_score":0.9997477},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4416777983","doi":"10.25965/reasci.595","title":"La recherche francophone en sciences infirmières : une analyse bibliométrique de 2010 à 2024","year":2025,"lang":"fr","type":"article","venue":"Recherche et Avancées en Sciences Infirmières","topic":"Health Sciences Research and Education","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Trois-Rivières","funders":"","keywords":"French; Context (archaeology); Human science; Nous","score_opus":0.5177952207572991,"score_gpt":0.618883468335819,"score_spread":0.10108824757851986,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416777983","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4278382,0.03528079,0.009961422,0.33986762,0.01874877,0.0026318294,0.00012357096,0.00037051248,0.16517729],"genre_scores_gemma":[0.3742067,0.11632933,0.28006384,0.014914183,0.003632699,0.0009958079,0.000031600674,0.00013011173,0.20969573],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9195079,0.07026557,0.0021154534,0.0021002817,0.0021515077,0.0038592578],"domain_scores_gemma":[0.86084235,0.13479917,0.00076682074,0.0012328462,0.00090458087,0.0014542391],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","sts","research_integrity","insufficient_payload"],"category_scores_codex":[0.19428918,0.00086052914,0.0011651352,0.0070514595,0.0033237075,0.00093756634,0.004070533,0.004371464,0.0024766198],"category_scores_gemma":[0.09992871,0.00074355764,0.00037190257,0.06262807,0.00670589,0.0026992871,0.0011758026,0.011810983,0.0012730315],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020181791,0.0009615521,0.13031985,0.0027877593,0.00019068937,0.0000495619,0.098175116,0.0026928524,0.0058251023,0.021314923,0.5365977,0.20088303],"study_design_scores_gemma":[0.0012739713,0.0010034142,0.09080978,0.0048937,0.00010901883,0.00005254827,0.11460817,0.015171448,0.0033763705,0.11595057,0.65135425,0.0013967528],"about_ca_topic_score_codex":0.06672084,"about_ca_topic_score_gemma":0.031599663,"teacher_disagreement_score":0.32495344,"about_ca_system_score_codex":0.0033810078,"about_ca_system_score_gemma":0.04469196,"threshold_uncertainty_score":0.99950457},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4416794994","doi":"10.1007/s12672-025-04144-0","title":"Research progress and frontier trends in liver cancer immunotherapy in the post-COVID-19 era (2020–2024): a visualization analysis based on bibliometric methods","year":2025,"lang":"en","type":"article","venue":"Discover Oncology","topic":"Cancer Immunotherapy and Biomarkers","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Natural Science Foundation of Ningxia Province","keywords":"Multidisciplinary approach; Frontier; Translational research; Pandemic; Liver cancer; Immunotherapy; Visualization; Cancer immunotherapy; 2019-20 coronavirus outbreak","score_opus":0.05701420962065415,"score_gpt":0.5152336314925254,"score_spread":0.45821942187187126,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416794994","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8780962,0.052123323,0.00689123,0.05435435,0.0007860864,0.0020810028,0.000072541254,0.00007026629,0.0055249976],"genre_scores_gemma":[0.97618175,0.004699247,0.0008546326,0.016098637,0.000062698986,0.0006351106,0.000100664576,0.000027148726,0.001340109],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99652505,0.0018034753,0.0004142857,0.0004835752,0.00039140816,0.00038223376],"domain_scores_gemma":[0.9983587,0.0009842356,0.00009250481,0.0003897043,0.00011403137,0.000060810395],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0038313977,0.00017598936,0.0005133979,0.037676446,0.00009306984,0.00007388176,0.00020983959,0.00022037292,0.0006738422],"category_scores_gemma":[0.00019300093,0.00011935461,0.00015978196,0.0919062,0.00031052786,0.0000967209,0.000041583702,0.00044527208,8.978283e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.011608295,0.0010531959,0.3458446,0.00009574414,0.00068288407,0.00011603282,0.002593711,0.00015920441,0.00069801835,0.00019037782,0.006245474,0.63071245],"study_design_scores_gemma":[0.006514888,0.0010878928,0.9061417,0.000060122384,0.0002098069,0.000003818092,0.0009822586,0.0075618355,0.00012507445,0.000054052154,0.07712388,0.0001346566],"about_ca_topic_score_codex":0.0076970183,"about_ca_topic_score_gemma":0.005003846,"teacher_disagreement_score":0.6305778,"about_ca_system_score_codex":0.0010364315,"about_ca_system_score_gemma":0.0010482407,"threshold_uncertainty_score":0.9989108},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4416827541","doi":"10.11113/ijic.v15n2.540","title":"Recent Trends on Multi-omics Studies in Cancer Research: A Bibliometric Study","year":2025,"lang":"","type":"article","venue":"International Journal of Innovative Computing","topic":"Bioinformatics and Genomic Networks","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Universiti Teknologi Malaysia","keywords":"Bibliometrics; China; Cancer; Web of science; MEDLINE; Citation","score_opus":0.20925023719318095,"score_gpt":0.4994601127338166,"score_spread":0.2902098755406356,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416827541","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96987516,0.016756134,0.0033913648,0.0023982972,0.006105866,0.0002684548,0.000025303385,0.0000025104778,0.0011769076],"genre_scores_gemma":[0.98559433,0.01161179,0.0009292257,0.00067922,0.0008661032,0.000004718757,0.000009246363,0.000015749021,0.00028964315],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9963809,0.000356714,0.001783901,0.00032173118,0.00075166783,0.00040507838],"domain_scores_gemma":[0.9892697,0.00032156013,0.0010510122,0.00019659127,0.009105986,0.000055198223],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.004377022,0.0002686795,0.0004820436,0.03094779,0.00013623822,0.00015027505,0.000944384,0.00016157981,0.000013352009],"category_scores_gemma":[0.0014404977,0.00024123586,0.000109900546,0.03537382,0.00021968757,0.000018852796,0.0008533427,0.0011313072,0.0000017976655],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009344921,0.0019120132,0.03847328,0.00003405646,0.0030873897,0.00009370913,0.0029330384,0.007495879,0.0016332571,0.00033296747,0.005753458,0.9373165],"study_design_scores_gemma":[0.06071865,0.017434657,0.5134879,0.015260232,0.0002594634,0.00023280137,0.09395023,0.064090736,0.02856133,0.0022809482,0.20067903,0.003044014],"about_ca_topic_score_codex":0.00002424818,"about_ca_topic_score_gemma":0.00006189383,"teacher_disagreement_score":0.93427247,"about_ca_system_score_codex":0.0008059952,"about_ca_system_score_gemma":0.000620255,"threshold_uncertainty_score":0.9851298},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4416982862","doi":"10.58812/esle.v3i03.800","title":"Bibliometric Analysis of Outcome-Based Education (OBE) in Higher Education: Trends, Themes, and Future Directions","year":2025,"lang":"","type":"article","venue":"The Eastasouth Journal of Learning and Educations","topic":"Engineering Education and Curriculum Development","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Employability; Scholarship; Higher education; Scopus; Curriculum; Work (physics); Quality (philosophy)","score_opus":0.01489765467143318,"score_gpt":0.2938090435001856,"score_spread":0.27891138882875244,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416982862","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.77957726,0.16823536,0.0005936439,0.03034932,0.012641205,0.0002819147,0.000021538604,0.00005113898,0.008248587],"genre_scores_gemma":[0.9845541,0.0042962604,0.00125218,0.00009297425,0.00045098175,0.000028708462,0.000025693293,0.000021546344,0.009277539],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99806845,0.00018601956,0.0010404853,0.00020555117,0.00025756407,0.00024191741],"domain_scores_gemma":[0.99830884,0.00031072067,0.0004832415,0.00024291077,0.00048191907,0.0001723745],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0007542088,0.0002648103,0.0005183344,0.05824916,0.00027241337,0.00014567754,0.00021387676,0.00012312374,0.00050699484],"category_scores_gemma":[0.00018690086,0.00022133334,0.0001833748,0.09291426,0.000088281995,0.00015266702,0.000028102815,0.00073653605,0.0000017740218],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021625161,0.0012787294,0.5125379,0.00038270623,0.001506817,2.3311168e-7,0.010951998,0.019216834,0.000025459638,0.004661173,0.0032398244,0.4461767],"study_design_scores_gemma":[0.00024064747,0.000028592798,0.7960289,0.00028244854,0.0014114146,0.0000075728894,0.014638931,0.0006903573,0.0000057921475,0.00005347529,0.18643141,0.00018045293],"about_ca_topic_score_codex":0.00008059623,"about_ca_topic_score_gemma":0.000017551489,"teacher_disagreement_score":0.44599625,"about_ca_system_score_codex":0.00021677242,"about_ca_system_score_gemma":0.0014222845,"threshold_uncertainty_score":0.95242476},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4417049654","doi":"10.70594/brain/16.4/33","title":"Forty Years of Knowledge Development on Emotional Abuse and Suicide – AI-Assisted Bibliometric and Empirical Insights","year":2025,"lang":"","type":"article","venue":"BRAIN BROAD RESEARCH IN ARTIFICIAL INTELLIGENCE AND NEUROSCIENCE","topic":"Suicide and Self-Harm Studies","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Suicidal ideation; Multidisciplinary approach; Depression (economics); Suicide prevention; Empirical research; Observational study; Poison control; Human factors and ergonomics","score_opus":0.3547435748348595,"score_gpt":0.5028304564628628,"score_spread":0.14808688162800332,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417049654","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98738414,0.0073445975,0.00041273292,0.0021273696,0.00066748215,0.00073150103,0.000010395917,0.00001596759,0.0013058258],"genre_scores_gemma":[0.9914617,0.006693832,0.000069111855,0.0010528396,0.000049270096,0.000053909043,0.0000015163674,0.000016239057,0.0006015622],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9943063,0.0010911993,0.0011222687,0.0015061374,0.0009312453,0.0010428027],"domain_scores_gemma":[0.9937788,0.0050036213,0.00012948092,0.00040396088,0.00035965064,0.00032447762],"candidate_categories":["metaepi_narrow","bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0033954254,0.00033801558,0.0005781425,0.02411146,0.0007508341,0.00027639302,0.000540841,0.00026153194,0.00003222002],"category_scores_gemma":[0.0028290807,0.00032545475,0.000059973223,0.04257223,0.0035882168,0.00022906014,0.0006616068,0.0010281747,0.000028537146],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011067978,0.0028582497,0.13738388,0.00043228173,0.00005361195,0.00025780027,0.018429708,0.000029260467,0.018487144,0.014333423,0.0016797767,0.8049481],"study_design_scores_gemma":[0.00026887213,0.0009268503,0.96886843,0.00032148574,0.000009917707,0.000021556345,0.0020539796,0.0014127515,0.013149327,0.006447088,0.0061819297,0.00033779346],"about_ca_topic_score_codex":0.00020202005,"about_ca_topic_score_gemma":0.000525542,"teacher_disagreement_score":0.83148456,"about_ca_system_score_codex":0.000089749024,"about_ca_system_score_gemma":0.00057210407,"threshold_uncertainty_score":0.9999198},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4417348245","doi":"10.1002/sd.70515","title":"Mapping the Research Landscape of Sustainable Development Goals in Business, Management, and Accounting: A Bibliometric and Thematic Synthesis","year":2025,"lang":"en","type":"article","venue":"Sustainable Development","topic":"Corporate Social Responsibility Reporting","field":"Business, Management and Accounting","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Workplace Health, Safety and Compensation Commission","funders":"","keywords":"Thematic map; Sustainable development; Sustainability; Multidisciplinary approach; Thematic analysis; Content analysis","score_opus":0.04177073718012804,"score_gpt":0.28512112714723264,"score_spread":0.24335038996710462,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417348245","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97246975,0.001435372,0.00020424445,0.0010156499,0.00004492942,0.0017258513,1.3901949e-7,0.000056424375,0.023047626],"genre_scores_gemma":[0.9869367,0.00011488519,0.0011563128,0.00012072105,0.000036671525,0.0004480835,0.0000039708325,0.000028936538,0.011153752],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99637616,0.000089889225,0.0010649399,0.00060441875,0.0007609411,0.0011036791],"domain_scores_gemma":[0.9965917,0.0008080395,0.0004703466,0.00039242912,0.0017128977,0.000024623863],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.011112767,0.0002915897,0.0004906625,0.026355527,0.0010232495,0.00080562284,0.00044354322,0.000109208406,0.00003436347],"category_scores_gemma":[0.005362888,0.00023962716,0.00003309027,0.07170407,0.00020849686,0.00085890025,0.0023047004,0.0002574862,0.000007660513],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023371188,0.00040298057,0.6631234,0.030959187,0.00034038213,0.00077287864,0.005080681,0.00003948317,0.00007992216,0.11670271,0.001992196,0.18027245],"study_design_scores_gemma":[0.00045130574,0.0000021492356,0.7859014,0.0005715895,0.000025134068,0.0000021524145,0.1416175,0.00010485284,0.00013615924,0.0112686865,0.059641037,0.00027802706],"about_ca_topic_score_codex":0.0006333293,"about_ca_topic_score_gemma":0.00010474925,"teacher_disagreement_score":0.17999442,"about_ca_system_score_codex":0.00039578133,"about_ca_system_score_gemma":0.0008213894,"threshold_uncertainty_score":0.9846799},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W654809136","doi":"10.1007/s11192-015-1610-x","title":"Statistical tests for ‘related records’ search results","year":2015,"lang":"en","type":"article","venue":"Scientometrics","topic":"Statistics Education and Methodologies","field":"Mathematics","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Ranking (information retrieval); Computer science; Information retrieval; Subject (documents); Set (abstract data type); Measure (data warehouse); Negative binomial distribution; Raw data; Statistics; Data mining; Mathematics; Library science","score_opus":0.7040664114245264,"score_gpt":0.5759315830768087,"score_spread":0.12813482834771772,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W654809136","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.043558415,0.00014864157,0.91505843,0.0012544538,0.007629447,0.0010514054,0.0016708696,0.00033645728,0.029291859],"genre_scores_gemma":[0.043030504,0.0000123012205,0.9400946,0.000073236784,0.00009111114,0.000029224777,0.00009584359,0.000023049693,0.016550137],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9979845,0.00017484209,0.0004004607,0.00032460326,0.00071868877,0.00039687345],"domain_scores_gemma":[0.98651886,0.01172871,0.00010912856,0.00034072768,0.0009563878,0.00034615683],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.007557162,0.000109381384,0.00019814378,0.0013341297,0.00009481707,0.00010129992,0.00027214404,0.0000881114,0.00004854311],"category_scores_gemma":[0.20380263,0.00009452792,0.00003555599,0.0045024035,0.00014216629,0.000060080245,0.000074958254,0.00013652715,0.00012476383],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000088358414,0.00021911053,0.0010886531,0.000057439225,0.000017385242,0.000003619946,0.0018509795,0.000011025661,0.000025004847,0.29300123,0.6759457,0.027691457],"study_design_scores_gemma":[0.0030487522,0.0005810854,0.007504543,0.000023047654,0.000056346915,0.000018508168,0.002026144,0.003480154,0.00070877816,0.8694619,0.112693906,0.00039686356],"about_ca_topic_score_codex":0.000014147316,"about_ca_topic_score_gemma":0.000004842519,"teacher_disagreement_score":0.5764606,"about_ca_system_score_codex":0.0001573553,"about_ca_system_score_gemma":0.00037728393,"threshold_uncertainty_score":0.80290407},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"methods","study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W6889863990","doi":"10.3389/fpubh.2023.1133972.s002","title":"Table_1_Research hotspots and new trends in the impact of resistance training on aging, bibliometric and visual analysis based on CiteSpace and VOSviewer.pdf","year":2023,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Frontier; Intervention (counseling); Visualization; Web of science; Perspective (graphical); Inclusion (mineral); Field (mathematics)","score_opus":0.1823242012883776,"score_gpt":0.4497727298067692,"score_spread":0.2674485285183916,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6889863990","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0020422174,0.0031758384,6.8995384e-8,0.00008715913,0.000007793634,0.0003862189,0.99413496,0.00003765281,0.0001280687],"genre_scores_gemma":[0.01000274,0.00029386606,0.000005699384,0.00006423823,0.000082312305,0.0001066165,0.988746,0.000102260325,0.0005962857],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9951651,0.00066190213,0.00052010285,0.00113708,0.0016620912,0.00085369486],"domain_scores_gemma":[0.99528337,0.0027316841,0.0005402946,0.0008724227,0.00017212062,0.000400139],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0013278838,0.0006965767,0.0012313632,0.13475771,0.00016252845,0.00060961046,0.0006407324,0.0004680789,0.021871861],"category_scores_gemma":[0.005786848,0.00052058214,0.00027017258,0.21507004,0.000060808532,0.00020858235,0.0002459858,0.0013789005,0.00088331645],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023958071,0.000097480435,0.00031571626,0.00061085937,0.00035598094,0.00015528707,0.00012256594,0.00019212342,0.0000076038514,5.8714914e-8,0.99502575,0.0028770177],"study_design_scores_gemma":[0.00220868,0.00095470675,0.755223,0.008011441,0.00051325856,0.0000063387056,0.0001985483,0.0038475746,0.000009683061,0.000003295013,0.22810134,0.0009221596],"about_ca_topic_score_codex":0.002032125,"about_ca_topic_score_gemma":0.003753539,"teacher_disagreement_score":0.7669244,"about_ca_system_score_codex":0.00020549251,"about_ca_system_score_gemma":0.00036277197,"threshold_uncertainty_score":0.9998946},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6890000784","doi":"10.3389/fnagi.2022.1024163.s001","title":"Data_Sheet_1_Rehabilitation of motor function after stroke: A bibliometric analysis of global research from 2004 to 2022.zip","year":2022,"lang":"en","type":"dataset","venue":"Figshare","topic":"UAV Applications and Optimization","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Rehabilitation; Stroke (engine); Function (biology); Motor function; Quality of life (healthcare); Bibliometrics; Quality (philosophy)","score_opus":0.03465830544355915,"score_gpt":0.3162744216113006,"score_spread":0.28161611616774146,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6890000784","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006696832,0.0012135053,0.00013999047,0.000005785202,0.00006883879,0.00053462584,0.9972993,0.00003255273,0.00003572696],"genre_scores_gemma":[0.0019269155,0.0000538937,0.00030218673,0.00000762322,0.000051985022,0.0013790773,0.9962365,0.000019023593,0.00002281653],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982016,0.00007657044,0.00044789675,0.00037055698,0.0006995223,0.00020384684],"domain_scores_gemma":[0.9981458,0.00027897436,0.00011947782,0.0009362117,0.00043312702,0.00008641794],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00017702839,0.00017423304,0.00037725523,0.027959593,0.000044100372,0.000045045672,0.00045078754,0.00019853936,0.56660324],"category_scores_gemma":[0.0006847268,0.00019920275,0.0001574706,0.13693364,0.000007756951,0.00014262935,0.0002393137,0.000269179,0.00020324645],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000369249,0.00005358473,0.00008092382,0.00031976137,0.0004322363,6.0651e-7,0.000007945898,0.041566327,0.000024085346,4.843672e-7,0.9561678,0.0013093039],"study_design_scores_gemma":[0.00011390645,0.00013273858,0.050995544,0.0001250805,0.00039707468,7.7653254e-8,0.000029689825,0.005616245,0.000010885144,0.00000695903,0.9423841,0.0001876664],"about_ca_topic_score_codex":0.0006323148,"about_ca_topic_score_gemma":0.0005002916,"teacher_disagreement_score":0.5664,"about_ca_system_score_codex":0.00033474807,"about_ca_system_score_gemma":0.00007056947,"threshold_uncertainty_score":0.9830577},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6890014941","doi":"10.3389/frma.2021.683212.s001","title":"Table1_Profiling COVID-19 Genetic Research: A Data-Driven Study Utilizing Intelligent Bibliometrics.XLSX","year":2021,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Pandemic; Suite; Public health; Politics; Field (mathematics); Human genetics; Disease; Public policy","score_opus":0.7557046926519175,"score_gpt":0.5409339660861776,"score_spread":0.21477072656573992,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6890014941","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000016487327,0.017597266,0.0000019096503,0.00002916173,0.00031434573,0.0048749186,0.97670656,0.00038338918,0.00007596615],"genre_scores_gemma":[0.00010651721,0.000516394,0.0002946271,0.00028103986,0.0010033841,0.0019297986,0.9952164,0.00049653015,0.00015526844],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9811103,0.0037114066,0.0018434151,0.0046266443,0.0060851877,0.0026230814],"domain_scores_gemma":[0.9783213,0.0040266914,0.001073827,0.012189555,0.0024386416,0.0019499561],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics","open_science","insufficient_payload"],"category_scores_codex":[0.0042753234,0.0013376329,0.0017117094,0.038531326,0.0011936439,0.0027944236,0.011964571,0.0009974855,0.45042723],"category_scores_gemma":[0.14524454,0.0014514595,0.0002809935,0.08365937,0.00008133533,0.0007510242,0.025753524,0.004948395,0.15941276],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000048912167,0.0011537146,0.000042283114,0.0038972213,0.0005116029,0.005272494,0.00013165246,0.00031226067,0.000009577499,1.09696295e-7,0.9881927,0.00042748492],"study_design_scores_gemma":[0.00073510624,0.00030160384,0.00008594196,0.0040707826,0.00024847174,0.00009553133,0.0020862871,0.0007894669,0.000037322294,0.000005761429,0.9902019,0.0013418108],"about_ca_topic_score_codex":0.0022623942,"about_ca_topic_score_gemma":0.0029487158,"teacher_disagreement_score":0.2910145,"about_ca_system_score_codex":0.0021473705,"about_ca_system_score_gemma":0.0067534316,"threshold_uncertainty_score":0.9999375},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6892885160","doi":"10.5281/zenodo.12738098","title":"Scientometric Study of Corporate Communication Research in G20 Countries","year":2024,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Scientific Research and Technology","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Productivity; Presidency; Quarter (Canadian coin); Scientometrics; Resource (disambiguation); Emerging markets","score_opus":0.14192827786906886,"score_gpt":0.3340026721625103,"score_spread":0.19207439429344142,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6892885160","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9381502,0.0013440165,0.012081309,0.002866501,0.00032400858,0.0015874476,0.000051434028,0.0015839753,0.04201109],"genre_scores_gemma":[0.99884325,0.00011791829,0.0004051245,0.000009753271,0.000015773407,2.0596593e-7,0.000052445106,0.0001822124,0.000373304],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9962331,0.0011008851,0.0003373214,0.0006023922,0.0012203727,0.0005059519],"domain_scores_gemma":[0.9972252,0.00022717632,0.00007862877,0.0012687104,0.0010764669,0.0001237876],"candidate_categories":["sts","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.007535975,0.00009434043,0.00014401993,0.004710435,0.0013209798,0.0021326388,0.004288591,0.000059394282,0.00051097455],"category_scores_gemma":[0.0011008066,0.00009179181,0.000026653712,0.015412437,0.00055244414,0.0006862577,0.003968355,0.00056039105,0.0022915732],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006992922,0.0013284179,0.0005702785,0.00029929858,0.000060851988,0.00019131025,0.010468918,0.0001068834,0.0033143715,0.44089055,0.09940133,0.44329786],"study_design_scores_gemma":[0.0014346275,0.0021843999,0.013018828,0.00022813136,0.000005713538,0.00013509665,0.0049289986,0.026860282,0.0022172926,0.021976411,0.9266254,0.00038485028],"about_ca_topic_score_codex":0.000104353596,"about_ca_topic_score_gemma":0.0000055745545,"teacher_disagreement_score":0.827224,"about_ca_system_score_codex":0.00021738515,"about_ca_system_score_gemma":0.00002367521,"threshold_uncertainty_score":0.99997914},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6892965103","doi":"10.5281/zenodo.14208834","title":"Enhancing Citation Integrity through Bibliometrics","year":2024,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Citation; Citation analysis; Webometrics","score_opus":0.2038678357967614,"score_gpt":0.4065832195336073,"score_spread":0.20271538373684592,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6892965103","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06387359,0.0013673186,0.37617347,0.0087743625,0.0017352605,0.00055437034,0.000304745,0.0026728904,0.544544],"genre_scores_gemma":[0.99299514,0.00017813002,0.0012164086,0.00038489667,0.00038850782,3.5339134e-8,0.0003773999,0.00067183044,0.0037876738],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9960113,0.00060716516,0.0005498247,0.0006967069,0.0017696752,0.00036532004],"domain_scores_gemma":[0.9973073,0.0005510974,0.00012967693,0.00053263694,0.001311892,0.00016740829],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.008449766,0.00013510171,0.00017379544,0.0038514894,0.0017528584,0.014780755,0.003334333,0.00013826197,0.014103699],"category_scores_gemma":[0.02626929,0.00011230101,0.000082191385,0.02210403,0.00014390226,0.003651691,0.0017052591,0.0008790222,0.023135718],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000121808525,0.000028352666,0.000009723369,0.00002260942,0.000017083234,0.000011936197,0.002105793,0.00002191261,0.0012490004,0.019247498,0.66424847,0.31302547],"study_design_scores_gemma":[0.00011041895,0.000058947036,0.0004875112,0.00004529662,0.0000071722866,0.000057117315,0.0011931681,0.0014345415,0.00091759005,0.016793488,0.97875196,0.00014280445],"about_ca_topic_score_codex":0.000040603965,"about_ca_topic_score_gemma":5.388303e-7,"teacher_disagreement_score":0.92912155,"about_ca_system_score_codex":0.00017193165,"about_ca_system_score_gemma":0.000013226205,"threshold_uncertainty_score":0.9995467},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"methods","study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6892985362","doi":"10.5281/zenodo.13880761","title":"Analysis of Publications Related to Social Work Education Using Bibliometric Analysis Method","year":2024,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Social Work Education and Practice","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Thematic analysis; Bibliometrics; Content analysis; Competence (human resources); Citation analysis; Trend analysis; Higher education; Field (mathematics)","score_opus":0.10148230646193675,"score_gpt":0.42814682504571605,"score_spread":0.3266645185837793,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6892985362","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12314278,0.00079813384,0.09921606,0.06958143,0.0010805258,0.0013032808,0.00035192195,0.0022033632,0.7023225],"genre_scores_gemma":[0.98537564,0.000054588152,0.0026756553,0.00017075014,0.00015967044,1.8364058e-7,0.0009952248,0.0003100975,0.010258198],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99729043,0.0012240703,0.00032769723,0.0003501621,0.0005565056,0.0002511118],"domain_scores_gemma":[0.9978902,0.00023754443,0.00014906494,0.00025003223,0.0012775589,0.00019563732],"candidate_categories":["bibliometrics","sts","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.003026631,0.00008452617,0.000200579,0.043389633,0.004304972,0.0013856634,0.00061586255,0.00009566167,0.038998593],"category_scores_gemma":[0.0029718296,0.00009955964,0.00019976421,0.4914279,0.00011628515,0.00043485983,0.00019339687,0.00019479326,0.0014370609],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024449717,0.0003782914,0.00052197446,0.0000284025,0.0038636534,6.419607e-7,0.049881767,0.0007510424,0.00028973675,0.1921471,0.16157266,0.5905403],"study_design_scores_gemma":[0.000033102027,0.000012204531,0.020219186,0.000004882665,0.0015043819,6.274881e-7,0.005010021,0.00054346054,0.0000070576275,0.00026738868,0.9722889,0.00010878072],"about_ca_topic_score_codex":0.00028321103,"about_ca_topic_score_gemma":0.0000055747023,"teacher_disagreement_score":0.86223286,"about_ca_system_score_codex":0.0003441156,"about_ca_system_score_gemma":0.00006502633,"threshold_uncertainty_score":0.999651},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6893064476","doi":"10.5281/zenodo.14164212","title":"Mapping Open Science Scholarly Literature","year":2024,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Dalhousie University","funders":"Deutsche Forschungsgemeinschaft","keywords":"Presentation (obstetrics); Scholarly communication; Open science; Key (lock); Knowledge production","score_opus":0.09546984991447147,"score_gpt":0.3302817465086922,"score_spread":0.23481189659422072,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6893064476","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017028567,0.0008793411,0.47667482,0.01277496,0.0004514357,0.0010194443,0.00010436675,0.0024295482,0.50396323],"genre_scores_gemma":[0.9067843,0.0022840376,0.04764822,0.0016709534,0.0006371275,5.117295e-7,0.0014404055,0.0024465986,0.037087876],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9967559,0.00037833746,0.0002145271,0.0010466106,0.0010459294,0.00055868336],"domain_scores_gemma":[0.99729604,0.00005459942,0.000059252005,0.0015006376,0.0008250757,0.00026439904],"candidate_categories":["sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.005590617,0.00013407391,0.00010441752,0.0010147614,0.0035044763,0.23793873,0.01994872,0.00003882121,0.0009092214],"category_scores_gemma":[0.0026487939,0.0001283067,0.000033725544,0.0055535594,0.00025424158,0.11629295,0.026907783,0.0005963963,0.0062427223],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000007833456,0.00005565123,8.955279e-7,0.0001105017,0.000029946017,0.0001485796,0.0015698051,0.0000083439745,0.0071550794,0.67395115,0.08772826,0.22923397],"study_design_scores_gemma":[0.0001333176,0.00009097804,0.00024086762,0.00013918233,0.000002260267,0.00012278285,0.00014133395,0.0038394395,0.00029695037,0.0014079635,0.9934298,0.00015513755],"about_ca_topic_score_codex":0.0000136385315,"about_ca_topic_score_gemma":9.1765536e-8,"teacher_disagreement_score":0.9057015,"about_ca_system_score_codex":0.00022551141,"about_ca_system_score_gemma":0.00002677254,"threshold_uncertainty_score":0.99779284},"labels":[{"model":"gemma","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6893696318","doi":"10.5281/zenodo.3758061","title":"Recent history and geography of global peatland research revealed by bibliometrics","year":2020,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Peatlands and Wetlands Ecology","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université de Montréal","funders":"","keywords":"Bibliometrics; Presentation (obstetrics); Peat; Biogeosciences; Field (mathematics)","score_opus":0.04581522388314522,"score_gpt":0.252250692246523,"score_spread":0.2064354683633778,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6893696318","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.69101125,0.0022870838,0.00056857395,0.0053598047,0.00009714763,0.00067352736,0.00041266714,0.00033869775,0.29925123],"genre_scores_gemma":[0.9954745,0.0031037163,0.00015059397,0.00032596927,0.000052558167,5.488202e-8,0.00045214,0.00024988476,0.00019058576],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99819106,0.00036003883,0.00020990615,0.00039665142,0.0004959294,0.0003464098],"domain_scores_gemma":[0.9992414,0.00003622464,0.000075663964,0.00021340956,0.00013899578,0.0002943166],"candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0009041281,0.00009703626,0.00015089037,0.00032849185,0.0005848713,0.00010345544,0.0005960542,0.000067096546,0.016721906],"category_scores_gemma":[0.00063180615,0.00009526987,0.000029402498,0.0038647878,0.00046175288,0.00013999388,0.0013200664,0.00021448682,0.0009671949],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009027777,0.00010030601,0.015582274,0.000027878577,0.00001710286,0.000008853976,0.00027960044,0.0000059659696,0.0027245756,0.00014497816,0.92094356,0.060074616],"study_design_scores_gemma":[0.00045938027,0.0004526183,0.06204179,0.0000045712763,0.0000063229213,0.000019819807,0.000054255128,0.00011608693,0.0000377187,0.00007670919,0.9366354,0.00009534868],"about_ca_topic_score_codex":0.00006871519,"about_ca_topic_score_gemma":0.0000010929409,"teacher_disagreement_score":0.30446324,"about_ca_system_score_codex":0.00025735743,"about_ca_system_score_gemma":0.0000027551473,"threshold_uncertainty_score":0.9998107},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6894115518","doi":"10.5281/zenodo.7121385","title":"MAPPING THE INTELLECTUAL STRUCTURE OF SOCIAL FINANCE: A BIBLIOMETRIC OVERVIEW","year":2021,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Community Development and Social Impact","field":"Economics, Econometrics and Finance","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"Horizon 2020 Framework Programme","keywords":"Field (mathematics); Bibliometrics; Social research; Domain (mathematical analysis); Core (optical fiber); Social network analysis","score_opus":0.13107542723429133,"score_gpt":0.26692209674506595,"score_spread":0.13584666951077462,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6894115518","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8057419,0.0060404614,0.0024989634,0.0027713738,0.0002836257,0.00036671254,0.0008161849,0.00022242007,0.18125835],"genre_scores_gemma":[0.99733967,0.0011263579,0.00009949056,0.00014242475,0.0000710871,1.0444462e-8,0.00031638908,0.00023584656,0.00066870824],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990943,0.00012829619,0.00030728505,0.00017579577,0.000084922576,0.00020939548],"domain_scores_gemma":[0.9992293,0.000056829616,0.00016595522,0.00024916796,0.00026448164,0.000034224784],"candidate_categories":["sts","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00050742965,0.000082722385,0.00019233293,0.002012137,0.0015907694,0.0003828421,0.00065408205,0.000056704113,0.019351073],"category_scores_gemma":[0.0014410475,0.00008543405,0.00007490819,0.013898843,0.000104953135,0.00014516198,0.00079597835,0.00024815387,0.0011527481],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003966379,0.00027830907,0.0001625261,0.0002480607,0.00028751366,0.000009927703,0.06350275,0.000023863078,0.0016819554,0.21743767,0.3312075,0.38512024],"study_design_scores_gemma":[0.0002215801,0.000029791592,0.016613167,0.000012081586,0.000002537156,0.000013891084,0.0011885116,0.000066832625,0.00017064845,0.0034657302,0.9781047,0.00011052201],"about_ca_topic_score_codex":0.000020658355,"about_ca_topic_score_gemma":7.8852435e-7,"teacher_disagreement_score":0.6468972,"about_ca_system_score_codex":0.00009236686,"about_ca_system_score_gemma":0.0000062974464,"threshold_uncertainty_score":0.999709},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6901709454","doi":"10.60692/jgbph-scm60","title":"Are the most highly cited articles the ones that are the most downloaded? A bibliometric study of IRRODL","year":2015,"lang":"en","type":"article","venue":"Greater South Information System","topic":"Online and Blended Learning","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Athabasca University","funders":"","keywords":"Publishing; Bibliometrics; Impact factor; Peer review; Order (exchange); Citation analysis; Citation; Publication; Work (physics)","score_opus":0.10621178919153854,"score_gpt":0.28909143881578936,"score_spread":0.18287964962425082,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6901709454","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9940963,0.000031762247,0.00008420877,0.0024908301,0.00033726927,0.00084141747,0.00004117105,0.00011994942,0.001957098],"genre_scores_gemma":[0.99936146,9.2824575e-7,0.000010953517,0.0002360259,0.00016717565,0.00006017033,0.0000034580355,0.0000067541782,0.00015307388],"study_design_codex":"observational","study_design_gemma":"qualitative","domain_scores_codex":[0.99724513,0.0006534772,0.00052919646,0.00012033923,0.0011542089,0.0002976406],"domain_scores_gemma":[0.9977286,0.00011256374,0.0010080346,0.0004789452,0.0005829937,0.00008886246],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0025732152,0.0001518376,0.00022231284,0.0023218866,0.0008336601,0.0005626397,0.00064031605,0.00008205878,0.000010401833],"category_scores_gemma":[0.00038905148,0.00006657222,0.00005886992,0.019687194,0.00014969961,0.0007789518,0.00011674211,0.00019869982,0.00020707576],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000028211061,0.000013150328,0.6648474,0.000029719475,0.000053054548,0.000001429077,0.33208,0.00032079557,2.592716e-7,0.00014914431,0.000814933,0.0016618939],"study_design_scores_gemma":[0.0004914378,0.000045845278,0.19526023,0.000050633156,0.00004305331,0.0000031072182,0.798651,0.00037115178,0.000026434245,0.0000027839533,0.004965709,0.00008860687],"about_ca_topic_score_codex":0.0005609038,"about_ca_topic_score_gemma":0.00010609201,"teacher_disagreement_score":0.46958718,"about_ca_system_score_codex":0.000073649455,"about_ca_system_score_gemma":0.000116172596,"threshold_uncertainty_score":0.94590443},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6902105260","doi":"10.6084/m9.figshare.13642706.v1","title":"Analyzing Canadian ecological restoration literature with bibliometric analysis and a systematic map. Presented at ESMARConf2021","year":2021,"lang":"en","type":"other","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scope (computer science); Systematic review; Selection (genetic algorithm); Bibliometrics; Presentation (obstetrics); Process (computing)","score_opus":0.03169186853094417,"score_gpt":0.27163908405617154,"score_spread":0.23994721552522738,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6902105260","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00029290668,0.14336523,0.0000021516687,0.00008803524,0.000064566666,0.0029819869,0.80213463,0.0004169638,0.050653543],"genre_scores_gemma":[0.012773331,0.000035039047,0.00015422389,0.00003976036,0.00018572698,0.0006656007,0.70057714,0.000648021,0.28492114],"study_design_codex":"not_applicable","study_design_gemma":"systematic_review","domain_scores_codex":[0.9966388,0.0006041379,0.00044640928,0.0010419574,0.00065063627,0.00061806827],"domain_scores_gemma":[0.99696916,0.00020362859,0.0007287093,0.0009499436,0.00049950235,0.0006490231],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.00017651625,0.0006237677,0.0013641928,0.17418067,0.00017912868,0.0009005248,0.00036124635,0.0008859546,0.3166523],"category_scores_gemma":[0.0021427972,0.0004929177,0.0002430387,0.25857934,0.00001857447,0.00012398593,0.00020065036,0.0004817516,0.002617153],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009641585,0.00003075703,0.0032596781,0.039817818,0.0031558652,0.001700742,0.000061716644,0.000019833811,0.000008171783,0.0000014285098,0.9519267,0.0000076102137],"study_design_scores_gemma":[0.0014978684,0.000235828,0.14146392,0.4447627,0.009234161,0.00041786357,0.00013510043,0.0031857565,0.000024849773,0.0000024625183,0.39573175,0.0033077365],"about_ca_topic_score_codex":0.011810396,"about_ca_topic_score_gemma":0.674634,"teacher_disagreement_score":0.66282356,"about_ca_system_score_codex":0.0007681209,"about_ca_system_score_gemma":0.00042431476,"threshold_uncertainty_score":0.9997522},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"other","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"other","about_ca_system":false,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"agree"},{"id":"W6902115866","doi":"10.6084/m9.figshare.14927189","title":"Additional file 1 of Osteopathic empirical research: a bibliometric analysis from 1966 to 2018","year":2021,"lang":"en","type":"article","venue":"Figshare","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"CINAHL; MEDLINE; Bibliometrics; Empirical research","score_opus":0.8486557295322864,"score_gpt":0.6260689090531582,"score_spread":0.22258682047912826,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6902115866","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0014698143,0.00042558138,0.000012156855,0.00036998338,0.000052620275,0.00014618025,0.9863372,0.000018990466,0.011167503],"genre_scores_gemma":[0.018726084,0.000008081157,0.002554403,0.0003198078,0.000224505,0.00044315463,0.9676782,0.000017404916,0.010028387],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9788526,0.000811906,0.0008652634,0.0012294125,0.017450867,0.0007899157],"domain_scores_gemma":[0.8470082,0.1302154,0.0003116504,0.001813663,0.019499222,0.001151798],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.004296101,0.00016432187,0.0005560771,0.71941704,0.00027502116,0.001942593,0.0027842908,0.00021594805,0.9991601],"category_scores_gemma":[0.6895123,0.00013084593,0.0004968745,0.9700267,0.000051389772,0.00039758187,0.002500511,0.0005235072,0.05729236],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000803114,0.00013276888,0.0008632945,0.0000049024384,0.00006899134,0.00006747379,0.00003216135,0.000014829701,0.000021968199,0.0000010927362,0.9705923,0.028192172],"study_design_scores_gemma":[0.0000759944,0.000079494996,0.2673119,0.000110799825,0.0000073715087,0.0000019351578,0.000110445384,0.00079645147,0.00020035062,0.00037155856,0.7308152,0.00011851439],"about_ca_topic_score_codex":0.000055901975,"about_ca_topic_score_gemma":0.00010140178,"teacher_disagreement_score":0.94186777,"about_ca_system_score_codex":0.00014177998,"about_ca_system_score_gemma":0.0012399936,"threshold_uncertainty_score":0.9990935},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6902188819","doi":"10.6084/m9.figshare.19926471.v1","title":"Trends in Graves’ orbitopathy research in the past two decades: a bibliometric analysis","year":2022,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Web of science; China; Bibliometrics; Quality (philosophy); European union","score_opus":0.2327854665999361,"score_gpt":0.4586738128922817,"score_spread":0.2258883462923456,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6902188819","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00009531261,0.003044415,3.317044e-9,0.00006822511,0.00003987458,0.00060202245,0.995654,0.000054070886,0.00044208503],"genre_scores_gemma":[0.001632016,0.000053274918,0.0000025536294,0.000098333956,0.00030719937,0.0040429537,0.9935796,0.00013507607,0.00014895549],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9863838,0.00447232,0.0010215656,0.001622723,0.004828532,0.0016710355],"domain_scores_gemma":[0.99353665,0.0021970242,0.000478612,0.0032541633,0.00031502516,0.00021855303],"candidate_categories":["metaepi_narrow","bibliometrics","open_science","research_integrity","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0045153624,0.0006956818,0.001146368,0.72688323,0.00033218207,0.0007538118,0.0058804015,0.00046141935,0.85713595],"category_scores_gemma":[0.0043980684,0.0005751961,0.0006266908,0.88181365,0.000041821597,0.0002541678,0.0022466122,0.00581183,0.02159884],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000040997675,0.00039995005,0.00049944915,0.00027322798,0.00018829481,0.0021246625,0.0001424177,0.00015308952,4.418867e-7,3.6460395e-7,0.9950907,0.0010863559],"study_design_scores_gemma":[0.000638085,0.00010559966,0.05040347,0.0005679951,0.0001729307,0.000023763778,0.00030044053,0.000059933715,5.5410004e-7,0.000012310831,0.94715154,0.0005633963],"about_ca_topic_score_codex":0.004737303,"about_ca_topic_score_gemma":0.02225908,"teacher_disagreement_score":0.83553714,"about_ca_system_score_codex":0.0011710805,"about_ca_system_score_gemma":0.00034101558,"threshold_uncertainty_score":0.99966997},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6903349993","doi":"10.1097/hnp.0000000000000621&#13;\\n&#13;\\n","title":"A Scientometric Overview of the Current Status and Trends of Holistic Nursing Studies","year":2024,"lang":"en","type":"article","venue":"","topic":"Nursing education and management","field":"Nursing","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Holistic nursing; Web of science; Work (physics); Nursing research; Bibliometrics; Quality (philosophy); Citation; MEDLINE","score_opus":0.18544765742368396,"score_gpt":0.4786252509774978,"score_spread":0.29317759355381384,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6903349993","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.45413965,0.4652953,0.00014859288,0.010290348,0.049641434,0.0004023937,0.000019903713,0.00013542717,0.019926911],"genre_scores_gemma":[0.9968848,0.0007574774,0.000082283244,0.000041503936,0.00003118271,0.000002946444,8.4532843e-7,0.000006450795,0.0021924623],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9992957,0.00003697095,0.0001752094,0.00014116804,0.00021847263,0.00013248304],"domain_scores_gemma":[0.99961877,0.00008725847,0.000049700186,0.00015555341,0.00005713228,0.00003160153],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00017961708,0.00006688943,0.0001277016,0.0007590826,0.000047887745,0.00003112915,0.000070031114,0.000008819652,0.000028554388],"category_scores_gemma":[0.000096686075,0.00004199627,0.0000468727,0.0026434716,0.00020351105,0.00005463934,0.000032328666,0.000057026744,0.000001468312],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000052767587,0.000110689565,0.00029662432,0.000607851,0.000023538312,4.9144877e-8,0.0032921922,0.0000025977727,0.0000788041,0.0141550815,0.02515252,0.95627475],"study_design_scores_gemma":[0.00093240646,0.00041225128,0.51032245,0.012392997,0.0011621674,0.0000057839643,0.0080679655,0.0018617059,0.004494621,0.021214526,0.43867546,0.00045764167],"about_ca_topic_score_codex":0.000017617242,"about_ca_topic_score_gemma":0.000004352698,"teacher_disagreement_score":0.95581716,"about_ca_system_score_codex":0.00010650863,"about_ca_system_score_gemma":0.00001805766,"threshold_uncertainty_score":0.17125581},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6908558117","doi":"10.3389/fnagi.2022.1024163.s003","title":"Image_2_Rehabilitation of motor function after stroke: A bibliometric analysis of global research from 2004 to 2022.TIF","year":2022,"lang":"en","type":"other","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Rehabilitation; Stroke (engine); Function (biology); Motor function; Quality of life (healthcare); Bibliometrics; Quality (philosophy)","score_opus":0.05655062911334826,"score_gpt":0.3714382925620727,"score_spread":0.3148876634487245,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6908558117","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0034837634,0.008503871,0.000004094894,0.0000103005,0.000074689015,0.0013557636,0.9567189,0.00012647525,0.029722111],"genre_scores_gemma":[0.058909018,0.000011038654,0.0006007963,0.00002980868,0.00034700715,0.003622167,0.7874798,0.0010457522,0.14795457],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9943487,0.0007267434,0.0006619716,0.0010061355,0.002746031,0.0005104406],"domain_scores_gemma":[0.9960225,0.00047047628,0.0006453001,0.001429446,0.0012141895,0.00021806346],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.00055421115,0.00037936599,0.00095018145,0.18733163,0.00004767803,0.00006932994,0.00075548573,0.00037957504,0.94469774],"category_scores_gemma":[0.0064741657,0.00041476035,0.000658693,0.42190808,0.000034228728,0.00014536134,0.0006815485,0.00047027812,0.012782333],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004490733,0.00023692448,0.0063895914,0.0005189547,0.0022516246,0.000009272739,0.000051020023,0.00008297279,0.0006679299,8.1732367e-7,0.98713124,0.0022105852],"study_design_scores_gemma":[0.0004670365,0.000586255,0.57762957,0.0009616138,0.0009769787,2.0373872e-7,0.00016240534,0.00012493678,0.000035359724,0.00002006789,0.4186292,0.00040634634],"about_ca_topic_score_codex":0.0036017334,"about_ca_topic_score_gemma":0.002573428,"teacher_disagreement_score":0.9319154,"about_ca_system_score_codex":0.0012444638,"about_ca_system_score_gemma":0.00033036602,"threshold_uncertainty_score":0.9998304},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6908615347","doi":"10.3389/fnhum.2022.981571.s002","title":"Image_2_A scientometric analysis of the 100 most cited articles on magnetic resonance guided focused ultrasound.TIFF","year":2022,"lang":"en","type":"other","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Web of science; Bibliometrics; Citation analysis; Scientometrics; Informetrics","score_opus":0.054910085993447155,"score_gpt":0.29516412458796226,"score_spread":0.2402540385945151,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6908615347","genre_codex":"dataset","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0005966238,0.008889973,7.0527705e-8,0.000022572132,0.000084431784,0.0009995135,0.7787964,0.0003795833,0.21023086],"genre_scores_gemma":[0.014616612,0.000019886964,0.00014020716,0.00043330833,0.00020421135,0.00077422895,0.22436197,0.001980996,0.7574686],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9950221,0.00043127884,0.00060586625,0.0010319654,0.0022433614,0.0006653821],"domain_scores_gemma":[0.995548,0.00050700275,0.0009887369,0.0025164166,0.000288448,0.00015141437],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00027757083,0.0005627007,0.0008320178,0.01039782,0.0002278175,0.00016900977,0.0021000286,0.00023439825,0.8974644],"category_scores_gemma":[0.010100117,0.00045232454,0.0006170917,0.07851283,0.00008400628,0.000064886925,0.000537653,0.00048822866,0.010792966],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000170168,0.00016805493,0.00076766167,0.00008780424,0.00032628683,0.000014477909,0.0000750796,0.00012492396,0.00048897037,0.0000050544086,0.99621886,0.0017058335],"study_design_scores_gemma":[0.0006269719,0.00009144257,0.0636605,0.0011894986,0.00088844984,0.0000032405367,0.000024137731,0.0005931269,0.00050649326,0.000008132083,0.93183506,0.0005729199],"about_ca_topic_score_codex":0.00025186784,"about_ca_topic_score_gemma":0.00037555787,"teacher_disagreement_score":0.8866714,"about_ca_system_score_codex":0.00045037657,"about_ca_system_score_gemma":0.00019150366,"threshold_uncertainty_score":0.9997929},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6908677389","doi":"10.3389/fimmu.2024.1519498.s001","title":"Table 1_A bibliometric study on the impact of gut microbiota on the efficacy of immune checkpoint inhibitors in cancer patients: analysis of the top 100 cited articles.xlsx","year":2025,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Gut flora; Citation; Cancer; Immunotherapy; Web of science; Cancer immunotherapy; Ranking (information retrieval); China; Dysbiosis","score_opus":0.04834399244233295,"score_gpt":0.34054917857252254,"score_spread":0.2922051861301896,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6908677389","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.40857884,0.00033962418,1.0070539e-9,0.000020786265,0.000038819253,0.0018023977,0.5891976,0.000008830656,0.000013101523],"genre_scores_gemma":[0.43437672,0.000008117605,1.1530522e-7,0.000021546728,0.00001675882,0.00018679487,0.56534225,0.000031015254,0.000016693857],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9946976,0.00097625604,0.0018002669,0.0006745453,0.001293235,0.000558107],"domain_scores_gemma":[0.99075705,0.0020044611,0.0027780747,0.0034686516,0.0009283573,0.000063415],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00084449543,0.0006784113,0.0014559791,0.03670749,0.00012932975,0.000088488196,0.0026143547,0.00029514713,0.04444233],"category_scores_gemma":[0.009975525,0.00032776708,0.0011889611,0.25558847,0.00007235948,0.000092798444,0.001125977,0.000987392,0.00015550102],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027146214,0.0046504606,0.006034615,0.00018658639,0.0035685578,0.0000012493055,0.00022768881,0.0027033712,0.0018431962,1.1857369e-7,0.9804505,0.00006220313],"study_design_scores_gemma":[0.001708776,0.0005566345,0.97401965,0.006167684,0.0014798915,6.439833e-8,0.00023898509,0.00008016681,0.007655106,5.0802777e-7,0.007731998,0.00036052777],"about_ca_topic_score_codex":0.014892862,"about_ca_topic_score_gemma":0.0011428572,"teacher_disagreement_score":0.9727185,"about_ca_system_score_codex":0.0006815194,"about_ca_system_score_gemma":0.00054261443,"threshold_uncertainty_score":0.99991745},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6908933652","doi":"10.3389/fnut.2024.1485642.s003","title":"Data_Sheet_3_Mapping the evolution and impact of ketogenic diet research on diabetes management: a comprehensive bibliometric analysis from 2005 to 2024.CSV","year":2024,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Diabetes mellitus; Diabetes management; Bibliometrics; MEDLINE; Web of science; Clinical nutrition","score_opus":0.14401047732410727,"score_gpt":0.4117347885710367,"score_spread":0.2677243112469294,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6908933652","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017527152,0.016071491,1.2981731e-7,0.000028863971,0.000075288255,0.0016672688,0.9802018,0.0000734524,0.0001290073],"genre_scores_gemma":[0.0061660334,0.00022112949,0.000012093687,0.00003690191,0.00028017088,0.0006903383,0.9923379,0.00012819549,0.00012723595],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99311626,0.0008239336,0.00073065975,0.0016862655,0.0023922448,0.0012506123],"domain_scores_gemma":[0.99349344,0.0019788423,0.00041326782,0.003080586,0.00064401457,0.00038983286],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.00080211856,0.0007609175,0.0011390769,0.21045391,0.0003029862,0.00086401124,0.0022217014,0.00038355804,0.11780331],"category_scores_gemma":[0.0020398542,0.00056572654,0.00066041923,0.3364656,0.0000665312,0.00029403763,0.004154994,0.0015056506,0.08161485],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013478288,0.00007411695,0.00023030775,0.0014029257,0.009560123,0.000050447426,0.000020240574,0.00066557765,0.000027205326,3.5805147e-7,0.9876465,0.00018740674],"study_design_scores_gemma":[0.00026802238,0.00042552926,0.11876839,0.0060873786,0.0026130679,6.29874e-7,0.00014487033,0.00091185153,0.000007161808,0.00003275827,0.8701626,0.0005777158],"about_ca_topic_score_codex":0.0023001784,"about_ca_topic_score_gemma":0.0010943379,"teacher_disagreement_score":0.12601168,"about_ca_system_score_codex":0.0012284743,"about_ca_system_score_gemma":0.00017268622,"threshold_uncertainty_score":0.99967945},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6910677159","doi":"10.48550/arxiv.1502.03471","title":"A scientometric study of General Relativity and Quantum Cosmology from 2000 to 2012","year":2015,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Relativity and Gravitational Theory","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Centennial; General relativity; Publication; Einstein; Publishing; Theory of relativity; Cosmology","score_opus":0.09388425622658543,"score_gpt":0.23755995666577626,"score_spread":0.1436757004391908,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6910677159","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9858771,0.00006283135,0.011021141,0.000018702292,0.00022418862,0.0003875586,0.00023494437,0.000013284459,0.0021602772],"genre_scores_gemma":[0.997302,0.0000023295488,0.00012407762,0.00000922672,0.00007502389,0.000001466157,0.000054359625,0.000011120013,0.0024204091],"study_design_codex":"observational","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99864155,0.00030521728,0.00017296098,0.0005948127,0.00009850529,0.00018696875],"domain_scores_gemma":[0.9987806,0.0002593324,0.00021839805,0.00038910314,0.000171585,0.00018100273],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00037953406,0.00019885313,0.00038102566,0.0005941746,0.00008382625,0.00002209933,0.00028072295,0.00011375227,0.00009131936],"category_scores_gemma":[0.000024811021,0.00022234156,0.00008036238,0.0009241832,0.00012243907,0.00016002533,0.0006548636,0.00035957934,0.000035810306],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026151873,0.0014195241,0.58710253,0.000016802549,0.0005503083,0.000028956529,0.0030246526,0.10308681,0.000069799564,0.3027993,0.0008369436,0.0008028432],"study_design_scores_gemma":[0.0013138027,0.00030068264,0.1460601,0.00003114356,0.00021633683,2.2528249e-7,0.0015458993,0.007888838,0.000022460084,0.8420641,0.00019021243,0.00036619976],"about_ca_topic_score_codex":0.002816207,"about_ca_topic_score_gemma":0.000040623137,"teacher_disagreement_score":0.5392648,"about_ca_system_score_codex":0.000078130295,"about_ca_system_score_gemma":0.00016479347,"threshold_uncertainty_score":0.9066826},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6913030144","doi":"10.5281/zenodo.8003645","title":"Advances in Gas Hydrate Research from Energy, Environmental, Engineering And Scientific Perspectives as Informed by Comparative Bibliometric and Scientometric Analyses","year":2023,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Methane Hydrates and Related Phenomena","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Guelph","funders":"","keywords":"Clathrate hydrate; Methane; Greenhouse gas; Seriousness; Hydrate; Global warming; Permafrost; Natural gas; Biosphere","score_opus":0.06125005963277075,"score_gpt":0.3171968636244275,"score_spread":0.25594680399165676,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6913030144","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9819153,0.0051474744,0.000052450705,0.00009792829,0.000039454026,0.00020681033,0.0000766792,0.00013315234,0.012330762],"genre_scores_gemma":[0.9909533,0.0075473837,0.0000706089,0.0000069395833,0.000014717776,1.9254522e-7,0.0004121036,0.00016096438,0.0008337747],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.99771386,0.00025333802,0.00021403187,0.0006049924,0.0007359668,0.00047780844],"domain_scores_gemma":[0.9992672,0.00016601867,0.00006265445,0.00022912973,0.000039821138,0.00023515009],"candidate_categories":["bibliometrics","sts","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0012120452,0.00013962438,0.00016135439,0.010749207,0.0014485611,0.00093899213,0.00046446544,0.000051934094,0.0044469656],"category_scores_gemma":[0.00040159575,0.00013086984,0.000019807394,0.04602159,0.00077237445,0.0007231627,0.0014666956,0.00025678752,0.002238211],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027790593,0.0009364977,0.003289084,0.00014407907,0.0001904167,0.00010624096,0.044531886,0.01679656,0.51366824,0.0019863022,0.08423438,0.33383843],"study_design_scores_gemma":[0.0010210784,0.0004150606,0.02868377,0.000052803043,0.000015354211,0.000025222957,0.009142166,0.02533049,0.003823999,0.0008021656,0.9302349,0.00045298968],"about_ca_topic_score_codex":0.00017206569,"about_ca_topic_score_gemma":0.000002415424,"teacher_disagreement_score":0.84600055,"about_ca_system_score_codex":0.00023675362,"about_ca_system_score_gemma":0.000002489067,"threshold_uncertainty_score":0.9998514},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W6920064162","doi":"10.60692/8bxyr-cz327","title":"Prostate cancer research in India: A scientometric analysis of publications output during 2004-13","year":2015,"lang":"en","type":"article","venue":"Greater South Information System","topic":"Prostate Cancer Diagnosis and Treatment","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Prostate cancer; Scopus; Tamil; Cancer; Citation; Alternative medicine; Bibliometrics","score_opus":0.13959673815453272,"score_gpt":0.3493875244230271,"score_spread":0.2097907862684944,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6920064162","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9958612,0.000101093654,0.000044189346,0.00023945424,0.000096970914,0.0008971197,0.00024589882,0.000050320006,0.0024637743],"genre_scores_gemma":[0.9987912,0.0000047750377,0.000045290526,0.00002690031,0.00002256265,0.0006980095,0.00007402923,0.000006880898,0.0003303102],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99795896,0.000067229455,0.0006635855,0.00017951037,0.000805149,0.00032555405],"domain_scores_gemma":[0.99804634,0.000012646381,0.00025379195,0.00039776598,0.0010673356,0.00022209459],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0011206277,0.000113261216,0.00037819464,0.008894712,0.000068796915,0.000095923744,0.000107517655,0.000067086396,0.000015385669],"category_scores_gemma":[0.000069822665,0.00008797188,0.000083217536,0.0134985875,0.000055005956,0.0005291222,0.00006278847,0.00012144355,0.00008078549],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008094717,0.000016337273,0.9609712,0.00030332286,0.00028026808,0.0000036175222,0.037285678,0.00055290555,8.5203465e-7,0.000033388158,0.00010848188,0.0003630239],"study_design_scores_gemma":[0.00224736,0.00007832131,0.987711,0.00024506345,0.00020927018,0.000005008355,0.007958345,0.0008216937,0.00048939907,4.659564e-7,0.0001575265,0.000076527525],"about_ca_topic_score_codex":0.0002719904,"about_ca_topic_score_gemma":0.000004957063,"teacher_disagreement_score":0.029327333,"about_ca_system_score_codex":0.0011824954,"about_ca_system_score_gemma":0.0003540013,"threshold_uncertainty_score":0.7936659},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6920436288","doi":"10.60692/hcb1y-90n06","title":"Global Research Trends in Pediatric Trauma From 1968 to 2021: A Bibliometric Analysis","year":2021,"lang":"en","type":"article","venue":"Greater South Information System","topic":"COVID-19 and healthcare impacts","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Citation; Data collection; Field (mathematics); MEDLINE; Descriptive statistics; Web of science","score_opus":0.17795545511477454,"score_gpt":0.41018544715017236,"score_spread":0.23222999203539782,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6920436288","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99084884,0.000050917308,0.0019463275,0.0010925776,0.00018092329,0.00026026677,0.00035622646,0.00006314852,0.0052008],"genre_scores_gemma":[0.99845064,0.000002738223,0.00024849418,0.0005307615,0.00028730408,0.000041735922,0.0001932955,0.000007038924,0.00023797475],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99716926,0.00020496802,0.0008544488,0.00027960306,0.00096753123,0.0005241904],"domain_scores_gemma":[0.99800485,0.000047227662,0.00012784271,0.0005862528,0.0007205197,0.00051330845],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0011019264,0.0001623492,0.0005269871,0.10520319,0.00009449887,0.0002102619,0.00013270736,0.000179855,0.0003528353],"category_scores_gemma":[0.00025325734,0.00014099815,0.00017830745,0.3251036,0.00001122037,0.00035844228,0.000091435315,0.00020737483,0.00092443946],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013293281,0.00000809256,0.966115,0.00026787457,0.00009593514,0.00006956947,0.014216211,0.00007839027,2.1392634e-7,0.000015757088,0.0007110363,0.01828896],"study_design_scores_gemma":[0.0013499537,0.00009851123,0.992192,0.00007487118,0.00020355926,0.00001582743,0.004389806,0.0005318082,0.000023134826,9.149741e-7,0.0009971517,0.00012248511],"about_ca_topic_score_codex":0.0013805848,"about_ca_topic_score_gemma":0.00011858434,"teacher_disagreement_score":0.21990043,"about_ca_system_score_codex":0.00097042305,"about_ca_system_score_gemma":0.00038284928,"threshold_uncertainty_score":0.99985343},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6920720784","doi":"10.6084/m9.figshare.13642706","title":"Analyzing Canadian ecological restoration literature with bibliometric analysis and a systematic map. Presented at ESMARConf2021","year":2021,"lang":"en","type":"other","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scope (computer science); Systematic review; Selection (genetic algorithm); Bibliometrics; Presentation (obstetrics); Process (computing)","score_opus":0.03169186853094417,"score_gpt":0.27163908405617154,"score_spread":0.23994721552522738,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6920720784","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00029290668,0.14336523,0.0000021516687,0.00008803524,0.000064566666,0.0029819869,0.80213463,0.0004169638,0.050653543],"genre_scores_gemma":[0.012773331,0.000035039047,0.00015422389,0.00003976036,0.00018572698,0.0006656007,0.70057714,0.000648021,0.28492114],"study_design_codex":"not_applicable","study_design_gemma":"systematic_review","domain_scores_codex":[0.9966388,0.0006041379,0.00044640928,0.0010419574,0.00065063627,0.00061806827],"domain_scores_gemma":[0.99696916,0.00020362859,0.0007287093,0.0009499436,0.00049950235,0.0006490231],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.00017651625,0.0006237677,0.0013641928,0.17418067,0.00017912868,0.0009005248,0.00036124635,0.0008859546,0.3166523],"category_scores_gemma":[0.0021427972,0.0004929177,0.0002430387,0.25857934,0.00001857447,0.00012398593,0.00020065036,0.0004817516,0.002617153],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009641585,0.00003075703,0.0032596781,0.039817818,0.0031558652,0.001700742,0.000061716644,0.000019833811,0.000008171783,0.0000014285098,0.9519267,0.0000076102137],"study_design_scores_gemma":[0.0014978684,0.000235828,0.14146392,0.4447627,0.009234161,0.00041786357,0.00013510043,0.0031857565,0.000024849773,0.0000024625183,0.39573175,0.0033077365],"about_ca_topic_score_codex":0.011810396,"about_ca_topic_score_gemma":0.674634,"teacher_disagreement_score":0.66282356,"about_ca_system_score_codex":0.0007681209,"about_ca_system_score_gemma":0.00042431476,"threshold_uncertainty_score":0.9997522},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"other","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W6920784561","doi":"10.6084/m9.figshare.20678000","title":"Additional file 4 of Factors associated with reporting of the Prevention of Falls Network Europe (ProFaNE) core outcome set domains in randomized trials on falls in older people: a citation analysis and correlational study","year":2022,"lang":"en","type":"article","venue":"Figshare","topic":"Balance, Gait, and Falls Prevention","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba; George & Fay Yee Centre for Healthcare Innovation","funders":"","keywords":"Randomized controlled trial; Set (abstract data type); Core (optical fiber); Outcome (game theory); Poison control","score_opus":0.16231044937062433,"score_gpt":0.40347911232489053,"score_spread":0.2411686629542662,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6920784561","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.46347904,0.000008174948,0.0000021064604,0.0000068991926,0.000030988973,0.0019748774,0.53431886,0.000007219447,0.00017182542],"genre_scores_gemma":[0.63110566,4.423141e-7,0.000021583885,0.0000140536895,0.000012562687,0.0011774063,0.3673701,0.0000088961,0.00028928983],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99200827,0.004257696,0.0027481278,0.00024070467,0.0005677462,0.00017747768],"domain_scores_gemma":[0.9771638,0.014207004,0.008135959,0.00017958428,0.0002882364,0.000025415648],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0034050131,0.00013740714,0.0012708467,0.00029128935,0.00018831565,0.0000036805131,0.00011451691,0.00009076427,0.26934066],"category_scores_gemma":[0.040199585,0.00009828416,0.00025352274,0.0019879285,0.000015191232,0.00007485978,0.00010939526,0.00041248198,0.0000024058186],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0041741077,0.0015949869,0.82629913,0.0003027942,0.00094618544,0.0000040174477,0.00917694,0.002045367,0.000021285345,0.000034109347,0.15530735,0.000093696755],"study_design_scores_gemma":[0.019021548,0.000117804,0.97521156,0.0020942958,0.0001932925,1.9773314e-7,0.0023274694,0.00081639586,1.9290921e-7,0.00006870499,0.000064253865,0.00008426544],"about_ca_topic_score_codex":0.00006514238,"about_ca_topic_score_gemma":0.002250313,"teacher_disagreement_score":0.26933828,"about_ca_system_score_codex":0.00009536601,"about_ca_system_score_gemma":0.00025097132,"threshold_uncertainty_score":0.9678852},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6920790036","doi":"10.6084/m9.figshare.12885746","title":"Additional file 1 of The use of the PARIHS framework in implementation research and practice—a citation analysis of the literature","year":2020,"lang":"en","type":"article","venue":"Open MIND","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital; University of Ottawa","funders":"","keywords":"Data extraction; File format; Information system; Data analysis","score_opus":0.8418288736516842,"score_gpt":0.7290191403110461,"score_spread":0.11280973334063815,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6920790036","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.46285158,0.000009699745,0.000027045204,0.10680265,0.00011442781,0.0039434857,0.424246,9.776084e-7,0.002004097],"genre_scores_gemma":[0.9645428,0.000014922878,0.018358905,0.012089331,0.00009226571,0.00072600076,0.0032589,0.000014061068,0.00090278033],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99527144,0.0030753515,0.0006506885,0.00017393862,0.00063359574,0.0001949929],"domain_scores_gemma":[0.9742094,0.024078012,0.00075051846,0.0003183527,0.0005927327,0.00005095662],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0018373464,0.000052877796,0.00015521364,0.000115264105,0.0003744888,0.000032005693,0.00046046538,0.00007343862,0.2006184],"category_scores_gemma":[0.02553228,0.000029232624,0.00004056945,0.004323938,0.00018515528,0.00039677523,0.00045044717,0.00048029603,0.000010481065],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007701447,0.00003897131,0.022320611,0.00007336017,0.00006272629,1.7408867e-7,0.11321931,0.00004349688,0.0002651528,0.00054231763,0.85968596,0.0036709239],"study_design_scores_gemma":[0.0002653847,0.00004516279,0.5516978,0.00046500572,0.00004515175,2.3842276e-7,0.02701687,0.00023177672,0.00021220522,0.00034814203,0.4196315,0.000040760955],"about_ca_topic_score_codex":0.0005030659,"about_ca_topic_score_gemma":0.0020870527,"teacher_disagreement_score":0.52937716,"about_ca_system_score_codex":0.000056148005,"about_ca_system_score_gemma":0.00090645853,"threshold_uncertainty_score":0.9826761},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6920801032","doi":"10.6084/m9.figshare.19862814.v1","title":"Additional file 1 of A bibliometric analysis of the cannabis and cannabinoid research literature","year":2022,"lang":"en","type":"article","venue":"Figshare","topic":"Cannabis and Cannabinoid Research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Cannabis; Cannabinoid; Marijuana smoking; Systematic review; Bibliometrics; Scientific literature; Research data","score_opus":0.04354316672681682,"score_gpt":0.33133264943467905,"score_spread":0.2877894827078622,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6920801032","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027117177,0.0009793277,5.154409e-9,0.0017403417,0.0000107870455,0.00034239102,0.99173415,0.000008175456,0.0024731003],"genre_scores_gemma":[0.024583984,0.000006025587,0.000015466081,0.00015434869,0.000058864927,0.002771111,0.9213511,0.000019020434,0.051040124],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9973645,0.00020361382,0.00025583638,0.00028181862,0.0015852593,0.00030893748],"domain_scores_gemma":[0.99772006,0.00025878003,0.00012973434,0.0004823408,0.0012601231,0.00014896713],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00022064985,0.00010269118,0.00032162125,0.029815821,0.00026469628,0.000029371746,0.00029851668,0.000066039735,0.9918201],"category_scores_gemma":[0.007011517,0.00007545965,0.0002774081,0.18770124,0.00007251638,0.000047773967,0.0006697267,0.00066734175,0.000017837836],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004451129,0.00010121171,0.00011762233,0.00026509332,0.00021506006,0.00001616237,0.00016010836,0.000012702112,0.00006907202,0.0000050995545,0.9971732,0.0018201884],"study_design_scores_gemma":[0.00014832137,0.00017229634,0.21254738,0.00036580273,0.000053412852,0.000023767077,0.00014608914,0.00020032251,0.0001046827,0.0000073171595,0.78617555,0.000055055276],"about_ca_topic_score_codex":0.00012310431,"about_ca_topic_score_gemma":0.000050883165,"teacher_disagreement_score":0.9918023,"about_ca_system_score_codex":0.000119224256,"about_ca_system_score_gemma":0.00081806513,"threshold_uncertainty_score":0.98118037},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6921132696","doi":"10.6084/m9.figshare.26640319","title":"Additional file 1 of Sixty years of research in dental age estimation: a bibliometric study","year":2024,"lang":"en","type":"article","venue":"Figshare","topic":"Forensic Anthropology and Bioarchaeology Studies","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Table (database); Bibliometrics; Data file; Table of contents","score_opus":0.1558164522050577,"score_gpt":0.3786756981665016,"score_spread":0.2228592459614439,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6921132696","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016630734,0.00032474345,8.36906e-9,0.000066895,0.000050818086,0.00012758891,0.989339,0.000017600716,0.008410282],"genre_scores_gemma":[0.6522192,0.0000020403909,0.000031930926,0.000007548195,0.00009975917,0.00024389224,0.3454915,0.0000053878953,0.0018987946],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993581,0.00005231682,0.00014990196,0.000121498386,0.0001981222,0.000120018645],"domain_scores_gemma":[0.9976338,0.0021509577,0.000026974149,0.00006977388,0.00010605256,0.000012449126],"candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.000072142495,0.000048238424,0.00011911199,0.005356448,0.00006740646,0.000012631493,0.000095015,0.000034514782,0.99068147],"category_scores_gemma":[0.0008577569,0.00004249925,0.000037516813,0.0032438692,0.0011893043,0.000071944756,0.00013716325,0.00015196335,0.0010403963],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000064040855,0.00009562471,0.00003897211,0.000103301165,0.00003339818,0.00012108836,0.0031629645,4.4715756e-7,3.2098743e-8,0.0014342635,0.9919609,0.0030426295],"study_design_scores_gemma":[0.0002089375,0.0007060115,0.052666206,0.002819431,0.000008128914,0.000008374382,0.041190047,0.00016482089,0.000006423351,0.0012928735,0.90080154,0.00012719915],"about_ca_topic_score_codex":0.00028171515,"about_ca_topic_score_gemma":0.005239826,"teacher_disagreement_score":0.9896411,"about_ca_system_score_codex":0.00001342873,"about_ca_system_score_gemma":0.000052967855,"threshold_uncertainty_score":0.9997374},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6925270689","doi":"10.17605/osf.io/4me7c","title":"Bibliometric analysis of medical papers that use win ratio in the past decade (2012 to 2024)","year":2024,"lang":"en","type":"other","venue":"Open Science Framework","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Biomedical sciences; Behavioural sciences; MEDLINE; Health science; Health services research","score_opus":0.06455403815690414,"score_gpt":0.39941882798823775,"score_spread":0.3348647898313336,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6925270689","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0069382396,0.004569596,0.0019688038,0.010477364,0.003794666,0.0048971283,0.0010587599,0.0002244351,0.966071],"genre_scores_gemma":[0.1295751,0.0012151924,0.024281448,0.0090353405,0.0010614699,0.00071288407,0.00008847943,0.0017449219,0.83228517],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98840463,0.00043617326,0.0006571467,0.0016778511,0.007829891,0.000994311],"domain_scores_gemma":[0.99471897,0.0012506021,0.0004319573,0.0028779425,0.00011586949,0.00060464494],"candidate_categories":["metaepi_narrow","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.01313924,0.0005008671,0.0011069918,0.2767125,0.00012771708,0.0034426795,0.014953339,0.0008473277,0.036790606],"category_scores_gemma":[0.007852129,0.00032437107,0.00025836998,0.779644,0.0014922436,0.0008817478,0.0028260082,0.0016250111,0.007542449],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000152367975,0.00014214945,0.018077655,0.000027882712,0.0003537448,0.00007149766,0.0010868729,0.000096555144,0.000054104472,0.002862299,0.9711178,0.0060941847],"study_design_scores_gemma":[0.00012585461,0.000054295935,0.07701371,0.0014783875,0.0008488137,0.0000054395077,0.0008670708,0.0002875213,0.000014330639,0.00024455466,0.91845655,0.0006034681],"about_ca_topic_score_codex":0.003636627,"about_ca_topic_score_gemma":0.005810103,"teacher_disagreement_score":0.50293154,"about_ca_system_score_codex":0.00026189015,"about_ca_system_score_gemma":0.00097019435,"threshold_uncertainty_score":0.99992085},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"methods","study_design":"observational","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6927101395","doi":"10.3389/fneur.2023.1170731.s002","title":"Data_Sheet_2_Thirty years of research on traumatic brain injury rehabilitation: a bibliometric study.docx","year":2023,"lang":"en","type":"dataset","venue":"Figshare","topic":"Aquatic Ecosystems and Phytoplankton Dynamics","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Rehabilitation; Traumatic brain injury; Bibliometrics; Thematic analysis; Web of science; Public health; Head injury","score_opus":0.11756458243589658,"score_gpt":0.39497236172827416,"score_spread":0.2774077792923776,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6927101395","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0029324999,0.000018091574,5.0031172e-8,0.000043771837,0.00013624974,0.0012985532,0.99528694,0.00003819083,0.000245644],"genre_scores_gemma":[0.0014372856,0.000005856894,0.000030909683,0.000039066228,0.00008336022,0.00043808416,0.9975649,0.00003757348,0.00036297593],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9957885,0.00051775447,0.0006311693,0.0006629205,0.0019940666,0.0004055584],"domain_scores_gemma":[0.9930243,0.0049742637,0.0003210107,0.0014904728,0.00005330374,0.00013666076],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.001680896,0.00024499194,0.0004710337,0.007919645,0.00011126413,0.00007939214,0.0012326492,0.00029562446,0.16350135],"category_scores_gemma":[0.009390422,0.00024063463,0.000102510276,0.037293218,0.000048973634,0.00014998944,0.0008198259,0.0007709703,0.08689307],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014417832,0.00028243702,0.00010164636,0.0010236274,0.00002065728,0.000032140386,0.00013420523,0.000047909965,2.53682e-7,7.808407e-7,0.99681777,0.0015241337],"study_design_scores_gemma":[0.00020377136,0.0012063234,0.041369162,0.0019907518,0.000012400369,0.0000016597744,0.00034998005,0.00012216296,3.7685388e-7,0.000049894097,0.9544331,0.0002604032],"about_ca_topic_score_codex":0.0018100905,"about_ca_topic_score_gemma":0.0029343378,"teacher_disagreement_score":0.076608285,"about_ca_system_score_codex":0.0002695474,"about_ca_system_score_gemma":0.00008934399,"threshold_uncertainty_score":0.9989539},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6927560003","doi":"10.3389/fsurg.2022.881555.s001","title":"Table_1_Mapping Knowledge Structure and Themes Trends of Post-operative Rehabilitation of Hip Fractures in the Elderly: A Bibliometrics and Visualization Study.DOCX","year":2022,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Rehabilitation; Bibliometrics; Hip fracture; Web of science; Health care; MEDLINE","score_opus":0.04070221234237456,"score_gpt":0.3738047597388824,"score_spread":0.3331025473965078,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6927560003","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015837144,0.006301366,4.114067e-8,0.000015840418,0.000029422261,0.0011495609,0.9766291,0.000018453888,0.000019092688],"genre_scores_gemma":[0.13764283,0.000028061237,0.000008676522,0.000018780653,0.00003339561,0.00025115378,0.8619635,0.000047264082,0.0000063475004],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9966602,0.0010659451,0.0006817946,0.0005885486,0.00076675334,0.00023677749],"domain_scores_gemma":[0.99609184,0.0017526671,0.0008792135,0.0005939351,0.00063597533,0.000046341043],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0004758004,0.00043621057,0.0006904648,0.01422157,0.00016204533,0.00014147304,0.0005263667,0.00026633986,0.12503953],"category_scores_gemma":[0.007172463,0.0003288643,0.000065914406,0.022129847,0.00004784257,0.00035439487,0.0004430479,0.0005883295,0.000008714549],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007037346,0.00040889368,0.0001931161,0.0017719029,0.000088659435,0.0000047031886,0.011220615,0.00004807683,0.000068875146,0.000004691984,0.98392,0.0022001357],"study_design_scores_gemma":[0.0020809225,0.0045977896,0.121304214,0.0022683644,0.00027898696,0.000014440527,0.022004033,0.000110939545,0.00010992511,0.000066498855,0.8463886,0.00077530835],"about_ca_topic_score_codex":0.000258027,"about_ca_topic_score_gemma":0.0012537619,"teacher_disagreement_score":0.13753138,"about_ca_system_score_codex":0.00009331486,"about_ca_system_score_gemma":0.00015321773,"threshold_uncertainty_score":0.9999163},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6927587193","doi":"10.3389/fneur.2023.1237839.s003","title":"Presentation_2_Smartphone use in Neurology: a bibliometric analysis and visualization of things to come.PPTX","year":2023,"lang":"en","type":"article","venue":"Figshare","topic":"Mobile Health and mHealth Applications","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Relevance (law); Mobile phone; Bibliometrics; Scientific literature; Field (mathematics); Phone; Visualization; Citation; Focus (optics)","score_opus":0.1337374311889245,"score_gpt":0.4813603405864543,"score_spread":0.3476229093975298,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6927587193","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9232445,0.00024260013,0.00030031014,0.004227796,0.00020950566,0.008847027,0.061227646,0.0005170161,0.0011836452],"genre_scores_gemma":[0.8997958,0.00024556485,0.00022483643,0.00947602,0.000085376974,0.015532678,0.07371422,0.00004245443,0.0008830859],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99825,0.00027329213,0.00060173485,0.00028404375,0.00021144137,0.0003794558],"domain_scores_gemma":[0.997405,0.0015674626,0.00024616875,0.00025765246,0.0002600145,0.00026372698],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.00034214577,0.00008907116,0.00027986857,0.040349882,0.00020209976,0.000009857319,0.00011085549,0.00014902254,0.011910299],"category_scores_gemma":[0.0031828666,0.00009056871,0.000033912438,0.15648076,0.0000058322184,0.00018737651,0.00013900897,0.00021263592,0.001069552],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000052610867,0.000048550337,0.68839955,0.0010348879,0.000022578193,0.0000027109659,0.0014189846,0.00017510226,0.000033084918,0.00013859289,0.30413693,0.004536374],"study_design_scores_gemma":[0.00036080228,0.000041105803,0.9232071,0.00013782376,0.000014619112,1.652244e-7,0.000081374455,0.0029175803,0.00000832278,0.000032019565,0.07314035,0.000058706435],"about_ca_topic_score_codex":0.00078324624,"about_ca_topic_score_gemma":0.0007154321,"teacher_disagreement_score":0.23480755,"about_ca_system_score_codex":0.000026813204,"about_ca_system_score_gemma":0.00015503993,"threshold_uncertainty_score":0.99970824},"labels":[{"model":"gemma","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6931339759","doi":"10.5281/zenodo.7255702","title":"Bibliometric investigation of the integration of animal personality in conservation contexts","year":2022,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Data Analysis with R","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Memorial University of Newfoundland","funders":"","keywords":"Personality; Perception; Big Five personality traits","score_opus":0.06848129886701744,"score_gpt":0.2719039966556921,"score_spread":0.20342269778867467,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6931339759","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9748795,0.00005747582,0.01876552,0.00212031,0.00006737747,0.0003592014,0.00014534821,0.000114843475,0.0034904154],"genre_scores_gemma":[0.99901295,0.000005851467,0.0004829536,0.0001199524,0.000008064507,6.735775e-8,0.00024654742,0.00007956908,0.000044050004],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9979284,0.00079945795,0.0003217859,0.00023570392,0.0005926258,0.000122043006],"domain_scores_gemma":[0.99869597,0.00004765139,0.00027919628,0.00046473235,0.00047535327,0.000037122758],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0016399078,0.000062930565,0.00011357318,0.004040587,0.0005828006,0.00015316876,0.001554831,0.000019177747,0.0008381789],"category_scores_gemma":[0.0008174325,0.000057708767,0.000042370582,0.029940745,0.00012494416,0.00047005631,0.0015677296,0.00017817487,0.000039672923],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002802741,0.00070334447,0.010260304,0.00022536993,0.00012200753,0.0000075153794,0.018433306,0.0008820438,0.41201133,0.22157097,0.10471137,0.23079218],"study_design_scores_gemma":[0.0013169758,0.000825175,0.8001584,0.00006118581,0.000026161,0.000053453066,0.0011354244,0.063200526,0.01957376,0.0025185405,0.11083794,0.00029248846],"about_ca_topic_score_codex":0.00017023481,"about_ca_topic_score_gemma":0.0000027531878,"teacher_disagreement_score":0.78989804,"about_ca_system_score_codex":0.00013154514,"about_ca_system_score_gemma":0.000011853993,"threshold_uncertainty_score":0.9906783},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6938953243","doi":"10.60692/bfjw9-58v26","title":"Global academic output on COVID-19 and Guillain-Barre Syndrome: A bibliometric analysis","year":2022,"lang":"en","type":"article","venue":"Greater South Information System","topic":"Long-Term Effects of COVID-19","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Quartile; Bibliometrics; Ranking (information retrieval); Impact factor; Quarter (Canadian coin)","score_opus":0.035249804833544444,"score_gpt":0.2923769449952387,"score_spread":0.25712714016169425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6938953243","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9890878,0.000056683864,0.0050175567,0.001600728,0.00033236473,0.001277465,0.00058996934,0.00044935523,0.0015880378],"genre_scores_gemma":[0.9934879,0.0000010032971,0.00007870479,0.005789905,0.00005610905,0.00021336977,0.00014862843,0.000015481317,0.00020888084],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9967553,0.00024236951,0.0008895231,0.00036020405,0.0013237019,0.0004288743],"domain_scores_gemma":[0.9978326,0.00011114075,0.00052021036,0.00070724025,0.000114418406,0.0007144122],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0013748986,0.00032030948,0.0006987439,0.04160641,0.0004318736,0.00015080247,0.00030895887,0.0001830575,0.00013029505],"category_scores_gemma":[0.0006238244,0.00027456778,0.00024584358,0.06479465,0.00005658621,0.00041169638,0.00029214882,0.0003312972,0.0003032238],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033357306,0.000005296834,0.9842686,0.0016108245,0.0005889815,0.00014699309,0.007782245,0.0030721747,1.8899382e-7,0.00019229756,0.0015189502,0.00047987542],"study_design_scores_gemma":[0.0046871565,0.000606313,0.97478575,0.00010758079,0.0011310971,0.0019424922,0.0049267765,0.0063810907,0.0000046946097,0.000004761786,0.005011726,0.0004105854],"about_ca_topic_score_codex":0.000064602704,"about_ca_topic_score_gemma":3.4365937e-7,"teacher_disagreement_score":0.023188243,"about_ca_system_score_codex":0.0017007745,"about_ca_system_score_gemma":0.00029906837,"threshold_uncertainty_score":0.9999707},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6939288627","doi":"10.6084/m9.figshare.12703760.v1","title":"Additional file 1 of Quantifying professionalism in peer review","year":2020,"lang":"en","type":"article","venue":"Figshare","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Memorial University of Newfoundland; Carleton University; University of Victoria","funders":"","keywords":"Table (database); Inclusion (mineral); Diagram; Data flow diagram; File format","score_opus":0.9313384607065003,"score_gpt":0.6506207366814617,"score_spread":0.2807177240250386,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6939288627","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000058582646,0.0015561741,3.5227424e-7,0.0033885927,0.000018744102,0.00016190515,0.9881984,0.00000784903,0.0066621257],"genre_scores_gemma":[0.0014560734,0.000017728393,0.00031480566,0.0017562042,0.000049239996,0.00025832228,0.9928349,0.000006339474,0.0033063355],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99087226,0.00012111147,0.0004981175,0.0003522788,0.007937775,0.00021846704],"domain_scores_gemma":[0.9713029,0.020699635,0.0003891224,0.00039983966,0.0069310926,0.0002774022],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0015253241,0.000071708004,0.00023346969,0.005712451,0.00004641332,0.00016404867,0.0012108679,0.000057439192,0.99925435],"category_scores_gemma":[0.60238844,0.000049953236,0.00010642072,0.0892943,0.000010217098,0.000270212,0.00053759164,0.00021622058,0.014532097],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000016378516,0.00001872392,0.000024450299,0.000106426276,0.0000011254747,0.000007915633,0.000014867188,8.012828e-7,0.0000023755756,0.000005294814,0.9887109,0.011105489],"study_design_scores_gemma":[0.000056160236,0.000018014956,0.009121351,0.0052544377,3.2108497e-7,9.787993e-7,0.000030138088,0.00062679814,0.000012577462,0.000095545925,0.98472124,0.00006241915],"about_ca_topic_score_codex":0.0000017888418,"about_ca_topic_score_gemma":0.0000016376672,"teacher_disagreement_score":0.98472226,"about_ca_system_score_codex":0.00001923241,"about_ca_system_score_gemma":0.00027577268,"threshold_uncertainty_score":0.9862352},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6939306866","doi":"10.60692/dve4e-jq773","title":"Bibliometric profile of the global scientific research on methanol poisoning (1902–2012)","year":2015,"lang":"en","type":"article","venue":"Greater South Information System","topic":"Mycorrhizal Fungi and Plant Interactions","field":"Agricultural and Biological Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Methanol poisoning; Scopus; Methanol; Bibliometrics; Subject (documents); Web of science","score_opus":0.15091527132755386,"score_gpt":0.2961063334088549,"score_spread":0.14519106208130103,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6939306866","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96665514,0.00001310845,0.000026336525,0.00011494816,0.000996093,0.00032530478,0.0013993614,0.000050307088,0.030419378],"genre_scores_gemma":[0.99929905,1.7362639e-7,0.0000234963,0.000028071518,0.00007825012,0.000023471608,0.00008847184,3.7951895e-7,0.00045862098],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.998356,0.00019872053,0.00032637385,0.00011970505,0.00076565857,0.00023354522],"domain_scores_gemma":[0.998915,0.00012720806,0.00017665872,0.00012523707,0.00055601133,0.000099839686],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0013756431,0.0000848501,0.0001147138,0.0011664501,0.0003185368,0.0002899117,0.0003483104,0.000073811614,0.00023650205],"category_scores_gemma":[0.0001971187,0.000026781076,0.000067566194,0.024463778,0.000019254261,0.00049840217,0.00009680451,0.00012087808,0.002596305],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020012984,0.000017197304,0.9496648,0.00012562807,0.00003563998,0.0000013761781,0.00752598,0.000117821735,0.00034614283,0.004525984,0.030229425,0.007209918],"study_design_scores_gemma":[0.00069170684,0.0010229255,0.7487349,0.0014130009,0.00005164981,0.00022376678,0.08872559,0.0039653634,0.021382486,0.00010971746,0.13290949,0.00076939166],"about_ca_topic_score_codex":0.0001322787,"about_ca_topic_score_gemma":0.000003220998,"teacher_disagreement_score":0.20092984,"about_ca_system_score_codex":0.000099609344,"about_ca_system_score_gemma":0.000028408129,"threshold_uncertainty_score":0.99818027},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6939558766","doi":"10.6084/m9.figshare.28914933.v1","title":"Mapping global research on artificial intelligence in physical therapy: a bibliometric analysis from 1990 to 2023","year":2025,"lang":"en","type":"dataset","venue":"Figshare","topic":"Mycorrhizal Fungi and Plant Interactions","field":"Agricultural and Biological Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Bibliometrics; Inclusion (mineral); Physical activity","score_opus":0.15907454160283876,"score_gpt":0.382375023420935,"score_spread":0.22330048181809622,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6939558766","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025371748,0.00023003177,1.5874782e-7,0.00021959232,0.00012041616,0.00033989892,0.9960965,0.000020443716,0.0004357982],"genre_scores_gemma":[0.0015853478,0.0002143369,0.0000034395987,0.00024044962,0.00065955496,0.00019309239,0.99705577,4.5894137e-7,0.00004752526],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99787426,0.00024051397,0.00028393773,0.00065687683,0.0005226373,0.00042179812],"domain_scores_gemma":[0.99235207,0.0070992555,0.00007276451,0.00018625146,0.00015189104,0.00013775929],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0001528885,0.00023326094,0.00039624478,0.008806483,0.00015709337,0.00028556425,0.0008725859,0.00025103448,0.32538286],"category_scores_gemma":[0.0028258276,0.00010237121,0.00024591057,0.20529944,0.0000018227815,0.00005596134,0.00028019855,0.0006715416,0.012863179],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000053621938,0.00016602955,0.000048321734,0.000010076995,0.000074215706,0.000026427178,0.0000063373504,0.000040483083,0.000046052064,0.0000041030003,0.95472354,0.044800784],"study_design_scores_gemma":[0.000005206151,0.00013270519,0.0035542015,0.0009156485,0.0000131372835,3.1812053e-7,0.000065765074,0.00015112077,0.00004107968,0.00029485824,0.9946174,0.00020856234],"about_ca_topic_score_codex":0.015093249,"about_ca_topic_score_gemma":0.0075267754,"teacher_disagreement_score":0.31251967,"about_ca_system_score_codex":0.000111375644,"about_ca_system_score_gemma":0.00003356376,"threshold_uncertainty_score":0.99146533},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6939856199","doi":"10.6084/m9.figshare.c.3882952.v1","title":"Applications of altmetrics for Canadian institutions","year":2017,"lang":"en","type":"other","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Altmetrics; Social media; Citizen science; Field (mathematics)","score_opus":0.16152203117321842,"score_gpt":0.36092005323995974,"score_spread":0.19939802206674132,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6939856199","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[9.770441e-10,0.00089055195,0.000011447146,0.000005936374,0.000017764984,0.00095127564,0.5191561,0.00006204507,0.47890487],"genre_scores_gemma":[0.00006870028,0.0000029547464,0.000320741,0.000017966977,0.00025050275,0.002912675,0.554251,0.00041853698,0.4417569],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99928397,0.000005897233,0.000108929824,0.00023020161,0.0001291972,0.0002418056],"domain_scores_gemma":[0.998248,0.000028688912,0.0003950298,0.00091442274,0.000199419,0.00021446377],"candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.000023241395,0.00016348396,0.0002158087,0.0011902022,0.00018566755,0.00003179676,0.0006695486,0.00032752097,0.19904013],"category_scores_gemma":[0.0015475344,0.00018452923,0.000105391286,0.00032540027,0.000017570123,0.00003260936,0.00003972183,0.000102378704,0.027275547],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[2.6952986e-7,0.000010651272,0.0000012455822,0.00035268252,0.000033959644,6.067917e-7,0.0000033569643,8.674583e-7,5.3847106e-7,0.0003881711,0.9972072,0.0020004348],"study_design_scores_gemma":[0.00011707842,0.0000055172636,0.000022081318,0.0017122423,0.000028779923,9.960619e-7,0.0000010976567,0.0000055787727,0.000009940076,0.000017812841,0.9979014,0.00017748967],"about_ca_topic_score_codex":0.049536232,"about_ca_topic_score_gemma":0.6418434,"teacher_disagreement_score":0.59230715,"about_ca_system_score_codex":0.00019311199,"about_ca_system_score_gemma":0.0020576715,"threshold_uncertainty_score":0.97348183},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W6941972853","doi":"10.13386/j.issn1002-0306.2023110239","title":"Research Status and Hotspots Analysis of Cranberry Food Based on Bibliometric","year":2024,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Mycorrhizal Fungi and Plant Interactions","field":"Agricultural and Biological Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"CRANBERRY JUICE; Bibliometrics; Food composition data; Agriculture; Content analysis","score_opus":0.33619512819234737,"score_gpt":0.5583793190338242,"score_spread":0.22218419084147684,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6941972853","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9665368,0.024393333,0.000009262482,0.0002020127,0.0002661332,0.00021875839,0.0017050637,0.000023430446,0.0066451603],"genre_scores_gemma":[0.96789354,0.03157903,0.000019753443,0.000080506325,0.0000937285,0.000014001525,0.00014377809,0.0000027830588,0.00017286201],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.997474,0.00036785705,0.00050665095,0.00040028084,0.00087926636,0.00037191177],"domain_scores_gemma":[0.9913005,0.0077945753,0.0001873452,0.00012585439,0.00033312672,0.0002585888],"candidate_categories":["bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0017772685,0.00017271761,0.0005125926,0.014682626,0.00023218876,0.0013158154,0.00096498727,0.00009893388,0.042833697],"category_scores_gemma":[0.0006038382,0.0000755988,0.00024291994,0.07515511,0.000038625167,0.00082908216,0.00027656482,0.0004462767,0.000031905147],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00025188265,0.00026156797,0.74582815,0.000080961196,0.0006202639,0.000024169944,0.000040803898,0.00026884838,0.16159135,0.00022078776,0.030403938,0.060407262],"study_design_scores_gemma":[0.000045535762,0.00009824969,0.974971,0.00035189604,0.00021558724,0.0000028709571,0.00007877326,0.001609197,0.0041104774,0.0011453631,0.017192593,0.00017847736],"about_ca_topic_score_codex":0.0040258,"about_ca_topic_score_gemma":0.00033884367,"teacher_disagreement_score":0.22914283,"about_ca_system_score_codex":0.000041658743,"about_ca_system_score_gemma":0.000045547138,"threshold_uncertainty_score":0.99972093},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6945215779","doi":"10.22059/jipa.2023.363888.3376","title":"Identifying Key References and International Trends in Transparency: A Scientometric Analysis of Scientific Productions over the Last Decade","year":2023,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Transparency (behavior); Citation; Field (mathematics); Social media; Bibliometrics; Scientometrics; Bibliographic coupling; Key (lock)","score_opus":0.4166971683269905,"score_gpt":0.555231391958715,"score_spread":0.13853422363172446,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6945215779","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9931829,0.0008284853,0.000020512309,0.00029440352,0.0004658323,0.00011615646,0.00017729348,0.000015281517,0.004899153],"genre_scores_gemma":[0.99678576,0.002347131,0.000010642667,0.000021478043,0.000016687129,0.000021405842,0.00007812643,0.00000703129,0.0007117383],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99777776,0.00010606143,0.0005285027,0.00043431527,0.0008967191,0.00025662797],"domain_scores_gemma":[0.9990888,0.00011144863,0.00036847693,0.00026846767,0.000062821186,0.00009999143],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0018229401,0.00013177209,0.00030773107,0.0044333325,0.0003168012,0.0010167564,0.0016326872,0.000040980158,0.07934462],"category_scores_gemma":[0.00018840867,0.00010061303,0.00012802455,0.027818441,0.0005352743,0.0013531478,0.0006709052,0.00015804035,0.00002326055],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014343143,0.00011145158,0.963966,0.0000075440526,0.00012400048,0.0000020843363,0.0007124191,0.0003551146,0.018385062,0.000093599265,0.007840992,0.008387364],"study_design_scores_gemma":[0.00015522627,0.0000026024954,0.99330604,0.000030316369,0.00011006191,0.0000012747835,0.00069597317,0.00034155158,0.0014681469,0.00035567913,0.0034219173,0.00011120561],"about_ca_topic_score_codex":0.00088135444,"about_ca_topic_score_gemma":0.0030272328,"teacher_disagreement_score":0.07932136,"about_ca_system_score_codex":0.00019200768,"about_ca_system_score_gemma":0.00001920153,"threshold_uncertainty_score":0.9928458},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W6946022951","doi":"10.3389/fnhum.2022.981571.s001","title":"Image_1_A scientometric analysis of the 100 most cited articles on magnetic resonance guided focused ultrasound.TIFF","year":2022,"lang":"en","type":"other","venue":"Figshare","topic":"Chemical synthesis and alkaloids","field":"Chemistry","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Web of science; Bibliometrics; Citation analysis; Scientometrics; Informetrics","score_opus":0.03062549124684077,"score_gpt":0.2538663707263436,"score_spread":0.22324087947950283,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6946022951","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0037483685,0.005414683,4.3539046e-8,0.000047254052,0.000042991596,0.00018626524,0.3676294,0.00016344826,0.62276757],"genre_scores_gemma":[0.03565735,0.0000530049,0.000070123875,0.0003238905,0.0003343942,0.00033528192,0.068156615,0.00049864926,0.8945707],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99786377,0.000038851067,0.00035716017,0.00056409714,0.0008439875,0.00033213085],"domain_scores_gemma":[0.99779195,0.0005335574,0.00036227555,0.0011514492,0.00006926101,0.00009153594],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.000061709696,0.00030170835,0.000494971,0.00088644476,0.00010394498,0.0000713085,0.001031666,0.00020316515,0.93769735],"category_scores_gemma":[0.0035242254,0.00022837128,0.00045177297,0.0097460635,0.00003951017,0.000018122268,0.00028378854,0.00031012716,0.0003005939],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000062715108,0.00011357744,0.00042071744,0.00013853116,0.00015758646,0.0000044353005,0.000018732304,0.000006454373,0.011524663,0.0000024067851,0.9794371,0.008169498],"study_design_scores_gemma":[0.00024025959,0.000011626688,0.0020245216,0.0008610548,0.00031900004,0.0000010832106,0.000014985388,0.00011383205,0.036551427,0.000008565602,0.95953745,0.00031622106],"about_ca_topic_score_codex":0.00008847577,"about_ca_topic_score_gemma":0.000030432717,"teacher_disagreement_score":0.93739676,"about_ca_system_score_codex":0.00014391074,"about_ca_system_score_gemma":0.00005147862,"threshold_uncertainty_score":0.9312711},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6946057838","doi":"10.3389/fped.2024.1327118.s001","title":"Datasheet1_A bibliometric analysis of research on pediatric preoperative anxiety (2007–2022).pdf","year":2024,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Anxiety; Bibliometrics; Web of science; Psychological intervention; MEDLINE; Data collection","score_opus":0.14396194550861593,"score_gpt":0.43293198036029845,"score_spread":0.2889700348516825,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6946057838","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00004284,0.008416859,1.1778671e-8,0.000010166041,0.00023408321,0.0013946965,0.98871666,0.00016731536,0.0010173444],"genre_scores_gemma":[0.0001058534,0.0013426323,0.000008162994,0.000048367256,0.0013462402,0.0010282123,0.9947204,0.00023456376,0.0011655532],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9876872,0.0012367767,0.0012952948,0.0026072662,0.0056939824,0.0014795117],"domain_scores_gemma":[0.9898435,0.001952694,0.0007238467,0.0044002044,0.0025408925,0.0005388803],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0020876087,0.0010439028,0.0018023623,0.5235942,0.00028176073,0.0007451842,0.0037817226,0.001166194,0.8347614],"category_scores_gemma":[0.015494626,0.00095876784,0.00088813616,0.75557226,0.000063340565,0.00048483463,0.0030532868,0.004296141,0.77603996],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000698035,0.00043821082,0.00008037956,0.004042509,0.0027181022,0.00021546772,0.000022544315,0.0001546788,0.0000017392692,4.769173e-7,0.9921786,0.000077473676],"study_design_scores_gemma":[0.0002481266,0.00033746508,0.0050852858,0.0017247406,0.004278039,0.000005284503,0.000037640264,0.00017134019,0.000046338744,0.000002787314,0.987226,0.0008369496],"about_ca_topic_score_codex":0.00030131137,"about_ca_topic_score_gemma":0.00038711622,"teacher_disagreement_score":0.23197803,"about_ca_system_score_codex":0.00081161753,"about_ca_system_score_gemma":0.0013282958,"threshold_uncertainty_score":0.9992863},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6946221539","doi":"10.3389/fneur.2024.1442145.s002","title":"Data_Sheet_2_Tracing the evolving dynamics and research hotspots of spinal cord injury and surgical decompression from 1975 to 2024: a bibliometric analysis.XLSX","year":2024,"lang":"en","type":"dataset","venue":"Figshare","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Spinal cord injury; Decompression; Window of opportunity; Surgical decompression; Web of science; Spinal cord; Bibliometrics","score_opus":0.09595123589707295,"score_gpt":0.3952303303191762,"score_spread":0.29927909442210326,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6946221539","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014491773,0.0018260081,3.0029375e-7,0.00007275085,0.000069882524,0.0002609671,0.9828757,0.000013252584,0.0003893603],"genre_scores_gemma":[0.010261899,0.0004945869,0.000009483432,0.00003669582,0.000056511817,0.00006472247,0.9889704,0.00001225431,0.000093430885],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976142,0.0001286877,0.00030951176,0.00067558244,0.0009087152,0.00036331147],"domain_scores_gemma":[0.9985544,0.000457893,0.00011062449,0.00060623896,0.00004848789,0.00022233588],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.000376128,0.00022931541,0.0003576388,0.007010692,0.00021722204,0.0003853083,0.0006281677,0.00021303078,0.57633007],"category_scores_gemma":[0.00060387736,0.00016758303,0.000086939974,0.047774415,0.00008284067,0.00016055026,0.0027239346,0.00063404883,0.0033483785],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008176797,0.000027527802,0.00035233513,0.00014186284,0.000057760455,0.000042176816,0.000008638196,0.0000015859817,0.0000089662935,7.615604e-7,0.99515504,0.004121565],"study_design_scores_gemma":[0.0000805903,0.00017317929,0.13372649,0.0008170888,0.00013297451,0.0000071277923,0.00018299826,0.0005609166,0.0000081087555,0.0000071139334,0.8641055,0.00019790352],"about_ca_topic_score_codex":0.0012396394,"about_ca_topic_score_gemma":0.0022459682,"teacher_disagreement_score":0.57298166,"about_ca_system_score_codex":0.0002583618,"about_ca_system_score_gemma":0.000019657766,"threshold_uncertainty_score":0.99742764},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6949912290","doi":"10.5281/zenodo.4306413","title":"Technical Appendix to Economic Methodology: A Bibliometric Perspective","year":2020,"lang":"en","type":"other","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Sex and Gender in Healthcare","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval; Université de Sherbrooke","funders":"","keywords":"Perspective (graphical); Bibliometrics","score_opus":0.17287628781249562,"score_gpt":0.3830131059718255,"score_spread":0.21013681815932986,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6949912290","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000030748597,0.00094820233,0.003626207,0.0070008123,0.0002464028,0.0014669431,0.00064531656,0.0020516901,0.9839837],"genre_scores_gemma":[0.02682975,0.0050824704,0.07713074,0.015582057,0.012148686,0.0000019581062,0.023116816,0.098217845,0.74188966],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975438,0.00048104467,0.0002837873,0.0009061746,0.00034896465,0.00043621575],"domain_scores_gemma":[0.99830735,0.000059631016,0.00012410004,0.0006967035,0.00023868488,0.00057352433],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00074695615,0.00025014183,0.00048152154,0.011562636,0.00045407226,0.00019366357,0.00076860224,0.00029453886,0.10336979],"category_scores_gemma":[0.0018336396,0.00025791491,0.00011402776,0.008780989,0.0001238909,0.000036834143,0.0010761585,0.0006483538,0.0787204],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011650021,0.00005355494,0.0000026855992,0.0001785559,0.00009778095,0.000049238773,0.00046309296,0.0000010344567,0.00032325357,0.0026569413,0.9322894,0.06376791],"study_design_scores_gemma":[0.00042858138,0.0007283565,0.0002149051,0.00007987336,0.000052781314,0.0003734482,0.00075449114,0.000012265178,0.00006407079,0.00011596263,0.99695176,0.00022348184],"about_ca_topic_score_codex":0.00020791334,"about_ca_topic_score_gemma":0.0000027375154,"teacher_disagreement_score":0.242094,"about_ca_system_score_codex":0.000574186,"about_ca_system_score_gemma":0.000022601296,"threshold_uncertainty_score":0.9999873},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6950084680","doi":"10.5281/zenodo.7310985","title":"Demonstrating Librarian Research Impact through Bibliometrics","year":2022,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Methane Hydrates and Related Phenomena","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Bibliometrics; Information system; Citation analysis","score_opus":0.07284009682942479,"score_gpt":0.2955462552007163,"score_spread":0.22270615837129154,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6950084680","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25405148,0.00022040913,0.0010290178,0.0009327511,0.000151039,0.00062379916,0.00018017902,0.0005573801,0.74225396],"genre_scores_gemma":[0.9955471,0.000109270106,0.0007612554,0.00012608041,0.00008434188,1.4381605e-7,0.000543009,0.00086006976,0.0019687004],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9961631,0.0012044486,0.00027334283,0.000506306,0.0011948333,0.0006580173],"domain_scores_gemma":[0.99901724,0.00008562824,0.00009914091,0.00048160652,0.00008647352,0.00022993126],"candidate_categories":["sts","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0028907012,0.00013830645,0.00013428497,0.001053701,0.006863692,0.0007803788,0.0014395514,0.000055259425,0.19753018],"category_scores_gemma":[0.00051826687,0.0001321826,0.00006979833,0.013395209,0.0002935946,0.0005197506,0.004521249,0.0009234875,0.009710866],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015542822,0.00066527363,0.000378644,0.000035822617,0.000107697866,0.0001034835,0.006059007,0.015978524,0.025543923,0.005971005,0.78520554,0.15979564],"study_design_scores_gemma":[0.00036286024,0.00057522516,0.0014108056,0.0000056153362,0.000006778082,0.00015092126,0.0009609167,0.0018973919,0.00028933247,0.0018691064,0.99227256,0.0001985117],"about_ca_topic_score_codex":0.0001735302,"about_ca_topic_score_gemma":2.0340777e-7,"teacher_disagreement_score":0.74149567,"about_ca_system_score_codex":0.00046548987,"about_ca_system_score_gemma":0.0000053068848,"threshold_uncertainty_score":0.99442923},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6950286384","doi":"10.5281/zenodo.6363363","title":"Bibliometrics","year":2022,"lang":"en","type":"book-chapter","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Scientific Research and Technology","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Bibliometrics; Scientific literature; Visibility; Field (mathematics); Thematic map; Portrait","score_opus":0.062297906367189855,"score_gpt":0.2528011756629457,"score_spread":0.19050326929575584,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6950286384","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000106256275,0.00039288224,0.024509072,0.0012981732,0.00039736784,0.00039674118,0.0002165108,0.0020233996,0.9707552],"genre_scores_gemma":[0.006833541,0.0031000057,0.0056093754,0.00057726516,0.00046127674,2.2501689e-7,0.004725848,0.0056083794,0.9730841],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99664867,0.00014454336,0.0003184122,0.00096190907,0.0013455062,0.0005809442],"domain_scores_gemma":[0.99712014,0.00005786717,0.00018558718,0.0016551906,0.0007049076,0.00027632044],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.0014598266,0.0002295441,0.00023445628,0.007845238,0.0030502393,0.0021161337,0.0068302294,0.00017545036,0.060568664],"category_scores_gemma":[0.00069602666,0.00025200692,0.000119606564,0.004456043,0.00028505488,0.00045445136,0.0099319285,0.0009834887,0.019705312],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000038244084,0.000022551529,3.5156873e-8,0.000014851228,0.000022662334,0.00005668604,0.000051013827,0.000002707544,0.00006236709,0.42154434,0.32602906,0.25218993],"study_design_scores_gemma":[0.00019099205,0.00026149114,0.0000060335583,0.000011374998,0.0000041680028,0.00020420148,0.00001042065,0.00033606493,0.000051027873,0.015404572,0.9832728,0.00024685313],"about_ca_topic_score_codex":0.000006578152,"about_ca_topic_score_gemma":1.2002069e-7,"teacher_disagreement_score":0.6572437,"about_ca_system_score_codex":0.00030199985,"about_ca_system_score_gemma":0.000014409569,"threshold_uncertainty_score":0.9999932},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6957836299","doi":"10.6068/dp14ba845f22996","title":"Most Recent Data (2003). Statistics Canada. CANSIM: Science and Technology - Research and Development | Country: Canada | Table: Survey of innovation, selected service industries, type and location of organizations with which innovative business units cooperated and collaborated in order to develop products or processes | Variable: Organization in Europe, Testing laboratoriesá, Consultants | Units: %, 2003. Data-Planet™ Statistical Ready Reference by Conquest Systems, Inc. Dataset-ID: 075-001-182.","year":2015,"lang":"en","type":"other","venue":"Data Planet","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Economic statistics; Official statistics; Census; Service (business); Publication; Statistics education; Statistical analysis; Summary statistics; Big data","score_opus":0.060368978145685495,"score_gpt":0.2809606725088093,"score_spread":0.2205916943631238,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6957836299","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":"evaluation","model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00004747672,0.00041450752,0.0000148408335,0.000012022504,0.00006249336,0.0016265122,0.9977666,0.000039377286,0.000016185306],"genre_scores_gemma":[0.00004248689,0.00035237183,0.0018661458,0.00005359157,0.000007441371,0.000011568138,0.99740803,0.00018567592,0.00007266232],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99541926,0.0005026036,0.0011450538,0.0012927883,0.0010957312,0.00054453453],"domain_scores_gemma":[0.071819514,0.00049595966,0.0007884935,0.001122787,0.9256116,0.00016165682],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0026699235,0.00045502372,0.000699994,0.00077573775,0.0002237094,0.0002502155,0.0011266628,0.00030588315,0.00002529867],"category_scores_gemma":[0.95492715,0.00042479121,3.9323704e-11,0.9873588,0.0006636912,0.0006130741,0.0013282329,0.00058343273,0.0000011674077],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022883102,0.00008315578,0.009651798,0.0011854089,0.00007240093,0.000033992645,0.000016685992,0.000035200672,0.00019192351,0.00011625459,0.98834896,0.000035415218],"study_design_scores_gemma":[0.0009814115,0.000087005516,0.0012446862,0.00032601922,0.000032809086,0.000057117195,0.00054455706,0.00076576613,0.000011463009,9.843559e-10,0.99545556,0.0004936116],"about_ca_topic_score_codex":0.9961723,"about_ca_topic_score_gemma":0.99944746,"teacher_disagreement_score":0.9865831,"about_ca_system_score_codex":0.00040384548,"about_ca_system_score_gemma":0.96548104,"threshold_uncertainty_score":0.9998204},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"not_applicable","genre":"dataset","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W6957921271","doi":"10.60692/334bv-cx045","title":"Understanding the structure, characteristics, and future of collective intelligence using local and global bibliometric analyses","year":2022,"lang":"en","type":"article","venue":"Greater South Information System","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Collective intelligence; Scopus; Field (mathematics); Bibliometrics; Web of science; Scientometrics","score_opus":0.6239676538400103,"score_gpt":0.47354080569349677,"score_spread":0.15042684814651353,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6957921271","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6893403,0.00010417858,0.30891776,0.000036463072,0.0003633905,0.00022777241,0.0006211252,0.00001127384,0.00037777514],"genre_scores_gemma":[0.999774,0.0000034428112,0.00013146218,0.000036022382,0.00003270185,0.0000032316334,0.0000035153453,0.0000027656529,0.000012902816],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9949546,0.00026311903,0.0007794798,0.00023106713,0.0035276772,0.0002440951],"domain_scores_gemma":[0.99775237,0.000264703,0.0006018416,0.00030375912,0.0009473498,0.00012999507],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0049394188,0.00012148911,0.0002799032,0.10038493,0.00072020706,0.0014616194,0.00061035657,0.000053798096,0.000083503284],"category_scores_gemma":[0.00071840914,0.00007206674,0.000051551866,0.38094217,0.0001890494,0.0006677129,0.0007016678,0.00015067105,0.0000043829714],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000101205565,0.000003831779,0.9544839,0.00007338993,0.00007044859,0.000003553343,0.015201495,0.0009921644,0.000005793058,0.0021661618,0.0001790641,0.026718985],"study_design_scores_gemma":[0.00035107258,0.00015015481,0.6478035,0.000011088537,0.000030401501,0.00014710681,0.21754707,0.13292323,0.00015953026,0.00045217888,0.0002344595,0.00019022686],"about_ca_topic_score_codex":0.00002619854,"about_ca_topic_score_gemma":3.3360234e-7,"teacher_disagreement_score":0.3104337,"about_ca_system_score_codex":0.0005361908,"about_ca_system_score_gemma":0.00013531906,"threshold_uncertainty_score":0.99957496},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6958264603","doi":"10.60692/x790x-kek25","title":"Thematic trends and knowledge structure on cognitive behavior therapy for insomnia: A bibliometric and visualization analysis","year":2022,"lang":"en","type":"article","venue":"Greater South Information System","topic":"Sleep and related disorders","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Cognition; Visualization; Thematic analysis; MEDLINE; Trend analysis; Scientific literature; Randomized controlled trial","score_opus":0.03825590389313698,"score_gpt":0.30916414309307716,"score_spread":0.27090823919994017,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6958264603","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9928897,0.00014132363,0.0036140943,0.000010443207,0.00025933364,0.0006342459,0.00043368895,0.00008381481,0.0019333488],"genre_scores_gemma":[0.9989917,0.0000014972685,0.000010124149,0.00008995321,0.00001732071,0.0005185262,0.0001982807,0.00001212569,0.0001604277],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.999004,0.0001303737,0.00036290346,0.00017521104,0.00016984894,0.00015766689],"domain_scores_gemma":[0.9994163,0.000040865056,0.00024761673,0.00014125454,0.00010078502,0.00005318006],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00022364483,0.00015346611,0.00023869822,0.019571768,0.00030163943,0.00009402756,0.00006596191,0.00010573343,0.00023871502],"category_scores_gemma":[0.000010028914,0.00011943962,0.00007462611,0.016812958,0.00002739639,0.00016930266,0.000031011237,0.00009303509,0.000016347056],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007867884,0.00008807583,0.36315355,0.00024033185,0.0034258834,0.000002510253,0.49716842,0.00010027022,0.0000012964583,0.0015359739,0.00031480126,0.13318212],"study_design_scores_gemma":[0.014222355,0.0011344757,0.7578,0.000052270796,0.002764174,0.000070992995,0.21559662,0.007383434,0.000037445498,0.0000061810633,0.00037824505,0.00055382465],"about_ca_topic_score_codex":0.000005721918,"about_ca_topic_score_gemma":3.3564442e-7,"teacher_disagreement_score":0.39464647,"about_ca_system_score_codex":0.000036737558,"about_ca_system_score_gemma":0.000008270696,"threshold_uncertainty_score":0.99154055},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W6958371923","doi":"10.6084/m9.figshare.12885746.v1","title":"Additional file 1 of The use of the PARIHS framework in implementation research and practice—a citation analysis of the literature","year":2020,"lang":"en","type":"article","venue":"Figshare","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital; University of Ottawa","funders":"","keywords":"Data extraction; File format; Information system; Data analysis","score_opus":0.8468371197052131,"score_gpt":0.6984544193934029,"score_spread":0.14838270031181022,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6958371923","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038564785,0.00000678152,8.018937e-7,0.016677791,0.0000207365,0.0007910578,0.97847474,0.0000030389956,0.00016856592],"genre_scores_gemma":[0.5544877,0.000009638585,0.0024915428,0.03398729,0.0002142244,0.0031405468,0.4052509,0.000027039918,0.00039112792],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99601245,0.002326793,0.0005767235,0.00015514185,0.0007186662,0.00021024563],"domain_scores_gemma":[0.9644613,0.03348635,0.000766132,0.00030685344,0.0009257633,0.000053560078],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00051990594,0.00005794464,0.00013824207,0.00013118926,0.00034073734,0.000015180568,0.0002999379,0.000090198315,0.73838955],"category_scores_gemma":[0.09493505,0.000032723594,0.00005826166,0.004775387,0.000052947227,0.0002723935,0.00028286886,0.0005710845,0.000022861295],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011590689,0.000008850155,0.0019543597,0.00018015651,0.000018211715,6.8767555e-8,0.018083757,0.000019053005,0.000019478237,0.00012153232,0.9794177,0.0001652532],"study_design_scores_gemma":[0.00012944145,0.000025299261,0.54077494,0.0019954648,0.000017547849,1.6680583e-7,0.006581535,0.00023730127,0.000047964335,0.00018645919,0.44997066,0.000033188408],"about_ca_topic_score_codex":0.00011885244,"about_ca_topic_score_gemma":0.00061748934,"teacher_disagreement_score":0.7383667,"about_ca_system_score_codex":0.000060192953,"about_ca_system_score_gemma":0.0007284168,"threshold_uncertainty_score":0.9126887},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6958436139","doi":"10.6084/m9.figshare.21505976.v1","title":"Trends in Graves’ orbitopathy research in the past two decades: a bibliometric analysis","year":2022,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Web of science; China; Bibliometrics; Quality (philosophy); European union","score_opus":0.2327854665999361,"score_gpt":0.4586738128922817,"score_spread":0.2258883462923456,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6958436139","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00009531261,0.003044415,3.317044e-9,0.00006822511,0.00003987458,0.00060202245,0.995654,0.000054070886,0.00044208503],"genre_scores_gemma":[0.001632016,0.000053274918,0.0000025536294,0.000098333956,0.00030719937,0.0040429537,0.9935796,0.00013507607,0.00014895549],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9863838,0.00447232,0.0010215656,0.001622723,0.004828532,0.0016710355],"domain_scores_gemma":[0.99353665,0.0021970242,0.000478612,0.0032541633,0.00031502516,0.00021855303],"candidate_categories":["metaepi_narrow","bibliometrics","open_science","research_integrity","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0045153624,0.0006956818,0.001146368,0.72688323,0.00033218207,0.0007538118,0.0058804015,0.00046141935,0.85713595],"category_scores_gemma":[0.0043980684,0.0005751961,0.0006266908,0.88181365,0.000041821597,0.0002541678,0.0022466122,0.00581183,0.02159884],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000040997675,0.00039995005,0.00049944915,0.00027322798,0.00018829481,0.0021246625,0.0001424177,0.00015308952,4.418867e-7,3.6460395e-7,0.9950907,0.0010863559],"study_design_scores_gemma":[0.000638085,0.00010559966,0.05040347,0.0005679951,0.0001729307,0.000023763778,0.00030044053,0.000059933715,5.5410004e-7,0.000012310831,0.94715154,0.0005633963],"about_ca_topic_score_codex":0.004737303,"about_ca_topic_score_gemma":0.02225908,"teacher_disagreement_score":0.83553714,"about_ca_system_score_codex":0.0011710805,"about_ca_system_score_gemma":0.00034101558,"threshold_uncertainty_score":0.99966997},"labels":[{"model":"gemma","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6958690911","doi":"10.6084/m9.figshare.20677997.v1","title":"Additional file 3 of Factors associated with reporting of the Prevention of Falls Network Europe (ProFaNE) core outcome set domains in randomized trials on falls in older people: a citation analysis and correlational study","year":2022,"lang":"en","type":"article","venue":"Figshare","topic":"Distributed and Parallel Computing Systems","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba; George & Fay Yee Centre for Healthcare Innovation","funders":"","keywords":"Randomized controlled trial; Core (optical fiber); Set (abstract data type); Outcome (game theory); Poison control","score_opus":0.12854278938666552,"score_gpt":0.32643215469267517,"score_spread":0.19788936530600965,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6958690911","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2861233,0.000008171964,0.00022660317,0.0000068016666,0.000036525318,0.0011931896,0.7122955,0.000013000084,0.00009691558],"genre_scores_gemma":[0.7676621,3.5437772e-8,0.00007300288,0.000004900451,0.0000063074785,0.00022236397,0.23196667,0.0000037883942,0.00006083952],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.9955397,0.0017631736,0.0017624744,0.00023947343,0.00057938416,0.0001157808],"domain_scores_gemma":[0.9852483,0.009196832,0.005127734,0.0002002003,0.00020870638,0.000018191586],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0022565897,0.00011089457,0.00095003226,0.00022030537,0.00007241362,0.000019589992,0.00025537386,0.000034244145,0.064945176],"category_scores_gemma":[0.021311995,0.00007894874,0.00019653038,0.0026534314,0.000009116297,0.00007355026,0.0001556996,0.00014952828,6.7836584e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0015372701,0.0011944892,0.34475663,0.00015255697,0.001392635,0.000016617834,0.011224922,0.45739284,0.000004669976,0.00015443403,0.18193784,0.00023507692],"study_design_scores_gemma":[0.009322141,0.00012923132,0.9580098,0.0009877064,0.000064447,0.0000011069408,0.00029185452,0.030999484,9.0451516e-7,0.000041587176,0.000057331465,0.000094410134],"about_ca_topic_score_codex":0.000030713927,"about_ca_topic_score_gemma":0.00024578723,"teacher_disagreement_score":0.6132532,"about_ca_system_score_codex":0.000042797714,"about_ca_system_score_gemma":0.00013849135,"threshold_uncertainty_score":0.9869319},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6958809712","doi":"10.6084/m9.figshare.c.6665805.v1","title":"The 100 most cited papers on total anomalous pulmonary venous connection: a bibliometric analysis","year":2023,"lang":"en","type":"other","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Annals; Citation; Bibliometrics; Citation analysis; Science Citation Index; Web of science; Impact factor; Citation index; MEDLINE","score_opus":0.03775398544670241,"score_gpt":0.27881211685392177,"score_spread":0.24105813140721935,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6958809712","genre_codex":"dataset","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000375747,0.003565964,1.6556923e-8,0.000023769358,0.00018075579,0.0005953621,0.5579731,0.0024916404,0.43513185],"genre_scores_gemma":[0.015217747,0.000090562884,0.0000020767334,0.000085991414,0.0011525594,0.00079613324,0.32282832,0.0040035034,0.6558231],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9964079,0.00019377556,0.00040109645,0.0010196744,0.0011701752,0.00080736255],"domain_scores_gemma":[0.99681145,0.0007315426,0.0006410302,0.001391105,0.00018758989,0.00023728957],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.00014606884,0.00070332625,0.0007573636,0.16227545,0.00045253485,0.0003663731,0.0007883454,0.0006504418,0.47564417],"category_scores_gemma":[0.0026341246,0.0005644043,0.0006303344,0.4698986,0.00002860297,0.00006563674,0.00023865963,0.000615922,0.16840377],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000041408133,0.00014536263,0.00006759007,0.00010629306,0.0024606963,0.00059802335,0.000015294065,0.000116346295,0.000007690137,0.0000015866763,0.98843706,0.0080026295],"study_design_scores_gemma":[0.00022198616,0.00018766896,0.028806908,0.0012330723,0.00052568456,0.000052450112,0.000038174177,0.0003181842,0.0000040581367,0.0000016631866,0.9678946,0.0007155474],"about_ca_topic_score_codex":0.00040192553,"about_ca_topic_score_gemma":0.0010095149,"teacher_disagreement_score":0.30762318,"about_ca_system_score_codex":0.0003978211,"about_ca_system_score_gemma":0.00013718348,"threshold_uncertainty_score":0.99968076},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6959036259","doi":"10.6084/m9.figshare.c.7228732.v1","title":"Insight into the research history and trends of total anomalous pulmonary venous connection: a bibliometric analysis","year":2024,"lang":"en","type":"other","venue":"Figshare","topic":"Peanut Plant Research Studies","field":"Agricultural and Biological Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Annals; Bibliometrics; Citation; Total anomalous pulmonary venous connection; Citation analysis; Impact factor; Web of science; Pulmonary medicine","score_opus":0.10615591343823519,"score_gpt":0.3104730446133025,"score_spread":0.2043171311750673,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6959036259","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0022158169,0.24092527,4.2776334e-9,0.0004083319,0.0000915253,0.00026286737,0.11000158,0.00012970244,0.6459649],"genre_scores_gemma":[0.13116556,0.0015202804,0.0000029093633,0.00002099065,0.00072440016,0.00021026644,0.06813398,0.000008095898,0.79821354],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99834603,0.00017551702,0.00015310638,0.00041908148,0.0006210568,0.0002852068],"domain_scores_gemma":[0.9990184,0.00057734066,0.00008648058,0.00011310033,0.00012539349,0.00007931694],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00022814034,0.0001691457,0.00033572325,0.01281358,0.00011283642,0.000058516765,0.0003078831,0.0001802447,0.1869336],"category_scores_gemma":[0.00045942943,0.000055790133,0.00015785752,0.06934361,0.00008578377,0.000033805696,0.00048175754,0.00038709329,0.00012518227],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000006679106,0.000044588753,0.000051826737,0.00016488013,0.00031361208,0.00007922117,0.00014173373,7.154197e-8,0.00015277488,0.0000020187847,0.94106334,0.057979275],"study_design_scores_gemma":[0.000017517394,0.00018666267,0.03164482,0.00033820665,0.000053266485,0.000014396681,0.000087263754,0.000014896105,0.000006277031,0.000008375145,0.967506,0.00012235285],"about_ca_topic_score_codex":0.0017595107,"about_ca_topic_score_gemma":0.0016686785,"teacher_disagreement_score":0.23940499,"about_ca_system_score_codex":0.0001010125,"about_ca_system_score_gemma":0.000020474945,"threshold_uncertainty_score":0.99837536},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","insufficient_payload"],"domain":null,"study_design":"observational","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6963817357","doi":"10.22067/ijaaf.2025.45706.1477","title":"An Analysis of scientific research in the field of international financial reporting standards: A Scientometric study","year":2025,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Tree-ring climate responses","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scientometrics; Centrality; Scopus; Latent Dirichlet allocation; Relevance (law); Transparency (behavior); Field (mathematics); Globalization; China","score_opus":0.347863184203711,"score_gpt":0.6283153267886448,"score_spread":0.2804521425849338,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963817357","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9940522,0.0021536169,0.000042850952,0.000114046925,0.00043953632,0.00036400327,0.0001670234,0.0000056486256,0.0026611015],"genre_scores_gemma":[0.9995034,0.00030360208,0.00006290721,0.000030383357,0.000023644496,0.0000038343646,0.000015983474,0.0000026855753,0.000053573112],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99413675,0.0009898478,0.0015487142,0.0004355075,0.0025699511,0.00031924405],"domain_scores_gemma":[0.99426436,0.0027745008,0.0012878921,0.0005838627,0.0010133076,0.00007609579],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.028638793,0.00012252956,0.0005864843,0.012355732,0.00031322293,0.0012754592,0.0037151347,0.00005223804,0.003292601],"category_scores_gemma":[0.011198843,0.000089182264,0.00015210033,0.023444189,0.00028938043,0.0011267771,0.0002830743,0.00034870705,9.421272e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031688478,0.00032910533,0.96656466,0.000031635474,0.00016060633,0.000035419842,0.00053739524,0.0024020025,0.0012438728,0.000016660228,0.00088077126,0.02748097],"study_design_scores_gemma":[0.00021190578,0.000044474837,0.99452126,0.0001512907,0.000094944226,0.0000011601676,0.0012636989,0.0013343923,0.0014859055,0.00041444454,0.00039627496,0.00008026324],"about_ca_topic_score_codex":0.010329596,"about_ca_topic_score_gemma":0.0103293825,"teacher_disagreement_score":0.027956568,"about_ca_system_score_codex":0.000036589063,"about_ca_system_score_gemma":0.0006376257,"threshold_uncertainty_score":0.9997613},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6963891123","doi":"10.22034/ijism.2023.1977996.0","title":"Bibliometric Analysis and Visualization of Scientific Publications of Iran University of Medical Sciences during 1980-2020","year":2024,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Ophthalmology and Visual Health Research","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Visualization; Information visualization; Data visualization; Web of science; Population; Citation; Scopus","score_opus":0.5254124890958658,"score_gpt":0.6705567993968441,"score_spread":0.1451443103009783,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6963891123","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98533696,0.012864916,0.00028032757,0.0004426889,0.0001164447,0.00022930832,0.000030899137,0.000010773881,0.0006876819],"genre_scores_gemma":[0.9906228,0.008970527,0.00007897399,0.0000057664784,0.00002411269,0.0000011890426,0.000014507069,0.0000073215247,0.00027479255],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99711376,0.0002480777,0.00060910854,0.00033878605,0.0014802989,0.0002099722],"domain_scores_gemma":[0.9980044,0.0005202089,0.0003992081,0.00023702298,0.0005521589,0.00028703877],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.004616125,0.00009612393,0.0005857147,0.06465583,0.00027836926,0.00019658382,0.0009855349,0.000118666314,0.00596008],"category_scores_gemma":[0.00094485615,0.00008444138,0.00013474014,0.1590188,0.0011330916,0.0009270989,0.00052104745,0.00023508392,0.0000017935007],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000097018434,0.0002983218,0.97544795,0.0012284704,0.00034100385,0.000018803745,0.00017943491,0.000008107587,0.01739986,0.00055682496,0.0005768985,0.0038472728],"study_design_scores_gemma":[0.0003053659,0.000046991976,0.9870509,0.0004946761,0.0002775104,0.000023963354,0.00013564574,0.0013613801,0.009844145,0.00023931859,0.00015673152,0.000063363776],"about_ca_topic_score_codex":0.0010391185,"about_ca_topic_score_gemma":0.000044646775,"teacher_disagreement_score":0.09436297,"about_ca_system_score_codex":0.00003918962,"about_ca_system_score_gemma":0.00068329397,"threshold_uncertainty_score":0.9949486},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6965047741","doi":"10.3389/fpubh.2023.1133972.s005","title":"Table_4_Research hotspots and new trends in the impact of resistance training on aging, bibliometric and visual analysis based on CiteSpace and VOSviewer.pdf","year":2023,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Frontier; Intervention (counseling); Visualization; Web of science; Perspective (graphical); Inclusion (mineral); Field (mathematics)","score_opus":0.18347450707663365,"score_gpt":0.45043762042510566,"score_spread":0.266963113348472,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6965047741","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0020709594,0.002888529,7.53213e-8,0.000085340325,0.000007922398,0.0003827857,0.9943853,0.00003741372,0.00014167074],"genre_scores_gemma":[0.011277734,0.00024134207,0.0000061525316,0.00006739889,0.00008428816,0.000105847066,0.98760074,0.000101661266,0.0005148044],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9951758,0.00068352796,0.0005140602,0.0011336554,0.0016481772,0.00084477145],"domain_scores_gemma":[0.9952806,0.0027481057,0.00053704454,0.00086617074,0.00017007075,0.00039799005],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0013495325,0.00069343543,0.0012230857,0.1380029,0.00016147591,0.00061257265,0.0006350831,0.0003945887,0.01989197],"category_scores_gemma":[0.0055085677,0.00051869644,0.0002670243,0.21636856,0.000060785467,0.00020897834,0.00024620336,0.0013695495,0.00086274784],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024056544,0.000094996685,0.0003409688,0.0005347773,0.00035798378,0.00015749509,0.0001150611,0.0002471011,0.0000074790496,6.907469e-8,0.99462444,0.0032790746],"study_design_scores_gemma":[0.0030199164,0.0009836358,0.76722634,0.008195571,0.00056551374,0.0000071497075,0.0001887125,0.005107233,0.000009574815,0.0000040449477,0.213681,0.0010113014],"about_ca_topic_score_codex":0.002588894,"about_ca_topic_score_gemma":0.004268886,"teacher_disagreement_score":0.78094345,"about_ca_system_score_codex":0.00020254856,"about_ca_system_score_gemma":0.00035285467,"threshold_uncertainty_score":0.9999152},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6965212874","doi":"10.3389/fnut.2024.1485642.s001","title":"Data_Sheet_1_Mapping the evolution and impact of ketogenic diet research on diabetes management: a comprehensive bibliometric analysis from 2005 to 2024.CSV","year":2024,"lang":"en","type":"dataset","venue":"Figshare","topic":"Biomedical and Chemical Research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Diabetes mellitus; Diabetes management; Bibliometrics; MEDLINE; Web of science; Clinical nutrition","score_opus":0.13015359248684022,"score_gpt":0.4285460180750685,"score_spread":0.29839242558822826,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6965212874","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023878892,0.007440192,1.796201e-7,0.00036845557,0.000035069388,0.0007409488,0.98879284,0.000018019202,0.0002163864],"genre_scores_gemma":[0.004263494,0.0005670747,0.000009107951,0.00012609894,0.0003050727,0.0002578094,0.9940646,0.000022333732,0.0003843779],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99601823,0.00015187325,0.00038701383,0.0007909412,0.0019374613,0.0007144904],"domain_scores_gemma":[0.9962851,0.0016291558,0.000084429426,0.0011260482,0.00034071045,0.0005345795],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.00034775527,0.0003032877,0.00069452834,0.044410974,0.00010203195,0.00014719793,0.00070551207,0.00034370983,0.13573004],"category_scores_gemma":[0.0019550843,0.0001818661,0.0003737753,0.11468242,0.00008611078,0.000047558206,0.001703936,0.0014465105,0.0052314503],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017398465,0.00008532199,0.00011780931,0.0020971606,0.003431157,0.000060089587,0.000006146362,0.0000021900053,0.00008621919,2.6201093e-7,0.9921637,0.0017759547],"study_design_scores_gemma":[0.00029847719,0.0006640936,0.039193317,0.0052912193,0.0009977382,8.376018e-7,0.000067796005,0.00023334914,0.000038287773,0.000038266175,0.9529917,0.00018486494],"about_ca_topic_score_codex":0.0006192402,"about_ca_topic_score_gemma":0.00005870145,"teacher_disagreement_score":0.1304986,"about_ca_system_score_codex":0.00036424104,"about_ca_system_score_gemma":0.0001541417,"threshold_uncertainty_score":0.99554306},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6965247951","doi":"10.3389/fpubh.2022.847420.s001","title":"Data_Sheet_1_Antimicrobial Stewardship in Surgery: A Literature Bibliometric Analysis.PDF","year":2022,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Citation; Web of science; Stewardship (theology); Citation analysis; MEDLINE; Observational study","score_opus":0.0583893831281018,"score_gpt":0.3092390181927466,"score_spread":0.25084963506464475,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6965247951","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00002637366,0.0044830674,6.816284e-9,0.00001563469,0.00025564406,0.00071351934,0.99426335,0.00017316655,0.00006923237],"genre_scores_gemma":[0.000056379038,0.00011748096,0.0000064300175,0.00032595996,0.00038237975,0.00071067456,0.997934,0.00020687305,0.00025983015],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99240553,0.0010242407,0.001273928,0.0020382595,0.0018817893,0.001376274],"domain_scores_gemma":[0.99378175,0.001078298,0.0011873165,0.0032729616,0.00032560207,0.00035406285],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.00095936854,0.001180454,0.0022728376,0.4881391,0.00022299867,0.0014875561,0.003056664,0.0010412361,0.9829996],"category_scores_gemma":[0.008834179,0.0012961702,0.0013654436,0.7142071,0.000016803886,0.0008349662,0.0024893074,0.003283607,0.10450698],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007613082,0.000349375,0.000099205135,0.001678785,0.0009566188,0.0026857567,0.000014702098,0.00005236177,0.0000042374163,3.8161705e-8,0.9939712,0.00011157425],"study_design_scores_gemma":[0.00033217238,0.000027374646,0.0074197724,0.002697199,0.0007868674,0.000054654858,0.0000056986446,0.000009604725,0.0000070240067,8.2790257e-7,0.9873789,0.0012799172],"about_ca_topic_score_codex":0.00047709604,"about_ca_topic_score_gemma":0.0016687248,"teacher_disagreement_score":0.87849265,"about_ca_system_score_codex":0.00073506456,"about_ca_system_score_gemma":0.0008842233,"threshold_uncertainty_score":0.999549},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6976347620","doi":"10.60692/09n5e-amd09","title":"Bibliometric Analysis of Literature on Acute Respiratory Distress Syndrome Treatments Published Between 2000 and 2019","year":2020,"lang":"en","type":"article","venue":"Greater South Information System","topic":"Respiratory Support and Mechanisms","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"ARDS; Acute respiratory distress; Bibliometrics; Extracorporeal membrane oxygenation; Modalities; Web of science; MEDLINE; Science Citation Index","score_opus":0.037508543879153686,"score_gpt":0.2535730064098743,"score_spread":0.2160644625307206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6976347620","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9961015,0.000039859435,0.0006523055,0.00007674761,0.00007630122,0.0004626944,0.0016849801,0.00009407577,0.0008115424],"genre_scores_gemma":[0.99871206,8.51791e-7,0.000058884874,0.00044785845,0.000044290347,0.000022490594,0.0005329003,0.000011370985,0.0001693054],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9982466,0.00005395285,0.00074694114,0.00022599088,0.0005216056,0.00020491694],"domain_scores_gemma":[0.9986304,0.000011904117,0.0004075762,0.00038698892,0.0002314376,0.0003317004],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00023528331,0.0002224336,0.0007231603,0.02778706,0.00005579953,0.00018001722,0.00011138833,0.0001824969,0.000055414293],"category_scores_gemma":[0.000033723365,0.00015973902,0.00018222701,0.04456308,0.000022112863,0.0007763357,0.00004226189,0.00012718933,0.00010048224],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021245609,0.0000061416604,0.9817006,0.00051964185,0.0026087968,0.00006640451,0.012622585,0.000007023525,0.000016299815,0.000017433957,0.00039148767,0.0018311327],"study_design_scores_gemma":[0.0020391024,0.00078788947,0.991642,0.00019597272,0.0025242635,0.000019452307,0.0008267617,0.00018143891,0.00088492635,1.6085238e-7,0.0007336879,0.00016432616],"about_ca_topic_score_codex":0.0000023713753,"about_ca_topic_score_gemma":1.8066975e-8,"teacher_disagreement_score":0.016776022,"about_ca_system_score_codex":0.000075309756,"about_ca_system_score_gemma":0.000053398664,"threshold_uncertainty_score":0.98323214},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6976740840","doi":"10.6084/m9.figshare.27938184.v1","title":"A bibliometric analysis of the 100 most-influential papers in the field of anti-diabetic drugs","year":2024,"lang":"en","type":"article","venue":"Figshare","topic":"Management, Economics, and Public Policy","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Scopus; Citation; Citation analysis; Web of science; Field (mathematics); Original research","score_opus":0.02114180454974881,"score_gpt":0.2654074224908753,"score_spread":0.24426561794112647,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6976740840","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5940821,0.0010166486,0.0000014757073,0.0053514685,0.0004988849,0.00087149953,0.014151964,0.000102399965,0.38392356],"genre_scores_gemma":[0.9945233,0.000010036948,0.0000013507532,0.0032739413,0.00021810907,0.00003410522,0.001723767,0.000009944719,0.00020546629],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99909824,0.000015950785,0.0003133095,0.00018377944,0.00019548392,0.00019322339],"domain_scores_gemma":[0.9991361,0.000258752,0.00018385277,0.00036699214,0.000048164424,0.0000061656237],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0002547043,0.00010490754,0.00021606838,0.024456237,0.000045561614,0.00028990224,0.0006411042,0.000049762497,0.048575994],"category_scores_gemma":[0.000736435,0.00007063334,0.00022812995,0.088436775,0.0000118619155,0.00033340298,0.00024947664,0.00010906521,0.00015555113],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010625739,0.00014594673,0.03190379,0.0030535923,0.0012336883,0.0000071179547,0.0006895113,0.0010294007,0.000038589846,0.006698925,0.88737726,0.06781154],"study_design_scores_gemma":[0.00018574562,0.0000059806007,0.36090186,0.00031843106,0.00038629395,1.03924485e-7,0.00020589647,0.013665294,0.00003307195,0.00022503085,0.6239176,0.0001547184],"about_ca_topic_score_codex":0.0008579778,"about_ca_topic_score_gemma":0.00017651978,"teacher_disagreement_score":0.4004412,"about_ca_system_score_codex":0.000011410591,"about_ca_system_score_gemma":0.000023975033,"threshold_uncertainty_score":0.9866007},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6977321263","doi":"10.60692/tz830-ra793","title":"Global Research Mapping of Psycho-Oncology Between 1980 and 2021: A Bibliometric Analysis","year":2022,"lang":"en","type":"article","venue":"Greater South Information System","topic":"Cancer survivorship and care","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Productivity; Quality (philosophy); Psychosocial; Scientometrics; Scientific literature","score_opus":0.11051257120606353,"score_gpt":0.3486990394433388,"score_spread":0.23818646823727527,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6977321263","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98606217,0.000055680037,0.0032041227,0.00011919247,0.00019148986,0.00031597895,0.00038126626,0.000031553922,0.009638561],"genre_scores_gemma":[0.9995925,0.0000011109179,0.0000525471,0.00005062096,0.00007734458,0.000079658654,0.000066313965,0.0000039081187,0.00007599256],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99810386,0.00027798172,0.00055761245,0.00014518102,0.00068688486,0.00022847131],"domain_scores_gemma":[0.998971,0.00003504116,0.00021708125,0.0003059543,0.000349055,0.00012187601],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0019204404,0.000090449415,0.00042059438,0.036511432,0.00019453964,0.000042019656,0.00011543861,0.00007116153,0.000111922614],"category_scores_gemma":[0.000035910904,0.00008170258,0.00011343467,0.105188526,0.00004463367,0.00015593001,0.00016083482,0.00019209369,0.00002941718],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009204892,0.0000023250107,0.9513738,0.00032414155,0.00033410825,0.000003494031,0.040170178,0.000017185614,4.9231545e-7,0.00007811393,0.00011890916,0.0074851895],"study_design_scores_gemma":[0.0010170062,0.00023729079,0.87525594,0.000035670993,0.00025529953,0.000035805493,0.11966773,0.00031984286,0.000011216924,0.0000020244636,0.0030803932,0.00008174619],"about_ca_topic_score_codex":0.00034249522,"about_ca_topic_score_gemma":0.0000053027247,"teacher_disagreement_score":0.07949755,"about_ca_system_score_codex":0.00042505228,"about_ca_system_score_gemma":0.00010605304,"threshold_uncertainty_score":0.97440886},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6977338435","doi":"10.6084/m9.figshare.20677991.v1","title":"Additional file 1 of Factors associated with reporting of the Prevention of Falls Network Europe (ProFaNE) core outcome set domains in randomized trials on falls in older people: a citation analysis and correlational study","year":2022,"lang":"en","type":"article","venue":"Figshare","topic":"Balance, Gait, and Falls Prevention","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba; George & Fay Yee Centre for Healthcare Innovation","funders":"","keywords":"Core (optical fiber); Randomized controlled trial; Set (abstract data type); Outcome (game theory); Poison control","score_opus":0.1635515612711525,"score_gpt":0.4035147341800521,"score_spread":0.23996317290889962,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6977338435","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.46202153,0.000008095448,0.0000020805294,0.000006919926,0.000030900086,0.001968696,0.53578365,0.0000072014313,0.0001709392],"genre_scores_gemma":[0.6292855,4.3659594e-7,0.00002175087,0.000014033195,0.00001253342,0.001194409,0.36916968,0.000008887781,0.00029277176],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99200636,0.004261479,0.002746252,0.00024076771,0.0005676702,0.00017746922],"domain_scores_gemma":[0.977078,0.014290834,0.008135271,0.00017976243,0.0002906944,0.000025421037],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0034040234,0.00013741525,0.0012704897,0.00029125737,0.00018825116,0.0000036726196,0.000114506656,0.00009076429,0.27623168],"category_scores_gemma":[0.04066279,0.00009828942,0.00025351983,0.001988751,0.000015188064,0.00007480288,0.00010942624,0.0004125626,0.0000024254675],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0040983637,0.0015873959,0.8206068,0.00029910728,0.00094233925,0.0000039473884,0.009125445,0.0020590543,0.00002141405,0.00003312394,0.16113079,0.00009218763],"study_design_scores_gemma":[0.018878942,0.00011634602,0.975353,0.0020947705,0.0001936229,1.976623e-7,0.0023318222,0.0008133365,1.9288936e-7,0.00006799542,0.000065511034,0.00008428027],"about_ca_topic_score_codex":0.00006501538,"about_ca_topic_score_gemma":0.0022301269,"teacher_disagreement_score":0.27622926,"about_ca_system_score_codex":0.000095358395,"about_ca_system_score_gemma":0.00025164106,"threshold_uncertainty_score":0.96741813},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W6977353442","doi":"10.6084/m9.figshare.25264615.v1","title":"Research Output of Canadian Journal of Information and Library Science (CJILS): A Bibliometric Analysis from 1993-2021","year":2024,"lang":"en","type":"article","venue":"Figshare","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Citation analysis; Information science; Information system; National library","score_opus":0.5917176489081988,"score_gpt":0.541196785067119,"score_spread":0.05052086384107979,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6977353442","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6834046,0.03746009,0.00015464007,0.0014116493,0.0007065255,0.0004574746,0.25057843,0.000023880639,0.025802674],"genre_scores_gemma":[0.9977025,0.00026424747,0.00045836478,0.00002274972,0.00008176222,0.000004850224,0.0011778242,0.0000055999626,0.00028213288],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9881393,0.00017983891,0.00089741655,0.00036502312,0.009863662,0.00055478234],"domain_scores_gemma":[0.9828112,0.007467972,0.000305874,0.0006008837,0.0076617794,0.0011522498],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.013790701,0.00009768192,0.00031598608,0.93580306,0.00029090667,0.010908549,0.0029066456,0.00010626594,0.07663972],"category_scores_gemma":[0.1006407,0.0000680476,0.00017253384,0.97319645,0.00019359583,0.0059669223,0.00077358837,0.00045722054,0.0007658435],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000007798048,0.000018361145,0.036808267,0.00004156295,0.000089133035,0.000049054408,0.00054533355,0.000060953833,0.00006758655,0.00013405514,0.5265614,0.43561652],"study_design_scores_gemma":[0.00013120494,0.00013698277,0.62932056,0.0002359129,0.000021002053,0.000007097106,0.0006460043,0.011874629,0.0011540574,0.0014767313,0.35487446,0.00012133652],"about_ca_topic_score_codex":0.0066128266,"about_ca_topic_score_gemma":0.00068669923,"teacher_disagreement_score":0.5925123,"about_ca_system_score_codex":0.00009692041,"about_ca_system_score_gemma":0.0047498657,"threshold_uncertainty_score":0.99966604},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W6977386344","doi":"10.6084/m9.figshare.1378926","title":"Proposal for Alfred P. Sloan Grant #2014-3-25 \"to support greater understanding of social media in scholarly communication and the actual meaning of various altmetrics\"","year":2015,"lang":"en","type":"article","venue":"Figshare","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Meaning (existential); Foundation (evidence); Scholarly communication; Period (music); Social media; Identity (music); Information center","score_opus":0.23441171087190804,"score_gpt":0.2818770805864573,"score_spread":0.04746536971454923,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6977386344","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.21466218,0.001566061,0.000041991825,0.003449716,0.00080189225,0.00640367,0.38519523,0.000226585,0.38765267],"genre_scores_gemma":[0.99389297,0.0000011301278,0.000063072584,0.000079694466,0.00008325652,0.00006873991,0.0055090617,0.000017068456,0.0002850064],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.99909556,0.00007117931,0.00029714985,0.0001181103,0.00024991942,0.00016808801],"domain_scores_gemma":[0.9989456,0.00038106632,0.00015711479,0.00015190631,0.00031134448,0.00005297439],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00045713253,0.0001071612,0.0002377474,0.00018164935,0.00017515461,0.00052578247,0.00027472884,0.000048333277,0.0033583548],"category_scores_gemma":[0.0011705728,0.00007755988,0.000061016835,0.000059802504,0.000097205055,0.00079565594,0.00013159546,0.00014433735,0.000014220362],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031605063,0.00005344065,0.000078172096,0.00017264487,0.000045370318,0.0000018670944,0.39102653,0.0000015491727,0.0000035503538,0.5276801,0.07979507,0.0008256335],"study_design_scores_gemma":[0.0185668,0.0019634853,0.003280536,0.0024873654,0.00021218871,0.000015799467,0.3604457,0.0002958684,0.0004388104,0.28189138,0.32904762,0.0013544367],"about_ca_topic_score_codex":0.00008512248,"about_ca_topic_score_gemma":0.0034536887,"teacher_disagreement_score":0.7792308,"about_ca_system_score_codex":0.000069332214,"about_ca_system_score_gemma":0.00010330101,"threshold_uncertainty_score":0.9975527},"labels":[{"model":"gemma","categories":["scholarly_communication","metaresearch"],"domain":"evaluation","study_design":"not_applicable","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["scholarly_communication","bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6977675007","doi":"10.6084/m9.figshare.14716728.v1","title":"Supplementary Material for: A Scientometric Analysis of the 100 Most Cited Articles on Pallidotomy","year":2021,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Pallidotomy; Neurosurgery; Citation; Web of science; Bibliometrics; Citation analysis","score_opus":0.07892211001078472,"score_gpt":0.33869877026274453,"score_spread":0.2597766602519598,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6977675007","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0013477681,0.000053878735,1.480782e-8,0.000023541099,0.00014231018,0.0010065164,0.99739623,0.000023806431,0.000005916109],"genre_scores_gemma":[0.00068643194,0.0000018350822,0.000017376951,0.00020779937,0.0002012169,0.00067096035,0.99811476,0.00006379533,0.00003581909],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9966906,0.00019238515,0.00060679787,0.0007139337,0.0012663872,0.0005299406],"domain_scores_gemma":[0.9967347,0.00028687328,0.0007024749,0.0016958113,0.00045909747,0.00012108909],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0002098388,0.00040980047,0.00073060184,0.0052956557,0.000190883,0.0002679587,0.0013804266,0.00021639914,0.7105743],"category_scores_gemma":[0.0025299066,0.00031466482,0.0006030986,0.02981154,0.00002444218,0.00006616743,0.0007821715,0.00021763773,0.0011843009],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005948179,0.0001999393,0.000039581264,0.00038543463,0.0014903457,0.000011594187,0.000010557606,0.00008096114,0.00018895483,2.2399271e-7,0.99750036,0.0000325946],"study_design_scores_gemma":[0.00050091726,0.00006198596,0.0018996663,0.0011049781,0.0023970613,9.85719e-7,0.000020596415,0.00016873794,0.0051508537,0.0000017877364,0.98836523,0.0003271755],"about_ca_topic_score_codex":0.00028317748,"about_ca_topic_score_gemma":0.0011687449,"teacher_disagreement_score":0.70939,"about_ca_system_score_codex":0.00026854663,"about_ca_system_score_gemma":0.00022274865,"threshold_uncertainty_score":0.99993056},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6979867845","doi":"","title":"Análisis bibliométrico sobre el alumnado con altas capacidades (2010- 2018)","year":2022,"lang":"es","type":"article","venue":"Repositorio Institucional de la Universidad de Alicante (Universidad de Alicante)","topic":"Child and Adolescent Psychosocial and Emotional Development","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Context (archaeology); Meaning (existential); Quarter (Canadian coin)","score_opus":0.013003551709487629,"score_gpt":0.2743781812275686,"score_spread":0.26137462951808094,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6979867845","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.77457094,0.0026106387,0.0013823957,0.01664296,0.00781422,0.0011670239,0.0020629284,0.0005295168,0.19321938],"genre_scores_gemma":[0.97011167,0.0035614457,0.0006363794,0.003686548,0.002182448,0.000033255837,0.0003775375,0.00016022054,0.019250479],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9915299,0.0018968229,0.0009256305,0.0018479448,0.0016676596,0.0021320435],"domain_scores_gemma":[0.9956176,0.00065534073,0.00089322263,0.0010770009,0.00037840597,0.0013784061],"candidate_categories":["metaepi_narrow","sts","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0014677754,0.0010894862,0.00111549,0.003675153,0.0032150242,0.0005792216,0.0019816658,0.000893491,0.0034466623],"category_scores_gemma":[0.00009306545,0.001376093,0.0010113386,0.00572462,0.00094991695,0.00068847334,0.00091747515,0.002288253,0.00026619423],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0033385784,0.0037513191,0.08895204,0.0003730628,0.0035213365,0.009630025,0.011935138,0.0010444423,0.025474947,0.394781,0.45112297,0.0060751433],"study_design_scores_gemma":[0.0044015306,0.00041569956,0.17571713,0.0005227683,0.00065435487,0.0031727974,0.009925481,0.0003341627,0.0004453081,0.0013562028,0.80150026,0.0015543167],"about_ca_topic_score_codex":0.0086486805,"about_ca_topic_score_gemma":0.00012207398,"teacher_disagreement_score":0.3934248,"about_ca_system_score_codex":0.004895367,"about_ca_system_score_gemma":0.0033659807,"threshold_uncertainty_score":0.9989247},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6980225120","doi":"","title":"Bibliometric analysis of scientific publications in the field of “inclusion in sports non-formal education” in the Web of Science Core Collection database","year":2024,"lang":"en","type":"article","venue":"The Scientific Issues of Ternopil Volodymyr Hnatiuk National Pedagogical University Series pedagogy","topic":"Psycholinguistics and Behavioral Studies","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Terminology; Field (mathematics); Bibliometrics; The Internet; Webometrics; Subject (documents); Ukrainian; Relevance (law); Data collection","score_opus":0.1002255589510121,"score_gpt":0.44662807961129336,"score_spread":0.34640252066028127,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6980225120","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98474497,0.00030617596,0.000018099663,0.0040771547,0.0005579293,0.00034852425,0.0001288389,0.0000070572423,0.009811224],"genre_scores_gemma":[0.99665534,0.00019760395,0.0001027561,0.000023791268,0.000030634117,0.000003605139,0.00005402294,0.0000021743238,0.0029300484],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99648553,0.00017853895,0.00048604148,0.00035283223,0.0022490693,0.00024799968],"domain_scores_gemma":[0.9972792,0.0007296606,0.00027676896,0.0003459958,0.001334822,0.0000335013],"candidate_categories":["bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.008176871,0.00010302941,0.00025588836,0.024421042,0.0010819338,0.00013860321,0.0016776255,0.00006497961,0.000121455254],"category_scores_gemma":[0.0014471756,0.000066183144,0.00011720617,0.123638906,0.0037758572,0.0006293519,0.00056414667,0.00019597681,8.6222593e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024000237,0.0030515061,0.33438984,0.00020599275,0.000090872236,0.0000124432845,0.21619816,0.0005557919,0.0049747955,0.42085645,0.0068472084,0.012576951],"study_design_scores_gemma":[0.0002740249,0.0001298196,0.8422743,0.00012803345,0.00016223769,0.0000016296206,0.10274978,0.0018796349,0.00030861836,0.0041959016,0.047727752,0.00016824194],"about_ca_topic_score_codex":0.005865531,"about_ca_topic_score_gemma":0.02134734,"teacher_disagreement_score":0.5078845,"about_ca_system_score_codex":0.00014559369,"about_ca_system_score_gemma":0.0022020424,"threshold_uncertainty_score":0.9989353},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6980234258","doi":"","title":"A BIBLIOMETRIC REVIEW OF RESEARCH PUBLICATIONS ON DIGITAL TWINPREDICTIVE MAINTENANCE SYSTEMS IN THE MARITIME INDUSTRY","year":2024,"lang":"en","type":"article","venue":"Memorial University Research Repository (Memorial University)","topic":"Digital Transformation in Industry","field":"Engineering","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Mitacs; King Fahd University of Petroleum and Minerals","keywords":"Nucleofection; Gestational period; TSG101; Diafiltration; Hyporeflexia; Proteogenomics; Hemopericardium; Articular cartilage damage","score_opus":0.05469788302027644,"score_gpt":0.2940216914128797,"score_spread":0.23932380839260325,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6980234258","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.043851834,0.0016914228,0.00023195139,0.0007917248,0.0055782227,0.001976906,0.0004579787,0.0005446275,0.94487536],"genre_scores_gemma":[0.9815423,0.0023045116,0.00001092502,0.0000075999014,0.0016199792,0.000005332245,0.000055807548,0.00003855707,0.014414988],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9953294,0.00089088024,0.0004029069,0.0004952763,0.0021348945,0.00074664364],"domain_scores_gemma":[0.99592716,0.00198409,0.000050487262,0.00071655156,0.00107763,0.00024409618],"candidate_categories":["bibliometrics","research_integrity"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0029842663,0.00022837837,0.00035260164,0.028971527,0.0004277347,0.0006153408,0.0017774956,0.0004897246,0.00004276932],"category_scores_gemma":[0.00062740623,0.0002242896,0.00017561701,0.086817674,0.00069151627,0.0022735845,0.0002506915,0.0032772352,0.0000985348],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011850008,0.0009603029,0.0014407381,0.01583446,0.00073628227,0.0056524305,0.0022991518,0.004437944,0.0005801589,0.2971486,0.66285545,0.0068694944],"study_design_scores_gemma":[0.0013406625,0.0004642109,0.0009318678,0.009127264,0.000053057272,0.00009217029,0.011357626,0.0018718297,0.0002983081,0.00022944025,0.97380364,0.00042990845],"about_ca_topic_score_codex":0.0003094779,"about_ca_topic_score_gemma":0.000009024201,"teacher_disagreement_score":0.93769044,"about_ca_system_score_codex":0.0015713287,"about_ca_system_score_gemma":0.0007025611,"threshold_uncertainty_score":0.99902225},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6986346476","doi":"","title":"Parkinson&amp;amp;rsquo;s Disease Related Depression and Anxiety: A 22-Year Bibliometric Analysis (2000-2022)","year":2023,"lang":"en","type":"review","venue":"Dove Medical Press (Taylor and Francis Group)","topic":"Parkinson's Disease Mechanisms and Treatments","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"China Academy of Chinese Medical Sciences","keywords":"Bibliometrics; Depression (economics); China; Anxiety; Disease; Subject (documents); Web of science","score_opus":0.04370141314666505,"score_gpt":0.332957388257843,"score_spread":0.28925597511117795,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6986346476","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00059366255,0.9898975,0.005722745,0.00005336014,0.0005023061,0.0015575817,0.00096996705,0.00031808644,0.0003848155],"genre_scores_gemma":[0.023497034,0.97139335,0.00020436224,0.00019486227,0.0003802349,0.0006373366,0.0032631424,0.0002097248,0.0002199758],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9929849,0.00042391167,0.0013576321,0.0020645815,0.0021758408,0.0009931382],"domain_scores_gemma":[0.99399406,0.0006176409,0.00052819843,0.0012929336,0.0001354451,0.0034317283],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0010719622,0.001136839,0.0037626754,0.0180729,0.00029959006,0.00018192995,0.0004124853,0.0011337891,0.0017003583],"category_scores_gemma":[0.0015894348,0.00081109797,0.0013347707,0.041079815,0.00031955718,0.00017988395,0.0005721496,0.0012256831,0.0002802193],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022506634,0.00059730944,0.00787118,0.009115933,0.004845047,0.0007627422,0.00005253235,3.133833e-7,5.286993e-7,0.000052250234,0.00024236794,0.97623473],"study_design_scores_gemma":[0.0036771381,0.00011078137,0.013561618,0.0010235009,0.036340587,0.00011119672,0.000009769873,0.003468344,3.391731e-7,0.00031223675,0.94043845,0.00094605406],"about_ca_topic_score_codex":0.00025001838,"about_ca_topic_score_gemma":0.00003204516,"teacher_disagreement_score":0.9752887,"about_ca_system_score_codex":0.00015162725,"about_ca_system_score_gemma":0.00004244055,"threshold_uncertainty_score":0.999434},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6991094829","doi":"","title":"The evaluation of the biorefinery research: A scientometric approach","year":2012,"lang":"en","type":"article","venue":"DSpace (Sirnak University)","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Biorefinery; Incentive; Scopus; Science Citation Index; Scientometrics; Citation analysis; Citation; Bibliometrics","score_opus":0.9046261129120148,"score_gpt":0.6202098476861282,"score_spread":0.2844162652258866,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6991094829","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.84691256,0.0017968318,0.0006606195,0.0016530425,0.0011382847,0.00062671385,0.000015367788,0.000014675663,0.14718187],"genre_scores_gemma":[0.98114735,0.00015499795,0.00019906508,0.000013083653,0.00009327569,0.000001532092,7.360107e-7,0.000006498176,0.018383484],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9730926,0.002557035,0.00030615664,0.00046392935,0.022608561,0.00097168196],"domain_scores_gemma":[0.9824051,0.00670364,0.00031588445,0.0015717195,0.008595056,0.000408569],"candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.11402765,0.00011182932,0.00017634868,0.073617436,0.0015029864,0.00078531937,0.004979956,0.000106654246,0.0000967215],"category_scores_gemma":[0.05958241,0.000056435976,0.0002013588,0.5006656,0.0010929324,0.0007951879,0.0020149422,0.00044887277,0.00011579011],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000093517076,0.0009051747,0.52124715,0.000012083893,0.00008300032,0.0000015246459,0.0018370395,0.00042651762,0.0025092624,0.10229474,0.111030184,0.25955978],"study_design_scores_gemma":[0.0007727093,0.000074668715,0.43731797,0.00000811304,0.000034484005,0.000003923009,0.0117533,0.008480994,0.001070349,0.0018705933,0.5384395,0.00017338226],"about_ca_topic_score_codex":0.00013888026,"about_ca_topic_score_gemma":0.000030036555,"teacher_disagreement_score":0.42740935,"about_ca_system_score_codex":0.0004760293,"about_ca_system_score_gemma":0.00057996454,"threshold_uncertainty_score":0.9997969},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W6991566876","doi":"","title":"Holistic analysis of global feminism publications: A bibliometric evaluation of feminism literature between 1975 and 2017","year":2021,"lang":"en","type":"article","venue":"Lincoln (University of Nebraska)","topic":"Sex and Gender in Healthcare","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Feminism; Bibliometrics; Popularity; Scientometrics; Periodical literature; Web of science","score_opus":0.17352152503655008,"score_gpt":0.39595287391977274,"score_spread":0.22243134888322266,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6991566876","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98566514,0.007017684,0.0004675302,0.0012357346,0.00005914552,0.0002775327,0.0006086592,0.000015924286,0.004652648],"genre_scores_gemma":[0.9961698,0.00057938945,0.0022523357,0.0000418316,0.00004250142,6.355154e-7,0.0005162205,0.0000041557387,0.00039312209],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99842656,0.00013843128,0.00025607922,0.0003303876,0.00071128516,0.00013725295],"domain_scores_gemma":[0.99701065,0.00023411424,0.00030569278,0.00048016311,0.0018117288,0.00015765996],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0006877424,0.000100766694,0.00053625193,0.0061510406,0.00008637346,0.000009781273,0.00014225244,0.00020388562,0.00013969949],"category_scores_gemma":[0.00040052782,0.000119426346,0.00017704621,0.042201944,0.00013689032,0.00012944218,0.000101016536,0.00012319842,0.0000015070032],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006356738,0.00026030984,0.92225516,0.00067712576,0.0017079001,0.000019988716,0.0028112647,0.000022398615,0.00029875673,0.0014303639,0.001167102,0.06928606],"study_design_scores_gemma":[0.0011758786,0.00013061226,0.9883836,0.00013661801,0.0035708908,0.000011281594,0.00396358,0.0011349412,0.00007585064,0.0006607646,0.00067523186,0.00008076947],"about_ca_topic_score_codex":0.00022304719,"about_ca_topic_score_gemma":0.00011771219,"teacher_disagreement_score":0.06920529,"about_ca_system_score_codex":0.000077290846,"about_ca_system_score_gemma":0.00054137415,"threshold_uncertainty_score":0.9781565},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6998711718","doi":"","title":"A bibliometric analysis of Serials Review from 1991 to 2020","year":2021,"lang":"en","type":"article","venue":"San José State University ScholarWorks (San Jose State University)","topic":"Library Collection Development and Digital Resources","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliographic coupling; Scopus; Bibliometrics; Citation; Publishing; Citation analysis; Bibliographic database; Web of science","score_opus":0.013982168243384496,"score_gpt":0.20670499027151892,"score_spread":0.19272282202813443,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6998711718","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8984603,0.002740641,0.065220095,0.0014144406,0.00047459354,0.0006458738,0.0008656901,0.0004417138,0.029736595],"genre_scores_gemma":[0.6715793,0.05141706,0.080694616,0.00433254,0.00011762167,0.0000011688284,0.0014502142,0.00011606624,0.19029142],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9968178,0.00053030805,0.00041402777,0.0010019575,0.0006983414,0.0005375334],"domain_scores_gemma":[0.99713993,0.0003372359,0.00035077642,0.00089418143,0.0006105894,0.0006672995],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00036212016,0.0003442236,0.00082609284,0.0415011,0.0004090625,0.0004157265,0.0017412018,0.00009810836,0.0008076823],"category_scores_gemma":[0.00020937459,0.00041732652,0.0004532357,0.29717612,0.00009060822,0.003941233,0.0014778258,0.0003333693,0.00008072078],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.003605042,0.0027203516,0.15680653,0.0014333805,0.028424835,0.028394487,0.02151718,0.016494615,0.0046917247,0.011382423,0.25682163,0.4677078],"study_design_scores_gemma":[0.0018715931,0.00020786762,0.1115127,0.00067701057,0.0009871688,0.0000064238825,0.001398459,0.0007461744,0.0016945963,0.000396856,0.87927264,0.0012284961],"about_ca_topic_score_codex":0.00023078994,"about_ca_topic_score_gemma":0.0003459987,"teacher_disagreement_score":0.622451,"about_ca_system_score_codex":0.0002540679,"about_ca_system_score_gemma":0.0004893405,"threshold_uncertainty_score":0.99982786},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6998819759","doi":"","title":"Bibliography of Canadian Bibliographies","year":2023,"lang":"en","type":"article","venue":"Project Muse (Johns Hopkins University)","topic":"Canadian Identity and History","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliography; Subject (documents); Government (linguistics)","score_opus":0.024106016532165972,"score_gpt":0.22352086327318468,"score_spread":0.19941484674101873,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6998819759","genre_codex":"other","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013662512,0.0000043836553,0.0000026778753,0.0005728605,0.00030045983,0.000214431,0.00012307105,0.0001918073,0.9849278],"genre_scores_gemma":[0.027479708,0.97221106,0.00004300157,0.000053381085,0.00006261213,0.0000010162782,0.000013020733,0.000011086436,0.00012510864],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.998808,0.000118009615,0.00010616783,0.00021775639,0.00028687876,0.0004631575],"domain_scores_gemma":[0.99922955,0.000059193597,0.000067797606,0.0002084489,0.00014680823,0.00028817673],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00028696223,0.000090181915,0.0001499779,0.989179,0.0009754576,0.00014804745,0.00044251143,0.00012298416,0.00021108476],"category_scores_gemma":[0.00003944739,0.00013428758,0.00020420604,0.92664254,0.00074627035,0.00044106392,0.000043849945,0.000114826355,0.0000882823],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000100592815,0.000020348827,0.004761028,0.000022145136,0.000052574185,0.00012812682,0.005223103,0.0000027826584,0.0000026899397,0.006865435,0.98170847,0.0012032123],"study_design_scores_gemma":[0.00016609077,0.00002319567,0.0032774392,0.000016675427,0.000030956977,3.7263885e-7,0.0040058573,0.0000019173806,0.000005522654,0.0000049832197,0.9923179,0.00014905854],"about_ca_topic_score_codex":0.9998792,"about_ca_topic_score_gemma":0.9999365,"teacher_disagreement_score":0.98480266,"about_ca_system_score_codex":0.000042902713,"about_ca_system_score_gemma":0.0010088983,"threshold_uncertainty_score":0.75025296},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"agree"},{"id":"W6998977819","doi":"","title":"A Bibliometric analysis of single mothers","year":2023,"lang":"en","type":"article","venue":"HRB National Drugs Library (Health Research Board)","topic":"Family Dynamics and Relationships","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Islamic Azad University","keywords":"Single mothers; Bibliometrics; Ranking (information retrieval); Web of science; Single parent; Research methodology; Single-subject design; Single point","score_opus":0.2183734956713305,"score_gpt":0.4509712243071274,"score_spread":0.2325977286357969,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6998977819","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.57551795,0.0014244857,0.00011164428,0.06788054,0.00033730618,0.0008971022,0.000416576,0.00040391233,0.35301045],"genre_scores_gemma":[0.98059505,0.0021097155,0.00051296706,0.00040900212,0.0001524398,0.000032944907,0.00024644015,0.000022809114,0.015918635],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99421746,0.0012152643,0.00041283012,0.00029045073,0.0032118247,0.00065214955],"domain_scores_gemma":[0.9957321,0.003145667,0.00014061874,0.0001733975,0.0004649153,0.0003432981],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0062580197,0.00008141653,0.00024936156,0.13516751,0.0009060047,0.00015609253,0.0004723116,0.00012366807,0.00048723296],"category_scores_gemma":[0.0014747914,0.00007954382,0.00016909123,0.4307673,0.00045623345,0.0006396513,0.00013566168,0.00033812795,0.00016364397],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004865575,0.00038354297,0.07837035,0.00012240434,0.0006067579,0.000006463152,0.013661602,0.0049821716,0.000051326802,0.660081,0.21937974,0.02230597],"study_design_scores_gemma":[0.00030313316,0.000102963,0.77281106,0.000040091967,0.000022591561,7.629408e-8,0.007438492,0.02272652,0.00000626537,0.03330621,0.1630781,0.00016447566],"about_ca_topic_score_codex":0.0016657748,"about_ca_topic_score_gemma":0.0002041939,"teacher_disagreement_score":0.6944407,"about_ca_system_score_codex":0.00022544082,"about_ca_system_score_gemma":0.0021297927,"threshold_uncertainty_score":0.8746346},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6999399753","doi":"","title":"Covid-19 and Emergency Medicine: A Scientometric Assessment of Global Publications","year":2021,"lang":"en","type":"article","venue":"Lincoln (University of Nebraska)","topic":"Emergency Medicine Education and Research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Annals; Citation; Citation impact; Quality (philosophy); Web of science; Citation analysis; Bibliometrics; Subject (documents)","score_opus":0.09809929389114645,"score_gpt":0.44845931793323984,"score_spread":0.3503600240420934,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6999399753","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7240801,0.0030463375,0.007672297,0.16708824,0.0009259198,0.0005717053,0.000057728288,0.000060290546,0.096497394],"genre_scores_gemma":[0.9733303,0.0018697914,0.003305649,0.00019565302,0.00010618218,8.9781037e-7,0.00006836251,0.0000054832203,0.021117669],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99860954,0.00007809541,0.00022712095,0.00027482776,0.00064624124,0.00016417539],"domain_scores_gemma":[0.99795544,0.00009242261,0.00014840989,0.00031939012,0.00081870845,0.0006656077],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0004999424,0.000079524136,0.0002778704,0.0005742362,0.00015446148,0.0000017422673,0.00015734017,0.000071399925,0.011963015],"category_scores_gemma":[0.0016243521,0.00008022553,0.000073949,0.005333894,0.0003778461,0.000086754866,0.00012636463,0.00012971504,0.00000500246],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000028440603,0.0006780452,0.86015946,0.0007280672,0.00012800682,0.000017996686,0.0010893206,0.000003805968,0.0026972638,0.019272106,0.11293825,0.0022592603],"study_design_scores_gemma":[0.00158467,0.00024116343,0.9193629,0.000054682154,0.00012077278,0.000029725654,0.010128526,0.00039403202,0.000016284472,0.00031940974,0.06767495,0.00007285285],"about_ca_topic_score_codex":0.00073028245,"about_ca_topic_score_gemma":0.00029239745,"teacher_disagreement_score":0.24925022,"about_ca_system_score_codex":0.00019228861,"about_ca_system_score_gemma":0.0027535732,"threshold_uncertainty_score":0.9889402},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7005565218","doi":"","title":"Research Productivity and Publication Trends of ‘Law Library Journal’ (1989-2021): A Scientometric Study","year":2024,"lang":"en","type":"article","venue":"Lincoln (University of Nebraska)","topic":"Blood disorders and treatments","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scientometrics; Productivity; Law library; Bibliometrics; Period (music); Scientific literature; Distribution (mathematics); Trend analysis","score_opus":0.028168298922233326,"score_gpt":0.30022926760786534,"score_spread":0.272060968685632,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7005565218","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98960155,0.002551573,0.000038242262,0.0009862449,0.00006579411,0.00015100582,0.000023759647,0.000007134044,0.0065747052],"genre_scores_gemma":[0.9941262,0.00017507037,0.00030146766,0.0000056550944,0.000053655353,5.777943e-7,0.000029120973,0.000007672335,0.0053005577],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9991147,0.00011437584,0.00008663485,0.00030045462,0.00024944887,0.00013438718],"domain_scores_gemma":[0.99947655,0.000021676798,0.00005372998,0.00021675641,0.00015383327,0.00007744677],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00041551463,0.000068199784,0.00011684609,0.00085993094,0.0001688949,0.000058254736,0.00019187588,0.000066281566,0.000068405454],"category_scores_gemma":[0.000030058853,0.00007077826,0.000057552064,0.0018637198,0.00020338365,0.00004989693,0.00025205337,0.00011376641,0.0000030353394],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000580625,0.0069088126,0.17495158,0.00032000526,0.0014333894,0.00012628996,0.0039668977,0.00001475503,0.031622216,0.0048797135,0.028927645,0.7462681],"study_design_scores_gemma":[0.007991437,0.010899417,0.6505179,0.0001973526,0.00055156427,0.00011730274,0.018924829,0.00042468685,0.02952129,0.0030119203,0.27702656,0.00081572373],"about_ca_topic_score_codex":0.000115853516,"about_ca_topic_score_gemma":0.00009267326,"teacher_disagreement_score":0.74545234,"about_ca_system_score_codex":0.000010069449,"about_ca_system_score_gemma":0.00012449027,"threshold_uncertainty_score":0.28862536},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7005988486","doi":"","title":"The state and evolution of Gold Open Access: A country level analysis","year":2018,"lang":"en","type":"other","venue":"Leiden Repository (Leiden University)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Web of science; Quarter (Canadian coin); China; Bibliometrics; State (computer science); Filter (signal processing)","score_opus":0.02455380905811092,"score_gpt":0.26295163211002615,"score_spread":0.23839782305191523,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7005988486","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012674059,0.0014072667,0.00054208725,0.000028707189,0.0005100305,0.0013972467,0.0006176625,0.00023669201,0.98258626],"genre_scores_gemma":[0.019663576,0.00019873933,0.0002176951,0.000009867342,0.00022317028,0.000004407223,0.000046395548,0.00043273158,0.9792034],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9963523,0.00064717507,0.0005410114,0.001079876,0.00083007413,0.0005495513],"domain_scores_gemma":[0.9953426,0.0002263706,0.0019348967,0.0017123906,0.0005423261,0.00024138782],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.000622221,0.00059122057,0.001030654,0.0034138807,0.0004954716,0.0010039238,0.005198328,0.0004600998,0.00028513063],"category_scores_gemma":[0.00009981288,0.00054161204,0.00029155496,0.0054192236,0.0011347247,0.0012283137,0.0037277641,0.00034520935,0.0003795829],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006960741,0.00014968963,0.25067443,0.00021858339,0.011438475,0.00042940964,0.00017112501,0.00007135873,0.0012936784,0.0033314694,0.73133224,0.00019345786],"study_design_scores_gemma":[0.0013265201,0.00012264616,0.12547901,0.0005077634,0.0068100933,0.000034604614,0.00043757437,0.00018621967,0.0003282493,0.0002177977,0.86354065,0.0010088986],"about_ca_topic_score_codex":0.05033275,"about_ca_topic_score_gemma":0.034716547,"teacher_disagreement_score":0.13220838,"about_ca_system_score_codex":0.0009899228,"about_ca_system_score_gemma":0.0007430059,"threshold_uncertainty_score":0.9997035},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7007717074","doi":"","title":"Analyse bibliométrique des revues Canadian Journal of Communication et Communication 1974-2005","year":2010,"lang":"fr","type":"other","venue":"Open MIND","topic":"Biomedical and Chemical Research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scholarly communication; Scientific communication; Context (archaeology)","score_opus":0.08520931918472802,"score_gpt":0.40842328934586986,"score_spread":0.32321397016114184,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7007717074","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.054214843,0.050006893,0.00009996473,0.029306771,0.00024794214,0.00094425806,0.00017719221,0.000001750278,0.86500037],"genre_scores_gemma":[0.22191012,0.044782214,0.06017545,0.00061691005,0.00089630665,0.00002959482,0.0008846605,0.00019665921,0.6705081],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99739224,0.0006220774,0.00076755055,0.00023320007,0.0004500023,0.00053492124],"domain_scores_gemma":[0.99558115,0.0005552483,0.0005387117,0.0013080365,0.0006089636,0.0014078877],"candidate_categories":["research_integrity","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00338649,0.00024980458,0.0006678025,0.0019190728,0.00021215825,0.00034080024,0.0025223335,0.0009485231,0.16471727],"category_scores_gemma":[0.0019078868,0.00020471877,0.00016089466,0.0017951238,0.0016571603,0.00022290433,0.00039811365,0.0025284507,0.00088715093],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013313834,0.00046869522,0.005811458,0.0001972389,0.00052057876,0.00007530009,0.002228189,7.0468707e-7,0.017533153,0.00030217302,0.08003978,0.8926896],"study_design_scores_gemma":[0.0009020557,0.00013060261,0.0032938153,0.003578821,0.00023707043,0.00010248171,0.00048713235,0.00004888617,0.0051918197,0.0005817956,0.9852439,0.00020161469],"about_ca_topic_score_codex":0.08667588,"about_ca_topic_score_gemma":0.5936864,"teacher_disagreement_score":0.9052041,"about_ca_system_score_codex":0.00038384463,"about_ca_system_score_gemma":0.003945287,"threshold_uncertainty_score":0.99989074},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W7008682128","doi":"","title":"Collaboration to stimulate innovation in the space sector and encourage cross-fertilization of Earth-Space RD: A study using bibliometrics and surveys of the Canadian space sector","year":2017,"lang":"en","type":"other","venue":"PolyPublie (École Polytechnique de Montréal)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Space (punctuation); Bibliometrics; Public sector; Work (physics)","score_opus":0.03148225110057918,"score_gpt":0.3009328390881242,"score_spread":0.269450587987545,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7008682128","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9722175,0.0012433934,0.013796596,0.00086625374,0.00027643424,0.007903802,0.0017685296,0.00017355282,0.0017539513],"genre_scores_gemma":[0.9898115,0.00008038204,0.0035470044,0.00011548027,0.000094625204,0.00018716388,0.000059372684,0.00053213496,0.0055723605],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9949726,0.0015147385,0.00086681225,0.0007711708,0.0011565274,0.0007181737],"domain_scores_gemma":[0.99476784,0.0002879007,0.0019590452,0.0017769587,0.00096977124,0.00023846824],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0058907554,0.00066791003,0.0008725148,0.018536257,0.00047074008,0.00086659286,0.000877264,0.0007425508,0.00003307191],"category_scores_gemma":[0.0031720723,0.0005459572,0.000069470036,0.027601676,0.00033037853,0.00048413078,0.00031318548,0.00061691075,0.0000028347413],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001959695,0.00043000048,0.94259006,0.00033947293,0.00018630215,0.000030554467,0.0048083724,0.0064408267,0.031516448,0.0013838531,0.011422388,0.00065575313],"study_design_scores_gemma":[0.0015681469,0.00035196714,0.96944064,0.00057752174,0.00013535797,0.000021300719,0.0007057375,0.017459484,0.0043126345,0.00008684936,0.0045946613,0.00074572524],"about_ca_topic_score_codex":0.75114816,"about_ca_topic_score_gemma":0.970876,"teacher_disagreement_score":0.21972783,"about_ca_system_score_codex":0.00072592933,"about_ca_system_score_gemma":0.0017127977,"threshold_uncertainty_score":0.9996992},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W7024578830","doi":"","title":"SCIENTOMETRIC ANALYSIS OF RESEARCH FROM A FEMINIST PERSPECTIVE","year":2013,"lang":"en","type":"article","venue":"Redalyc (Universidad Autónoma del Estado de México)","topic":"Face recognition and analysis","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Perspective (graphical); Subject (documents); Scopus; Latin Americans; Variety (cybernetics); Sample (material); Citation; Bibliometrics","score_opus":0.03727592367206298,"score_gpt":0.3065725764491382,"score_spread":0.26929665277707526,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7024578830","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8432962,0.00037235138,0.12660429,0.0017987683,0.00009208358,0.00026818487,0.000060748596,0.00019756469,0.027309822],"genre_scores_gemma":[0.9795635,0.000053971577,0.018459877,0.00012115101,0.000027255815,0.000010425244,0.00001838762,0.000012647089,0.0017327898],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9967428,0.00034889797,0.0003228986,0.0008420581,0.0010722468,0.00067108555],"domain_scores_gemma":[0.9965684,0.0006099762,0.00021158195,0.0009963273,0.0012183102,0.00039541145],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00086946815,0.00020923816,0.00052183674,0.010576654,0.00037233523,0.00039809674,0.0016223022,0.00016089858,0.00096765894],"category_scores_gemma":[0.00030187832,0.00022080353,0.0004923105,0.03853352,0.00038764125,0.00086737325,0.00053860067,0.0003610188,0.00036118407],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011831939,0.0034840822,0.19199218,0.00013523981,0.015778696,0.00078044325,0.04751984,0.0043393625,0.047385212,0.48887363,0.011241,0.18835199],"study_design_scores_gemma":[0.00085698126,0.00019382083,0.2134358,0.000056827987,0.0007927565,0.000009156617,0.024613941,0.7475048,0.0021857375,0.009037719,0.00074343994,0.0005690522],"about_ca_topic_score_codex":0.008413858,"about_ca_topic_score_gemma":0.00017546689,"teacher_disagreement_score":0.74316543,"about_ca_system_score_codex":0.00066290615,"about_ca_system_score_gemma":0.00034473452,"threshold_uncertainty_score":0.9999456},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7027041239","doi":"","title":"A Bibliometric Analysis of Research Trends in Spousal Support for Breast Cancer Patients","year":2025,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Cancer survivorship and care","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Breast cancer; Bibliometrics; Social support; Cancer; Web of science; Incidence (geometry); MEDLINE","score_opus":0.2890505722771048,"score_gpt":0.6271608764433242,"score_spread":0.3381103041662194,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7027041239","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9847168,0.007314007,0.000041425996,0.00036018982,0.00051293545,0.0005264533,0.00084956264,0.000012549177,0.0056661014],"genre_scores_gemma":[0.9923869,0.005018377,0.000019364557,0.00020202617,0.00008151963,0.00014641663,0.000095221745,0.000028794688,0.0020213798],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9965418,0.00021266668,0.0010207244,0.00047029063,0.0012561312,0.00049838744],"domain_scores_gemma":[0.99648273,0.0005239353,0.0004394425,0.00049033447,0.0018325114,0.00023103089],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0024967808,0.00021820857,0.0012763015,0.21746248,0.00010049099,0.00025646322,0.0012385856,0.00013853614,0.014694175],"category_scores_gemma":[0.00034772584,0.000193088,0.00044753286,0.28510725,0.00014357353,0.0006419482,0.0004900514,0.00042291972,0.0000012505219],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013428209,0.0005990193,0.8285556,0.00025549694,0.0010978198,0.000011159906,0.00015497614,0.00003276028,0.0014116137,0.00001609794,0.024828158,0.14169449],"study_design_scores_gemma":[0.0016407433,0.000039746927,0.99031585,0.00049189647,0.0008650702,0.0000012492187,0.000121099474,0.000089678826,0.0019590526,0.00014948909,0.0041655744,0.00016055416],"about_ca_topic_score_codex":0.016645858,"about_ca_topic_score_gemma":0.005991349,"teacher_disagreement_score":0.16176026,"about_ca_system_score_codex":0.00035897427,"about_ca_system_score_gemma":0.00050891255,"threshold_uncertainty_score":0.9899024},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7036118677","doi":"","title":"Bibliometric analysis using R on corporate governance mechanisms / Hamizah Hassan","year":2023,"lang":"en","type":"article","venue":"UiTM Institutional Repositories (Universiti Teknologi MARA)","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Corporate governance; Bibliometrics; Scopus; Quarter (Canadian coin); Information governance; Citation analysis","score_opus":0.1820170316637471,"score_gpt":0.3451119945746251,"score_spread":0.163094962910878,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7036118677","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.80191094,0.00006121575,0.1713948,0.0008366882,0.0063006203,0.00023679051,0.0002480679,0.00053978537,0.018471092],"genre_scores_gemma":[0.98498696,0.00002217426,0.0038904725,0.000108651206,0.00014535192,0.0000026972536,0.000109490495,0.000008215832,0.010725976],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99457186,0.00019128998,0.0005165912,0.0012770185,0.0029344603,0.0005087913],"domain_scores_gemma":[0.9963531,0.0008655175,0.0006894408,0.0013046892,0.00061462796,0.00017263042],"candidate_categories":["bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0027986157,0.00024894296,0.0004106633,0.049018234,0.0015527243,0.00088843453,0.0015924103,0.00014007623,0.00018541112],"category_scores_gemma":[0.002575712,0.0002239434,0.00029984865,0.2836594,0.00043543114,0.0009269438,0.0010972468,0.00021128959,0.0007521865],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000138779,0.0001523749,0.01806606,0.000009438747,0.0005920657,0.0017617089,0.00014743193,0.25676793,0.0009974136,0.6642237,0.03545763,0.021685526],"study_design_scores_gemma":[0.0009969362,0.0003185769,0.46188426,0.000069270835,0.0005531698,0.00008510398,0.0026920934,0.41271123,0.00096044043,0.057367772,0.061323892,0.001037256],"about_ca_topic_score_codex":0.00016137719,"about_ca_topic_score_gemma":0.000031368487,"teacher_disagreement_score":0.60685587,"about_ca_system_score_codex":0.00045164704,"about_ca_system_score_gemma":0.00024165996,"threshold_uncertainty_score":0.9997471},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W7037816534","doi":"","title":"Ética de la investigación: una visión bibliométrica desde WOS (1976-2023) (1976-2023)","year":2024,"lang":"es","type":"article","venue":"Dialnet (Universidad de la Rioja)","topic":"Diatoms and Algae Research","field":"Materials Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Context (archaeology); Medical practice; Ethics committee; Web of science","score_opus":0.016648551205979457,"score_gpt":0.31448318344798326,"score_spread":0.2978346322420038,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7037816534","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.88268733,0.0065734223,0.00089917437,0.0034023523,0.001181282,0.0007098814,0.00070154975,0.00064512476,0.10319989],"genre_scores_gemma":[0.9826198,0.0048926244,0.0023688881,0.0005067509,0.0009408191,0.00004197818,0.000060867045,0.00018600884,0.008382247],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99226755,0.0020515933,0.00056601973,0.0014918451,0.0014288324,0.0021941585],"domain_scores_gemma":[0.9938198,0.0032960714,0.00020579844,0.0010434105,0.00020643095,0.0014285328],"candidate_categories":["metaepi_narrow","scholarly_communication","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0037946985,0.00076926715,0.000837165,0.005349309,0.0006893434,0.0048827445,0.0018309003,0.0010417791,0.0047414764],"category_scores_gemma":[0.0009886479,0.00081644,0.00048120163,0.011582222,0.0019892824,0.0013396625,0.001056118,0.0014121813,0.0050379136],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004754038,0.0009946461,0.010292319,0.0030808519,0.00070795545,0.013731962,0.00391843,0.00047323602,0.3186472,0.35503286,0.25564343,0.037001703],"study_design_scores_gemma":[0.0025647406,0.0006696974,0.0514615,0.00244763,0.0006177869,0.0004072206,0.0016220923,0.016779875,0.0229653,0.028069662,0.8700108,0.0023836857],"about_ca_topic_score_codex":0.002653491,"about_ca_topic_score_gemma":0.00007063053,"teacher_disagreement_score":0.61436737,"about_ca_system_score_codex":0.00082524173,"about_ca_system_score_gemma":0.0028391336,"threshold_uncertainty_score":0.99942863},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reporting","study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","research_integrity"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7037849563","doi":"","title":"The evaluation of the research on the social sciences in Turkey: A scientometric approach","year":2021,"lang":"en","type":"article","venue":"DSpace (Sirnak University)","topic":"Distributed systems and fault tolerance","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Subject (documents); Nucleofection; Government (linguistics); Context (archaeology); Work (physics)","score_opus":0.2518332468355427,"score_gpt":0.38299869691802574,"score_spread":0.13116545008248304,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7037849563","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.78639054,0.0004549928,0.0056981724,0.023187743,0.0008088638,0.00080835493,0.000011949055,0.00003095074,0.18260841],"genre_scores_gemma":[0.997937,0.000012111489,0.00008999238,0.000037534974,0.000029058085,0.0000017819094,3.918882e-7,0.0000018378737,0.0018902518],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9960874,0.0014330978,0.00010012734,0.00031838857,0.0017646403,0.00029637522],"domain_scores_gemma":[0.99829865,0.0004758732,0.0000933612,0.00046937476,0.00064192835,0.000020804346],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.0063009313,0.00006710875,0.000089857036,0.00043924307,0.0015669283,0.00024475777,0.0020732349,0.000044297347,0.000003059943],"category_scores_gemma":[0.00047971006,0.000036432404,0.00006761933,0.01823961,0.0005424673,0.00020856089,0.0005135305,0.00026298687,0.000005368438],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000008461696,0.00014310022,0.0014516084,0.000008527681,0.000014627304,0.0000042476604,0.0024996833,0.004080697,0.0002831572,0.97926974,0.0059444765,0.0062916493],"study_design_scores_gemma":[0.0023910159,0.00018010568,0.1227863,0.00022476741,0.000037227077,0.0000132586865,0.057323616,0.5812283,0.0027305428,0.0058037546,0.22678256,0.00049854006],"about_ca_topic_score_codex":0.00014362077,"about_ca_topic_score_gemma":0.0001128662,"teacher_disagreement_score":0.97346604,"about_ca_system_score_codex":0.00022118002,"about_ca_system_score_gemma":0.0006673259,"threshold_uncertainty_score":0.9997329},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7044877122","doi":"","title":"Acupuncture for Chronic Obstructive Pulmonary Disease: A 38-Year Bibliometric Landscape of Global Research Trends and Knowledge Evolution (1986&amp;ndash;2024)","year":2025,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Acupuncture Treatment Research Studies","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Geopolitics; Global health; Acupuncture; Integrative medicine; Productivity; Convergence (economics); Global governance; Citation; China","score_opus":0.2430566242400939,"score_gpt":0.6061925454653551,"score_spread":0.36313592122526117,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7044877122","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.58170587,0.39135584,0.00037182934,0.0014124973,0.00060794596,0.0025405202,0.0006719241,0.000055359902,0.021278212],"genre_scores_gemma":[0.9749215,0.017539317,0.00025790537,0.000022949711,0.00031684292,0.00030295431,0.000089742905,0.0000398287,0.0065089464],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99596685,0.00041800973,0.0007837635,0.00079575455,0.0012356881,0.00079993816],"domain_scores_gemma":[0.9950983,0.0014105838,0.0003841452,0.0006898179,0.0019027716,0.0005143841],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0017797,0.00058026175,0.0015340359,0.02206795,0.00047421618,0.00036534932,0.0011813502,0.00021009114,0.0021630106],"category_scores_gemma":[0.0024154396,0.00033981501,0.0004219453,0.048188396,0.00062605314,0.0009228895,0.0017390226,0.00064721616,0.0000072061903],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029221307,0.00069182256,0.8053848,0.00157799,0.0011781107,0.000044398916,0.00007621192,0.00000849232,0.004732983,0.00039886608,0.13467146,0.048312742],"study_design_scores_gemma":[0.0019566251,0.00009837987,0.9680444,0.0009973904,0.00061449694,0.00002015078,0.00016008018,0.000077829805,0.0005532461,0.0077219433,0.01951469,0.00024075223],"about_ca_topic_score_codex":0.00026155892,"about_ca_topic_score_gemma":0.00015167106,"teacher_disagreement_score":0.39321566,"about_ca_system_score_codex":0.001210139,"about_ca_system_score_gemma":0.0012858872,"threshold_uncertainty_score":0.9999054},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7045556824","doi":"","title":"Bibliometric analysis on authentic leadership","year":2022,"lang":"en","type":"article","venue":"Universiti Putra Malaysia Institutional Repository (Universiti Putra Malaysia)","topic":"Education and Communication Studies","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Authentic leadership; Publish or perish; Scopus; Educational leadership; Publication; Leadership studies; Work (physics); Field (mathematics)","score_opus":0.06141262096592147,"score_gpt":0.285185714733808,"score_spread":0.2237730937678865,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7045556824","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7177887,0.00036391112,0.0006839575,0.008903121,0.00091570115,0.00036268035,0.00007246232,0.00031166253,0.27059782],"genre_scores_gemma":[0.9648614,0.00013246054,0.00027935515,0.0010376358,0.00022979826,0.000022843553,0.00014367644,0.000020458545,0.033272363],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9954384,0.000904777,0.0003683871,0.00079145,0.0017494534,0.0007475264],"domain_scores_gemma":[0.9975219,0.00043832016,0.0003627554,0.0007925017,0.0004565942,0.00042797925],"candidate_categories":["metaepi_narrow","bibliometrics","sts","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0008301139,0.00034565796,0.00044888377,0.0386752,0.007527861,0.00023266516,0.0015991612,0.00015919062,0.0031052304],"category_scores_gemma":[0.00006957559,0.00046753953,0.00062328135,0.088749535,0.0011803664,0.0010783466,0.0004421348,0.00070756214,0.00017330472],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023777518,0.0012972922,0.014967376,0.000035392994,0.002007026,0.0006641665,0.012945851,0.2677518,0.00020359151,0.6863716,0.011281535,0.0022365644],"study_design_scores_gemma":[0.0022981658,0.00048567128,0.053876583,0.0000509574,0.0026360787,0.00012539604,0.30196503,0.015200097,0.000111108064,0.00089376187,0.62054616,0.0018109796],"about_ca_topic_score_codex":0.0008774028,"about_ca_topic_score_gemma":0.000020986125,"teacher_disagreement_score":0.68547785,"about_ca_system_score_codex":0.002998162,"about_ca_system_score_gemma":0.0015946015,"threshold_uncertainty_score":0.9997776},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W7045715355","doi":"","title":"Bibliometric Analysis of Non-Vitamin K Antagonist Oral Anticoagulants (NOACS) in the Prevention of Venous Thrombosis and Pulmonary Embolism","year":2025,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Magnetic confinement fusion research","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Pulmonary embolism; Venous thrombosis; Thrombosis; Venous thromboembolism; Agriculture; Central venous catheter; Original research; Venous blood","score_opus":0.16564877550342746,"score_gpt":0.5420612918145787,"score_spread":0.37641251631115125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7045715355","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98356885,0.0018196867,0.0003740529,0.00016575718,0.000072248644,0.00056864705,0.0000837657,0.0000027565902,0.013344212],"genre_scores_gemma":[0.99615055,0.0033977218,0.00006796407,0.000021125059,0.000019985979,0.000029964447,0.000030391868,0.000009081969,0.00027318855],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9975366,0.00033472004,0.00088949664,0.0002984512,0.0006769117,0.00026382075],"domain_scores_gemma":[0.99816924,0.00042272866,0.0006274074,0.00044515933,0.00026653925,0.00006893514],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0017460709,0.00018827934,0.0009513616,0.03398592,0.0001196443,0.0003789365,0.0015274892,0.00005046863,0.021234272],"category_scores_gemma":[0.000054941494,0.00014970083,0.00021989281,0.06494167,0.00015569624,0.0002596696,0.0007313329,0.00022854301,8.864328e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000045855755,0.0010338125,0.7742581,0.00009722814,0.00070632866,0.000002986182,0.00009119079,0.000030398944,0.051962987,0.00009737726,0.0015847563,0.17008899],"study_design_scores_gemma":[0.00041153605,0.00002101726,0.9901819,0.00020005867,0.0004742411,2.6532163e-7,0.0003580878,0.001067413,0.006101722,0.00062771514,0.00043201557,0.00012402597],"about_ca_topic_score_codex":0.008332512,"about_ca_topic_score_gemma":0.0001192696,"teacher_disagreement_score":0.2159238,"about_ca_system_score_codex":0.000011286043,"about_ca_system_score_gemma":0.00015512314,"threshold_uncertainty_score":0.9982711},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7046278966","doi":"","title":"Contributions to the Field of Library and Information Sciences in Pakistan: A Bio Bibliometric Study of Dr. Saeed Ullah Jan","year":2022,"lang":"en","type":"article","venue":"Lincoln (University of Nebraska)","topic":"Magnetic confinement fusion research","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Children's Research Centre","keywords":"Honor; Publication; Field (mathematics); Publishing; Subject (documents); Commission; Czech","score_opus":0.009464664653716476,"score_gpt":0.2839063454145276,"score_spread":0.2744416807608111,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7046278966","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97892624,0.00002110619,0.00017749549,0.0062745325,0.000018526038,0.00051022833,0.00012377388,0.0000031789364,0.013944893],"genre_scores_gemma":[0.99946845,0.000004974712,0.00006205041,0.00003788934,0.000004810579,0.0000020049895,0.000007744353,8.197205e-7,0.0004112569],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99935627,0.00007237891,0.0001338065,0.00007740971,0.00027563906,0.000084523286],"domain_scores_gemma":[0.9994598,0.00024586733,0.00010798488,0.000110020854,0.000048225855,0.00002809016],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0002495558,0.000038059152,0.000109254215,0.0044733393,0.00018207174,0.0000143712205,0.00027946016,0.00000952558,0.007796697],"category_scores_gemma":[0.000013451103,0.000037576596,0.000024381696,0.015225453,0.00005919808,0.00015342222,0.0005132228,0.00007864929,9.4491014e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003260211,0.0014845094,0.6412124,0.000060258164,0.000077243516,0.000002919877,0.008381661,0.00049973634,0.0010276375,0.023797326,0.021123163,0.30200708],"study_design_scores_gemma":[0.0020996334,0.0019991223,0.7998784,0.000015119928,0.000020792966,1.7228773e-7,0.114563905,0.00206339,0.00026268468,0.00013577144,0.07886246,0.000098520635],"about_ca_topic_score_codex":0.0034700045,"about_ca_topic_score_gemma":0.00016269478,"teacher_disagreement_score":0.30190858,"about_ca_system_score_codex":0.0000035627288,"about_ca_system_score_gemma":0.00010699431,"threshold_uncertainty_score":0.9931103},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7046965258","doi":"","title":"Electricity Price Fundamentals in Deregulated Markets: A Bibliometric Analysis","year":2022,"lang":"en","type":"article","venue":"Dialnet (Universidad de la Rioja)","topic":"Magnetic confinement fusion research","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Electricity; Liberalization; Electricity market; Scopus; Electricity price forecasting; Electric power industry","score_opus":0.008167614891965078,"score_gpt":0.25805202005655586,"score_spread":0.2498844051645908,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7046965258","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8506321,0.00008347624,0.00068889174,0.0003480955,0.000023016855,0.00028809108,0.0000537147,0.000027726232,0.14785491],"genre_scores_gemma":[0.99315745,0.000015375952,0.00015333747,0.000059599173,0.000021359643,0.00004232496,0.00009803324,0.000011088043,0.0064414507],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99787194,0.00056883576,0.00019172292,0.0003448569,0.00054088526,0.00048177878],"domain_scores_gemma":[0.9989907,0.0004207606,0.000110873596,0.00028973786,0.00004742969,0.00014049403],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0008861133,0.00014414608,0.0002584474,0.042260416,0.0002524895,0.000101342936,0.00042708963,0.00003767944,0.27375662],"category_scores_gemma":[0.000021912332,0.000178128,0.0001699481,0.15336156,0.000051873776,0.000051156927,0.00038813098,0.00037038157,0.000021643558],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002262402,0.00093871204,0.7220308,0.000016627651,0.0007602472,0.00011859994,0.00062155136,0.00078955345,0.0036190865,0.00427388,0.015127485,0.25147718],"study_design_scores_gemma":[0.0015139813,0.00008611637,0.918519,0.0000027192139,0.00014288726,0.0000011574605,0.0017205904,0.01328074,0.00015781751,0.00029742805,0.06403239,0.00024516563],"about_ca_topic_score_codex":0.005578623,"about_ca_topic_score_gemma":0.00012077802,"teacher_disagreement_score":0.273735,"about_ca_system_score_codex":0.0001447073,"about_ca_system_score_gemma":0.00019860288,"threshold_uncertainty_score":0.96859473},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7047123583","doi":"","title":"Exploring paths for the normalization of altmetrics: Applying the Characteristic Scores and Scales. The 2016 Altmetrics Workshop, Bucharest, Romania","year":2016,"lang":"en","type":"article","venue":"Leiden Repository (Leiden University)","topic":"Superconducting and THz Device Technology","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Université de Montréal; Universiteit Leiden; Department of Science and Technology, Ministry of Science and Technology, India; Alfred P. Sloan Foundation","keywords":"Normalization (sociology); Altmetrics; Focus (optics); Probabilistic logic; Pattern recognition (psychology)","score_opus":0.0447820698840049,"score_gpt":0.21581563146882188,"score_spread":0.17103356158481697,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7047123583","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9762944,0.00037239364,0.021186946,0.0006016847,0.0004566677,0.00073775876,0.000021252607,0.00006331528,0.0002655572],"genre_scores_gemma":[0.99797094,0.00013252428,0.00007363809,0.000017684786,0.00032094232,0.000053425436,0.0000047452554,0.000020242274,0.0014058339],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9989176,0.00010281935,0.00024000369,0.00028682084,0.0001948161,0.00025795848],"domain_scores_gemma":[0.9971591,0.0018126506,0.0002631056,0.0005105536,0.00020974739,0.000044794677],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00035022225,0.00017066956,0.00019858703,0.00024785075,0.0007878276,0.00007534764,0.0006242321,0.00006230295,0.000006829267],"category_scores_gemma":[0.00014330537,0.00008813093,0.000111623456,0.0008851716,0.00027900236,0.00044765233,0.00023714453,0.00016475405,0.0000043679424],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006135682,0.00009224232,0.7924574,0.0000787804,0.0005402408,0.00001755979,0.001184416,0.000048137044,0.012358401,0.057029206,0.0004017288,0.13573052],"study_design_scores_gemma":[0.0063974047,0.00060986367,0.6132145,0.0018452768,0.0028195055,0.000117061696,0.11116846,0.0024292953,0.07480185,0.0039355536,0.1802234,0.0024378288],"about_ca_topic_score_codex":0.0009092712,"about_ca_topic_score_gemma":0.000028425819,"teacher_disagreement_score":0.17982168,"about_ca_system_score_codex":0.000048835227,"about_ca_system_score_gemma":0.00005193778,"threshold_uncertainty_score":0.60594124},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7062477308","doi":"","title":"Uncovering Impact: Bibliometric Approaches to Assessing Library Collections","year":2023,"lang":"en","type":"other","venue":"MacSphere (McMaster University)","topic":"Magnetic confinement fusion research","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Nucleofection; Gestational period; Dysgeusia; TSG101; Liquation; Diafiltration; Emperipolesis; Triacetin; Durvalumab","score_opus":0.050034517439323675,"score_gpt":0.2472954975809778,"score_spread":0.1972609801416541,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7062477308","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000075050426,0.000030937437,0.0015360425,0.00016229572,0.00018416806,0.0005421056,0.00017475252,0.00046789905,0.99682677],"genre_scores_gemma":[0.00038451876,0.000013333123,0.00075592013,0.000013327612,0.0002877089,0.0000058423575,0.00003836605,0.0004266367,0.99807435],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.998515,0.00006966014,0.00014735224,0.0004903664,0.00032786373,0.0004497534],"domain_scores_gemma":[0.99915344,0.000096125776,0.00011439221,0.00034541756,0.000017477061,0.000273124],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.000058106772,0.00031193116,0.0003085416,0.01895381,0.00018308203,0.00046518684,0.00047059957,0.00012016584,0.9530239],"category_scores_gemma":[0.0000055856385,0.0003328427,0.00018494231,0.053756226,0.000037960057,0.000108263324,0.0006802522,0.0002843463,0.0006214096],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000007609245,0.00004787421,0.0042057987,0.000028703462,0.00013310528,0.000016743752,0.000018784875,0.000052108677,0.0000072688113,0.00021046806,0.4850214,0.51025015],"study_design_scores_gemma":[0.00038925564,0.000051266627,0.0013533784,0.00012681844,0.000046248504,2.9257592e-7,0.0006790063,0.000104423,0.000021897893,0.000020067533,0.9968757,0.00033161842],"about_ca_topic_score_codex":0.003812431,"about_ca_topic_score_gemma":0.00044595756,"teacher_disagreement_score":0.95240253,"about_ca_system_score_codex":0.000044747714,"about_ca_system_score_gemma":0.00023459127,"threshold_uncertainty_score":0.9999124},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W7070471512","doi":"","title":"Overlap Between Humanities Faculty Citations and Library Monograph Collections 2005-2009","year":2012,"lang":"en","type":"article","venue":"CU Scholar (University of Colorado Boulder)","topic":"Library Collection Development and Digital Resources","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Subject (documents); Quarter (Canadian coin); Citation; Collection development; Plan (archaeology); Interlibrary loan; Resource (disambiguation); Digital humanities","score_opus":0.03142714228716827,"score_gpt":0.2122819147785895,"score_spread":0.1808547724914212,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7070471512","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9166445,0.0006839907,0.010980832,0.0012418504,0.00018629074,0.0002758573,0.00009460906,0.0003856658,0.0695064],"genre_scores_gemma":[0.93877953,0.00013685961,0.02236041,0.00018017455,0.00006363402,0.0000013639534,0.000048218215,0.000012498893,0.03841729],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990684,0.0000705332,0.00013041118,0.00021096427,0.00022858939,0.0002911134],"domain_scores_gemma":[0.99933517,0.00010256669,0.00011292477,0.00018992259,0.000053679425,0.00020574342],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.00012667953,0.00013219226,0.0001845338,0.0006729239,0.0013722647,0.00047268727,0.0004665935,0.00009795878,0.000109679095],"category_scores_gemma":[0.000022992634,0.00016003965,0.000085001404,0.0015516554,0.00016324724,0.009724685,0.00034589774,0.00020685849,0.00002280217],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000051000272,0.0003241749,0.68694484,0.00006424159,0.00030618554,0.000010212485,0.015007483,0.000015730742,0.00009717602,0.058879197,0.23133576,0.006963996],"study_design_scores_gemma":[0.00055975653,0.00008539455,0.48533717,0.00001961068,0.000028671178,0.0000054634743,0.0036657343,0.00013657018,0.00014760275,0.0026010377,0.50713986,0.00027312333],"about_ca_topic_score_codex":0.00002387859,"about_ca_topic_score_gemma":0.000010476789,"teacher_disagreement_score":0.27580413,"about_ca_system_score_codex":0.000024461398,"about_ca_system_score_gemma":0.00013330579,"threshold_uncertainty_score":0.9999278},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W7071087944","doi":"","title":"Science, Technology and Innovation Indicators in a Changing World: Responding to Policy Needs","year":2007,"lang":"en","type":"other","venue":"UNU Collections (United Nations University)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Exploit; Public policy; Policy development; Selection (genetic algorithm); Emerging technologies; Policy analysis; Work (physics)","score_opus":0.011739401911016759,"score_gpt":0.26319757210812245,"score_spread":0.2514581701971057,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7071087944","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0030877015,0.000014570884,0.0018523573,0.00070827425,0.0001420014,0.0009089875,0.00013796931,0.0008662189,0.9922819],"genre_scores_gemma":[0.0052657556,0.000022526132,0.0024746747,0.0000756062,0.000079327256,0.000008994765,0.00006680301,0.00024636753,0.99175996],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980501,0.000081587204,0.00026169737,0.00053519267,0.0004048954,0.0006665371],"domain_scores_gemma":[0.998611,0.000118722724,0.0003072625,0.00039469873,0.00041924248,0.0001491021],"candidate_categories":["metaepi_narrow","bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00069122034,0.0003019134,0.0002785489,0.9998903,0.0018850174,0.0001190501,0.0005257944,0.00041159152,0.00019618821],"category_scores_gemma":[0.00091017096,0.000432177,0.000031477153,0.99935615,0.0006505512,0.00032257088,0.00035809015,0.0006045713,0.00011031982],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000025398393,0.00008040508,0.0003403406,0.000008448674,0.000037860344,0.000015229868,0.0006392592,0.00007190256,0.0005117919,0.9875327,0.010407286,0.00032942783],"study_design_scores_gemma":[0.00068666803,0.000053094376,0.000004569397,0.00021751494,0.000036631292,0.000013672709,0.0026665644,0.00010905729,0.00008904571,0.00020795401,0.9955407,0.00037457063],"about_ca_topic_score_codex":0.00001600964,"about_ca_topic_score_gemma":0.00030465357,"teacher_disagreement_score":0.9873247,"about_ca_system_score_codex":0.0026704508,"about_ca_system_score_gemma":0.0014522773,"threshold_uncertainty_score":0.999813},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7071672398","doi":"","title":"Türkiye’nin Göreli Atıf Etkisi Üzerine Bilimetrik Bir Çalışma A Scientometric Study on Relative Citation Impact of Turkey","year":2009,"lang":"tr","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Maritime and Coastal Archaeology","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Citation impact; Field (mathematics); Field survey","score_opus":0.3361027803433506,"score_gpt":0.5678323810237951,"score_spread":0.23172960068044446,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7071672398","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9503856,0.009863447,0.000033210472,0.000291177,0.0007991424,0.0016793241,0.0004162007,0.000049618015,0.03648233],"genre_scores_gemma":[0.99215114,0.0040090973,0.000054561697,0.0001868699,0.00043523888,0.00003162282,0.00004327387,0.000075482814,0.0030127305],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.992953,0.0009977724,0.0022244528,0.001023983,0.0018557801,0.0009450364],"domain_scores_gemma":[0.99379385,0.0013668748,0.0023799615,0.00071937445,0.0012230177,0.0005169411],"candidate_categories":["metaepi_narrow","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0032850595,0.0008494425,0.0019950909,0.01111842,0.0008580128,0.0016900775,0.0030837473,0.00022492761,0.030785251],"category_scores_gemma":[0.0011372515,0.00070302037,0.00082467624,0.0065877764,0.0009044296,0.003585646,0.0014106835,0.0011150569,0.000080171776],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.005663405,0.016023962,0.74470973,0.00034494232,0.0037983144,0.0002668758,0.036496628,0.00089133345,0.014194359,0.01840424,0.041362494,0.11784372],"study_design_scores_gemma":[0.0024627282,0.002245756,0.9640798,0.00066962815,0.0004165059,0.000012813362,0.0032305755,0.00018470567,0.0013941148,0.02064955,0.0037925649,0.0008612391],"about_ca_topic_score_codex":0.010359786,"about_ca_topic_score_gemma":0.00025754268,"teacher_disagreement_score":0.21937008,"about_ca_system_score_codex":0.0003354515,"about_ca_system_score_gemma":0.00033310463,"threshold_uncertainty_score":0.9995421},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7073995987","doi":"","title":"Bibliometric analysis of blueberry (Vaccinium corymbosum L.) research publications based on Web of Science","year":2022,"lang":"en","type":"article","venue":"LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)","topic":"Theoretical and Computational Physics","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Web of science; Agriculture; Work (physics); Bibliometrics; Web site; Information science","score_opus":0.03166916161425209,"score_gpt":0.29568478399419035,"score_spread":0.26401562237993825,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7073995987","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.71003544,0.00006659944,0.007945496,0.005875264,0.00043880488,0.0005576768,0.0006419408,0.000127987,0.2743108],"genre_scores_gemma":[0.9986608,0.000002010535,0.0006403818,0.000046425084,0.00011924766,0.00018679915,0.00021801465,0.000019659878,0.00010664169],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99577576,0.00044234062,0.0006339648,0.00058243383,0.0020463853,0.0005191394],"domain_scores_gemma":[0.99632555,0.0009094438,0.00039348562,0.0006567055,0.0014377814,0.00027700214],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0018499912,0.00020796111,0.00042191852,0.03566951,0.0010438721,0.00009922793,0.0010621736,0.00006461309,0.00038583466],"category_scores_gemma":[0.0002657969,0.00020659927,0.0003150098,0.12673125,0.00016273998,0.00024823815,0.00042749528,0.00059592153,0.0000038861203],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010583552,0.0013596218,0.005513521,0.00001636079,0.00021893838,0.000001626755,0.00019043284,0.011967629,0.0026207003,0.9763712,0.00027472718,0.0013594028],"study_design_scores_gemma":[0.003082628,0.0008053342,0.28375345,0.00008712992,0.0008680978,0.000020208896,0.002414266,0.16924185,0.009832481,0.039613217,0.4891222,0.0011591231],"about_ca_topic_score_codex":0.000003684349,"about_ca_topic_score_gemma":0.000003616009,"teacher_disagreement_score":0.936758,"about_ca_system_score_codex":0.0002951992,"about_ca_system_score_gemma":0.0018413966,"threshold_uncertainty_score":0.9752604},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7080732969","doi":"10.1177/21582440251367078","title":"A Bibliometric Analysis of Publication Trends of Empirical Studies Over the Past 50 years Using the Child Attachment Studies Catalog and Data Exchange (CASCADE)","year":2025,"lang":"en","type":"article","venue":"SAGE Open","topic":"Geochemistry and Geologic Mapping","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; University of Calgary","funders":"","keywords":"Observational study; Ethnic group; Empirical research; Socioeconomic status; Bibliometrics; Population; Research design; Information exchange","score_opus":0.15820506134160872,"score_gpt":0.4317555586746182,"score_spread":0.27355049733300946,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7080732969","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.79032487,0.0968369,0.0135535775,0.085152164,0.00038195724,0.0012013657,0.00026878194,0.000059238053,0.012221127],"genre_scores_gemma":[0.9974993,0.0007600778,0.00063779374,0.0003127135,0.000017050343,0.000015962898,0.00002560712,0.0000010442515,0.00073046604],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9990243,0.000103591796,0.00024128627,0.00032133918,0.00018896737,0.000120569064],"domain_scores_gemma":[0.99828947,0.0002900572,0.00018952176,0.001084252,0.00013096812,0.000015740481],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0012553198,0.000081941784,0.0002740796,0.0046522007,0.00016283608,0.00011729224,0.0018886673,0.000027686894,0.000012147157],"category_scores_gemma":[0.0003365297,0.00004603751,0.000040948496,0.059671007,0.00017423896,0.0002864164,0.0050692647,0.00007439825,2.1435852e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002969684,0.00036449402,0.5321768,0.00040974017,0.011456843,0.000007047907,0.018198734,0.00046908992,0.00028598242,0.0017048401,0.066777736,0.368119],"study_design_scores_gemma":[0.00021898469,0.000021555144,0.96995,0.00006083623,0.00049433793,0.0000020679463,0.0013139416,0.007785809,0.00012189442,0.00023497232,0.019708106,0.00008746298],"about_ca_topic_score_codex":0.0001539663,"about_ca_topic_score_gemma":0.00008851583,"teacher_disagreement_score":0.43777323,"about_ca_system_score_codex":0.00001743492,"about_ca_system_score_gemma":0.000020390728,"threshold_uncertainty_score":0.9603161},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7081994849","doi":"10.1016/j.eneco.2025.108899","title":"A comparative bibliometric analysis of five major energy journals","year":2025,"lang":"en","type":"article","venue":"Energy Economics","topic":"Geochemistry and Geologic Mapping","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Notice; Energy (signal processing); Energy economics; Energy policy; Resource (disambiguation); Environmental impact of the energy industry; Energy market; Energy transition","score_opus":0.022381309250906412,"score_gpt":0.26284674562465815,"score_spread":0.24046543637375173,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7081994849","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25169873,0.002011001,0.5587548,0.0020574096,0.0004636768,0.000033305645,0.000018177943,0.00006982724,0.18489307],"genre_scores_gemma":[0.9899279,0.00026932405,0.0022799494,0.00034544308,0.000022731201,0.0000068939844,0.000008963249,9.91235e-7,0.0071378173],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9990267,0.00004520181,0.00038581213,0.00029944745,0.00005382042,0.00018898485],"domain_scores_gemma":[0.99884826,0.00022898454,0.00026185237,0.0004406304,0.00016124551,0.00005901619],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00022418727,0.000116462565,0.00045063483,0.023584606,0.00007211935,0.0000848767,0.00074931653,0.00007594889,0.0000884282],"category_scores_gemma":[0.000048165053,0.00011697741,0.00019045673,0.057480916,0.000044713328,0.00018319479,0.00026368798,0.00005771365,0.0000011565857],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000026135775,0.00023974937,0.017351707,0.000022432681,0.0056624063,0.000010229707,0.0005221795,0.24255782,0.0008466253,0.6941825,0.0076435944,0.030934623],"study_design_scores_gemma":[0.00032284562,0.000026489677,0.011886409,0.000015181352,0.0001967389,0.0000028391705,0.00009858428,0.84490067,0.030784016,0.027844915,0.08368952,0.00023178781],"about_ca_topic_score_codex":0.00028493573,"about_ca_topic_score_gemma":0.0001636003,"teacher_disagreement_score":0.73822916,"about_ca_system_score_codex":0.000041409447,"about_ca_system_score_gemma":0.00007630211,"threshold_uncertainty_score":0.98748225},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7084042380","doi":"10.6084/m9.figshare.30188249","title":"Unraveling retraction dynamics in COVID-19 research: Patterns, reasons, and implications","year":2025,"lang":"en","type":"other","venue":"Figshare","topic":"Plant Pathogenic Bacteria Studies","field":"Agricultural and Biological Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Quarter (Canadian coin); Quality (philosophy); Dynamics (music); Panel Study of Income Dynamics; Period (music)","score_opus":0.1323103782581213,"score_gpt":0.33868054365874284,"score_spread":0.20637016540062153,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7084042380","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0005393808,0.001613856,2.8292783e-7,0.0029747467,0.00006223753,0.0005185266,0.912308,0.00014614295,0.08183687],"genre_scores_gemma":[0.017120399,0.0038753857,0.0000522439,0.00060901686,0.0006887882,0.0006687333,0.8302922,0.000019039559,0.14667414],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990346,0.00011420058,0.00012940571,0.0003646038,0.00012861507,0.00022861903],"domain_scores_gemma":[0.99922967,0.00048464973,0.000085155625,0.00007523461,0.000041393472,0.00008392336],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00013898779,0.00013215533,0.00016767516,0.00006565686,0.00013658879,0.00006926531,0.00019541009,0.00032828675,0.039232254],"category_scores_gemma":[0.0015727194,0.00006242463,0.000029205727,0.0003003865,0.000009749307,0.000033802055,0.00019980173,0.00033969298,0.00013636741],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004164316,0.000015458569,0.0018074281,0.00014140242,0.000010345673,0.000007954605,0.000008772341,5.2060898e-8,0.00022879566,0.00004456294,0.98027205,0.017459005],"study_design_scores_gemma":[0.00005836981,0.000015107212,0.07935442,0.0024337359,0.000004075821,0.0000073633632,0.00018923094,0.0000109567345,0.0000027668084,0.00007311256,0.9176937,0.0001572162],"about_ca_topic_score_codex":0.0010249466,"about_ca_topic_score_gemma":0.036601573,"teacher_disagreement_score":0.0820157,"about_ca_system_score_codex":0.00019293993,"about_ca_system_score_gemma":0.000046385423,"threshold_uncertainty_score":0.98097795},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7087864565","doi":"","title":"Global Research Trends in Childhood Asthma and the Microbiome: A Bibliometric Analysis of 2000 to 2024","year":2025,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Gold and Silver Nanoparticles Synthesis and Applications","field":"Materials Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Pediatric Oncology Group","funders":"","keywords":"Thematic map; Service (business); Thematic analysis; Bibliometrics; Community health; Public health","score_opus":0.17029151604286316,"score_gpt":0.5585511339404221,"score_spread":0.3882596178975589,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7087864565","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98494077,0.010054336,0.000029293917,0.00094503985,0.000086332526,0.00027419263,0.00013706589,0.000006680293,0.0035263172],"genre_scores_gemma":[0.9971964,0.0020836215,0.00011003583,0.00011780092,0.000018647523,0.00005345636,0.000002658232,0.0000063376942,0.00041106518],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9976522,0.00042982624,0.0007280923,0.00038972063,0.000455805,0.00034439587],"domain_scores_gemma":[0.998361,0.0005470962,0.00023313453,0.00047085117,0.00025655632,0.00013137849],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0039589885,0.00014415302,0.00067110726,0.042856973,0.00021265453,0.0010323584,0.0018958303,0.000056567253,0.0050043575],"category_scores_gemma":[0.00029032695,0.000099190846,0.00015963598,0.20973007,0.00026696007,0.0004375456,0.0009257126,0.00013565947,0.00000904595],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000190824,0.00045685525,0.30198535,0.000020465857,0.00023808389,0.0000025332931,0.00023998231,0.00023675384,0.6206739,0.0010502998,0.021206854,0.05369808],"study_design_scores_gemma":[0.00043703613,0.0000053328604,0.92052424,0.00009371667,0.00012960153,0.0000010100908,0.00008328751,0.00005632971,0.07476703,0.0017894785,0.0020136756,0.0000992681],"about_ca_topic_score_codex":0.0033261739,"about_ca_topic_score_gemma":0.0005469231,"teacher_disagreement_score":0.6185389,"about_ca_system_score_codex":0.00008118935,"about_ca_system_score_gemma":0.0001226886,"threshold_uncertainty_score":0.9959052},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7093738341","doi":"","title":"Research on wildfires and remote sensing in the last three decades: a bibliometric analysis","year":2021,"lang":"en","type":"article","venue":"RepositóriUM (Universidade do Minho)","topic":"Fire effects on ecosystems","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Field (mathematics); Scopus; Earth observation; Bibliometrics; Trend analysis; Focus (optics)","score_opus":0.0237974223555845,"score_gpt":0.28297271924008577,"score_spread":0.25917529688450125,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7093738341","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98191124,0.00036542892,0.00032365692,0.0008845594,0.000051122708,0.00025097412,0.0000028120278,0.000026861217,0.016183313],"genre_scores_gemma":[0.9986382,0.00006037647,0.00083905424,0.00010861816,0.000032231153,7.051387e-7,0.0000049909568,0.000016818958,0.00029897192],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9967303,0.0007539652,0.0002171343,0.0007111777,0.001072662,0.00051472057],"domain_scores_gemma":[0.9979733,0.0010122239,0.00009069023,0.0007556152,0.00003827952,0.00012988715],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.001814227,0.00017852122,0.00027857232,0.016345121,0.00037622248,0.00030257003,0.00036265492,0.00014126507,0.00011767594],"category_scores_gemma":[0.00021173834,0.00015317091,0.00012349492,0.14271034,0.00021369751,0.0002726417,0.0003529199,0.00047710072,0.00013574389],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023330413,0.00049236394,0.46352017,0.00017936273,0.00087957596,0.015741933,0.010942828,0.0028262667,0.061530303,0.0004896064,0.004716171,0.43844813],"study_design_scores_gemma":[0.00039499562,0.00013204444,0.96230954,0.00009049695,0.00013191791,0.00027093999,0.0033950987,0.030164186,0.0013766085,0.00010533508,0.0013947566,0.0002340951],"about_ca_topic_score_codex":0.008455199,"about_ca_topic_score_gemma":0.008794572,"teacher_disagreement_score":0.49878937,"about_ca_system_score_codex":0.0003780225,"about_ca_system_score_gemma":0.000027056862,"threshold_uncertainty_score":0.9981476},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7095057608","doi":"10.22034/ijism.2025.2036425.1546","title":"Bibliometric analysis of Iranian Authors in High-Impact Medical Science Journals","year":2025,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"China; Bibliometrics; Citation; Citation analysis; Specialty; Citation impact; MEDLINE; Middle East","score_opus":0.7670192304732296,"score_gpt":0.7585526557240697,"score_spread":0.008466574749159905,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7095057608","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96772057,0.02188068,0.0025643564,0.00069009466,0.0010713887,0.0003983617,0.00010146122,0.000013647288,0.005559467],"genre_scores_gemma":[0.98307055,0.015899427,0.0002833873,0.0001845247,0.00005398226,0.0000108894665,0.000002957689,0.000018058365,0.00047621294],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9604772,0.001314792,0.0043598628,0.0014786981,0.030939547,0.0014299509],"domain_scores_gemma":[0.9730673,0.011525346,0.0025966812,0.0017839007,0.009195717,0.0018310866],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.14015329,0.00041792213,0.0024965522,0.932267,0.000553382,0.01531097,0.028213944,0.00032921834,0.04238243],"category_scores_gemma":[0.17344816,0.00028939274,0.00089417247,0.976887,0.0012697957,0.0068269814,0.0064069307,0.0012395588,0.000031808846],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000088051595,0.00048039143,0.81760377,0.0000138502955,0.00037523208,0.000053478532,0.000049430953,0.0008511286,0.008150949,0.00015721921,0.013010899,0.1591656],"study_design_scores_gemma":[0.0006510754,0.000026322168,0.9786157,0.0002022909,0.0001461435,0.000005071041,0.000103262246,0.0049044266,0.004072289,0.010004878,0.0009823772,0.00028617174],"about_ca_topic_score_codex":0.0046874583,"about_ca_topic_score_gemma":0.0002452989,"teacher_disagreement_score":0.16101193,"about_ca_system_score_codex":0.0005795046,"about_ca_system_score_gemma":0.003202426,"threshold_uncertainty_score":0.99995583},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7096841259","doi":"","title":"Bibliometric analysis of output and impact based on CRIS data: a case study on the registered output of a Dutch university","year":2014,"lang":"en","type":"article","venue":"","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Variety (cybernetics); Bibliometrics; Perspective (graphical); Order (exchange); Web of science; Information system","score_opus":0.7104587863483294,"score_gpt":0.5599914982932118,"score_spread":0.15046728805511755,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7096841259","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9887248,0.00003619725,0.006297264,0.0004160996,0.000052356827,0.0003572987,0.00026912885,0.000009516665,0.0038373377],"genre_scores_gemma":[0.9987784,0.000020689995,0.00014358018,0.00006815401,0.000009333605,6.6002457e-7,0.000004991104,0.000005830792,0.00096833793],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98959535,0.001047461,0.0006724826,0.0009458484,0.007382758,0.00035609453],"domain_scores_gemma":[0.9704817,0.022543043,0.00057254353,0.0036538471,0.0024087315,0.00034012782],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.036883797,0.00017909441,0.0006926326,0.75836414,0.0002143486,0.00066910044,0.0030344983,0.000076841294,0.0002491142],"category_scores_gemma":[0.04747846,0.00009085402,0.00026057073,0.9133988,0.00023391537,0.00034709251,0.0011215956,0.00019750014,0.000014274594],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019355741,0.0013612576,0.9079389,0.0000059239123,0.00045276748,0.00014923899,0.00025717582,0.0005970347,0.000024580831,0.00013091473,0.01122577,0.07766286],"study_design_scores_gemma":[0.0008088789,0.0015999783,0.77874756,0.0000038041537,0.00021422554,0.0000053528483,0.0046788263,0.21299137,0.00003578137,0.00003457784,0.000764987,0.000114642105],"about_ca_topic_score_codex":0.005917748,"about_ca_topic_score_gemma":0.00038183035,"teacher_disagreement_score":0.21239434,"about_ca_system_score_codex":0.00006538398,"about_ca_system_score_gemma":0.00014666184,"threshold_uncertainty_score":0.9917308},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7100142665","doi":"","title":"Günter Krampen &amp;amp; Han s-Werner Wahl Geropsychology and Psychology in the Last Quarter of the 20th Century Geropsychology and Psychology in the Last Quarter of the 20th Century Bibliometrical Results for the German-Speaking Versus Anglo-American Research Co","year":2016,"lang":"en","type":"article","venue":"","topic":"Aging and Gerontology Research","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Quarter (Canadian coin); Position (finance); History of psychology; Period (music); Experimental psychology","score_opus":0.1310074953426608,"score_gpt":0.47870274637475757,"score_spread":0.34769525103209675,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7100142665","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.89029706,0.0016969907,0.00041398406,0.07848579,0.0021861233,0.0024026367,0.0001470433,0.000033473312,0.024336936],"genre_scores_gemma":[0.9918111,0.0007626763,0.00012457378,0.005851163,0.0003443854,0.00046685096,0.000016783559,0.00007052837,0.00055194914],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.98453915,0.0089238,0.0015823545,0.0017716584,0.0009945111,0.0021885142],"domain_scores_gemma":[0.98329693,0.011729823,0.000688259,0.0037926442,0.00032964838,0.00016272278],"candidate_categories":["metaepi_narrow","sts","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.009427545,0.00063727476,0.0010307012,0.002603402,0.00057649595,0.00010283221,0.0037159973,0.00078157865,0.00037025227],"category_scores_gemma":[0.0009566788,0.00025559275,0.00033780202,0.006655581,0.0080688065,0.00017223912,0.00048961537,0.0024041564,0.000116162904],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.017309718,0.0029883378,0.32560593,0.00006551485,0.0007882247,0.000056629895,0.06608365,9.867526e-7,0.0025241869,0.008262268,0.50807387,0.06824069],"study_design_scores_gemma":[0.010000081,0.0013719941,0.8298692,0.00005688777,0.000059208138,0.0004395153,0.008360713,0.0000056109097,0.000013272758,0.0007966208,0.14868197,0.0003448952],"about_ca_topic_score_codex":0.00093675655,"about_ca_topic_score_gemma":0.0077061458,"teacher_disagreement_score":0.50426334,"about_ca_system_score_codex":0.00007498814,"about_ca_system_score_gemma":0.000085653824,"threshold_uncertainty_score":0.9999896},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"design_other","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7103995860","doi":"10.6084/m9.figshare.30543472.v1","title":"Advancements and Collaborative Dynamics in the Treatment of Retinoblastoma: A Bibliometric Analysis of Trends and Themes","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Citation analysis; Thematic analysis; Citation; Co-citation; Beijing; Web of science","score_opus":0.024311604284623024,"score_gpt":0.3307268067692036,"score_spread":0.30641520248458054,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7103995860","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003099472,0.024955226,6.5765896e-8,0.00002470842,0.000021917616,0.0011628142,0.9704695,0.000008399067,0.0002579258],"genre_scores_gemma":[0.0146503495,0.0075829546,0.000038765174,0.000012970274,0.00001516754,0.0007584117,0.9767848,0.00003659298,0.00011997525],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9946362,0.00079960655,0.0015804368,0.0012555367,0.0011455428,0.0005827041],"domain_scores_gemma":[0.9926742,0.002115318,0.0024942588,0.0015983304,0.0009855696,0.00013232658],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00025520136,0.0010670306,0.0027315896,0.267278,0.00014833295,0.00016488222,0.00083871785,0.00047741417,0.03028582],"category_scores_gemma":[0.0015143438,0.0008010016,0.0003660834,0.60790163,0.00016170154,0.00027319512,0.00042100568,0.0002952365,0.000018878396],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007460139,0.003405285,0.0026481892,0.0016988572,0.010352885,0.00006198966,0.0015962219,0.000120823264,0.0000036270594,0.0000049451664,0.94789803,0.031463154],"study_design_scores_gemma":[0.0070406045,0.006283835,0.11037573,0.0073546786,0.019677406,0.000005457112,0.0063427673,0.011859189,0.00011158232,0.0000075984217,0.8296376,0.001303579],"about_ca_topic_score_codex":0.00016438279,"about_ca_topic_score_gemma":0.005710401,"teacher_disagreement_score":0.34062365,"about_ca_system_score_codex":0.0008479917,"about_ca_system_score_gemma":0.00039719976,"threshold_uncertainty_score":0.99944407},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W7104044367","doi":"10.6084/m9.figshare.30543472","title":"Advancements and Collaborative Dynamics in the Treatment of Retinoblastoma: A Bibliometric Analysis of Trends and Themes","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Citation analysis; Thematic analysis; Citation; Co-citation; Beijing; Web of science","score_opus":0.024311604284623024,"score_gpt":0.3307268067692036,"score_spread":0.30641520248458054,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7104044367","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003099472,0.024955226,6.5765896e-8,0.00002470842,0.000021917616,0.0011628142,0.9704695,0.000008399067,0.0002579258],"genre_scores_gemma":[0.0146503495,0.0075829546,0.000038765174,0.000012970274,0.00001516754,0.0007584117,0.9767848,0.00003659298,0.00011997525],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9946362,0.00079960655,0.0015804368,0.0012555367,0.0011455428,0.0005827041],"domain_scores_gemma":[0.9926742,0.002115318,0.0024942588,0.0015983304,0.0009855696,0.00013232658],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00025520136,0.0010670306,0.0027315896,0.267278,0.00014833295,0.00016488222,0.00083871785,0.00047741417,0.03028582],"category_scores_gemma":[0.0015143438,0.0008010016,0.0003660834,0.60790163,0.00016170154,0.00027319512,0.00042100568,0.0002952365,0.000018878396],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007460139,0.003405285,0.0026481892,0.0016988572,0.010352885,0.00006198966,0.0015962219,0.000120823264,0.0000036270594,0.0000049451664,0.94789803,0.031463154],"study_design_scores_gemma":[0.0070406045,0.006283835,0.11037573,0.0073546786,0.019677406,0.000005457112,0.0063427673,0.011859189,0.00011158232,0.0000075984217,0.8296376,0.001303579],"about_ca_topic_score_codex":0.00016438279,"about_ca_topic_score_gemma":0.005710401,"teacher_disagreement_score":0.34062365,"about_ca_system_score_codex":0.0008479917,"about_ca_system_score_gemma":0.00039719976,"threshold_uncertainty_score":0.99944407},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7104451277","doi":"10.71781/22963","title":"Analyse bibliométrique de la contribution des postdoctorants canadiens à l’avancement des connaissances","year":2014,"lang":"fr","type":"dissertation","venue":"Papyrus : Institutional Repository (Université de Montréal)","topic":"Doctoral Education Challenges and Solutions","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Population; Context (archaeology); Face (sociological concept); Research methodology","score_opus":0.038002843255313355,"score_gpt":0.3451215453848234,"score_spread":0.3071187021295101,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7104451277","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.91820246,0.05334578,0.0019685654,0.0011179374,0.0026107298,0.00087229826,0.00037830978,0.00014874394,0.021355174],"genre_scores_gemma":[0.9544676,0.0076531176,0.0010147021,0.00025783916,0.00055944145,0.0001381843,0.0010544568,0.000054863267,0.034799814],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.99410474,0.0020746978,0.00094549026,0.00083020027,0.00066035584,0.0013845125],"domain_scores_gemma":[0.9944752,0.0009831432,0.000923908,0.0005134327,0.0018501253,0.0012542086],"candidate_categories":["metaepi_narrow","sts","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.0014013302,0.0006795989,0.00077117624,0.0014568287,0.02430228,0.000117186784,0.00054758287,0.0012966956,0.00046914248],"category_scores_gemma":[0.00095445587,0.00077683094,0.00043538318,0.0016937315,0.0018067996,0.0006244773,0.0001440764,0.0012792099,0.00015174624],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.0021866062,0.0016936393,0.3483237,0.0025785498,0.0019858782,0.0013260982,0.3635292,0.0074663823,0.014356996,0.23345347,0.010858246,0.012241218],"study_design_scores_gemma":[0.0026252547,0.00032880966,0.56440747,0.0025347467,0.0010106282,0.0003403814,0.182041,0.0024764368,0.001862725,0.0036454236,0.23753679,0.0011903574],"about_ca_topic_score_codex":0.43818444,"about_ca_topic_score_gemma":0.44359496,"teacher_disagreement_score":0.22980805,"about_ca_system_score_codex":0.02212926,"about_ca_system_score_gemma":0.00842534,"threshold_uncertainty_score":0.9999998},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"incentives","study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"medium"}],"label_agreement":"split"},{"id":"W7106034893","doi":"10.3138/jsp-2024-0049","title":"An Empirical Study of International Retracted Research Articles of Famous Chinese Universities from 2012 to 2022","year":2025,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Misconduct; Citation; Academic dishonesty; Ranking (information retrieval); Publishing; Empirical research; Scientific misconduct; Quality (philosophy); Academic integrity","score_opus":0.07012572944052395,"score_gpt":0.422144762125753,"score_spread":0.35201903268522905,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7106034893","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97537607,0.0001862841,0.00019028026,0.015003022,0.0006780205,0.000103230865,0.000023397919,0.000007796963,0.008431923],"genre_scores_gemma":[0.997892,0.000036020367,0.0005862086,0.00020489289,0.00045099406,4.6167239e-7,0.0000024290991,0.000003928377,0.00082305796],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99542254,0.001439201,0.0005770141,0.00014079991,0.0022079432,0.00021248781],"domain_scores_gemma":[0.99370056,0.001104919,0.00034073045,0.00014282174,0.0045277034,0.00018329071],"candidate_categories":["metaresearch","scholarly_communication","research_integrity"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.014912013,0.00006998151,0.00024153644,0.0011081652,0.0003124764,0.0024127413,0.0016145117,0.00045763867,0.00027545405],"category_scores_gemma":[0.011738339,0.000060970266,0.000068955625,0.0015224683,0.00013911873,0.030514542,0.00019440525,0.004861398,0.0000014940018],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00051881815,0.0007403609,0.8232947,0.0000061070527,0.00018779251,0.000027245267,0.13786744,0.000054916123,0.0056668967,0.0020496082,0.025267757,0.004318351],"study_design_scores_gemma":[0.0005799704,0.00029193578,0.62657565,0.00009953555,0.000022807335,0.0000012081442,0.355366,0.000054615848,0.00018806476,0.0050614025,0.011692849,0.00006597917],"about_ca_topic_score_codex":0.011340471,"about_ca_topic_score_gemma":0.0025850146,"teacher_disagreement_score":0.21749856,"about_ca_system_score_codex":0.00030875904,"about_ca_system_score_gemma":0.00084563525,"threshold_uncertainty_score":0.99862283},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7106478718","doi":"","title":"Hotspots and Frontiers in Arrhythmias During Pregnancy: A Bibliometric Analysis","year":2025,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Cardiovascular Issues in Pregnancy","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Web of science; Scientometrics; Principal (computer security); Pregnancy; Maternal health","score_opus":0.10550123067980056,"score_gpt":0.5047690590891654,"score_spread":0.3992678284093648,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7106478718","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6700306,0.32320455,0.0010534064,0.000110634544,0.00044654286,0.00064286584,0.000011687319,0.000033492262,0.0044662086],"genre_scores_gemma":[0.89654505,0.100902915,0.001085372,0.000074854506,0.00007507711,0.00006885805,0.0000070759647,0.000042252916,0.001198551],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99645954,0.00027530018,0.0010763281,0.0007476028,0.0009283697,0.0005128356],"domain_scores_gemma":[0.9977046,0.00020470504,0.00044695596,0.0010282618,0.00032598086,0.00028949248],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0010708066,0.00038153274,0.001978959,0.15641129,0.0001455074,0.0006681724,0.00110706,0.00019465452,0.0012405578],"category_scores_gemma":[0.00094907073,0.0003657649,0.0006696382,0.18413243,0.00016711757,0.0013466625,0.0011047962,0.00058362266,0.000003709917],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026359048,0.00015852989,0.91250527,0.0005616757,0.0039749146,0.0011850238,0.0001536129,0.0001769037,0.0026905576,0.000012002111,0.0020678707,0.07625002],"study_design_scores_gemma":[0.002099815,0.0000071759832,0.98395705,0.0041514044,0.0021722247,0.00012288922,0.00007863663,0.0001376125,0.0049744416,0.0009327405,0.0010455648,0.00032044837],"about_ca_topic_score_codex":0.00031357553,"about_ca_topic_score_gemma":0.000038079696,"teacher_disagreement_score":0.22651443,"about_ca_system_score_codex":0.00025020677,"about_ca_system_score_gemma":0.00017926692,"threshold_uncertainty_score":0.9998794},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7106673636","doi":"10.5281/zenodo.17715890","title":"Investigating Cognitive Profiles in Dyslexia, Dyscalculia, and Their Comorbidity: Bibliometric mapping","year":2025,"lang":"","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Reading and Literacy Development","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Cognition; Bridging (networking); Visualization; Field (mathematics); Cognitive map; Thematic analysis; Thematic map; Data visualization","score_opus":0.062201406650734904,"score_gpt":0.3091129236346217,"score_spread":0.24691151698388678,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7106673636","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6735123,0.0033983581,0.0032863226,0.0021495102,0.0006201548,0.0017266641,0.00042582664,0.00057950366,0.31430137],"genre_scores_gemma":[0.9929587,0.0006718298,0.00024845081,0.0005592484,0.000113051465,3.8016387e-7,0.0009222187,0.0008764265,0.0036496604],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9954345,0.0015329539,0.00081964117,0.0010473474,0.00034057623,0.00082493265],"domain_scores_gemma":[0.99799705,0.0003485379,0.00027273854,0.00044999772,0.0006473918,0.00028427618],"candidate_categories":["metaepi_narrow","bibliometrics","sts","scholarly_communication","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.0024541917,0.00038595084,0.00046227805,0.0186796,0.002366028,0.0017475723,0.000870103,0.00022397557,0.005476665],"category_scores_gemma":[0.00280217,0.00040767595,0.000067795925,0.034400214,0.0005807478,0.00030150777,0.0019086344,0.0009057999,0.0025598858],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000093852665,0.00046125916,0.0022301243,0.00065678015,0.00031473115,0.00004312969,0.030017687,0.000005689675,0.001803615,0.0053677107,0.059122775,0.8998827],"study_design_scores_gemma":[0.0022379272,0.00025919682,0.103119776,0.0017244989,0.000031293315,0.000112716836,0.0112646455,0.0009503212,0.0007303266,0.00092114834,0.8781135,0.000534659],"about_ca_topic_score_codex":0.00012945336,"about_ca_topic_score_gemma":8.5794295e-7,"teacher_disagreement_score":0.89934796,"about_ca_system_score_codex":0.0002672809,"about_ca_system_score_gemma":0.000020004803,"threshold_uncertainty_score":0.9998375},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7109213208","doi":"10.5539/jsd.v18n5p63","title":"Materials Used in the Construction of Artificial Reefs: A Bibliometric Review","year":2025,"lang":"","type":"article","venue":"Journal of Sustainable Development","topic":"Marine Sponges and Natural Products","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Fundação de Apoio à Pesquisa do Estado da Paraíba","keywords":"Scopus; Life-cycle assessment; Identification (biology); Thematic map; Originality; Conceptual framework; Scientific literature; Sustainability; Constructive","score_opus":0.012124565782091302,"score_gpt":0.2785728296968805,"score_spread":0.2664482639147892,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7109213208","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9081696,0.08569437,0.00008921342,0.004148542,0.00067472673,0.00073316455,0.0000024753547,0.0000010319162,0.000486866],"genre_scores_gemma":[0.94088405,0.056061972,0.0012462799,0.00040797488,0.00018564216,0.000009755399,0.000012403047,0.000009441268,0.0011824771],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9969211,0.0003162216,0.0017209951,0.00023502034,0.00042539675,0.00038125415],"domain_scores_gemma":[0.9972006,0.000048532576,0.0010857508,0.00028839664,0.0013287738,0.000047948928],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004513612,0.00021816368,0.0005987404,0.0068547036,0.000099729274,0.00006619908,0.00047250453,0.000170949,0.00007638231],"category_scores_gemma":[0.0012963975,0.00015291103,0.00012869136,0.01745809,0.00010400055,0.00002190175,0.00022437755,0.00026447637,8.967367e-7],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0036909804,0.0021976617,0.0121286195,0.056136873,0.0019529914,0.0010485044,0.0015456323,0.00004615341,0.26198837,0.012148011,0.04432112,0.60279506],"study_design_scores_gemma":[0.0016141075,0.00071723567,0.034040023,0.004170764,0.00030034216,0.00031063423,0.008045074,0.0000016441952,0.49055707,0.0010622476,0.45872712,0.0004537135],"about_ca_topic_score_codex":0.000019689673,"about_ca_topic_score_gemma":0.000004962741,"teacher_disagreement_score":0.60234135,"about_ca_system_score_codex":0.00014316555,"about_ca_system_score_gemma":0.0019588936,"threshold_uncertainty_score":0.8388034},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7110689416","doi":"","title":"Rehabilitación cardíaca en Latinoamérica: una revisión temática y análisis bibliométrico","year":2024,"lang":"es","type":"article","venue":"E-LIS Repository (University of Naples Federico II)","topic":"Cardiac Health and Mental Health","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Rehabilitation; Latin Americans; Work (physics); Bibliometrics; Scientific literature","score_opus":0.010722996787065588,"score_gpt":0.27420996056733965,"score_spread":0.2634869637802741,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7110689416","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8665193,0.04291147,0.00014790395,0.0073581035,0.0029519864,0.0017667516,0.00032006617,0.0005244286,0.077500045],"genre_scores_gemma":[0.9690219,0.015653132,0.0012943058,0.0002381971,0.0012794719,0.0000035012702,0.00008766397,0.00008947424,0.01233231],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9948466,0.0006551234,0.0009046349,0.0012834326,0.0012310925,0.0010791402],"domain_scores_gemma":[0.99591804,0.0009110356,0.00039041077,0.0011161134,0.00039459727,0.0012698339],"candidate_categories":["metaepi_narrow","sts"],"consensus_categories":[],"category_scores_codex":[0.001081347,0.0006087135,0.0015552272,0.0024052004,0.0014618884,0.00020291624,0.00055004685,0.00067375356,0.00048705097],"category_scores_gemma":[0.00024975397,0.00069033465,0.0010763642,0.006814277,0.0004872457,0.0006826164,0.0006402324,0.0010464309,0.00027808233],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0042956574,0.0034378276,0.1276486,0.105595216,0.0028874741,0.007987652,0.037239894,0.000029094881,0.03358228,0.0066618165,0.5527034,0.1179311],"study_design_scores_gemma":[0.0024144494,0.0034796938,0.14006151,0.009895815,0.0015833443,0.00070463703,0.022791645,0.00066531316,0.001098455,0.000037885195,0.81625503,0.0010122082],"about_ca_topic_score_codex":0.004181884,"about_ca_topic_score_gemma":0.000026704856,"teacher_disagreement_score":0.26355165,"about_ca_system_score_codex":0.0016226022,"about_ca_system_score_gemma":0.0021043213,"threshold_uncertainty_score":0.99983805},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7110829446","doi":"10.3389/feduc.2025.1610465.s011","title":"Table 2_Trends in neurology medical education: a bibliometric analysis (2000–2023).xlsx","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Impact factor; Neurology; Publishing; Table of contents; Web of science; MEDLINE; Trend analysis","score_opus":0.031406981771507476,"score_gpt":0.3426720156086018,"score_spread":0.3112650338370943,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7110829446","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000010948618,0.033043902,1.7849732e-7,0.00095928495,0.0012243587,0.0010013925,0.9606228,0.00016107199,0.0029760492],"genre_scores_gemma":[0.00013984466,0.00095101824,0.00001566704,0.004261549,0.0011731639,0.00412404,0.9834079,0.00017403325,0.0057527977],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9819437,0.0022040962,0.0035075753,0.0045913197,0.0046686986,0.003084592],"domain_scores_gemma":[0.9849917,0.0031749955,0.0021098584,0.005709269,0.0019329372,0.002081191],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"category_scores_codex":[0.0012603371,0.002237295,0.0041650343,0.6595162,0.0004108527,0.0009006908,0.007196561,0.00509759,0.98045444],"category_scores_gemma":[0.0885767,0.0026434667,0.0013248262,0.9033484,0.000114018505,0.00077208574,0.004089454,0.006470171,0.10813043],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017072096,0.0034263523,0.00087172876,0.0021490555,0.002511933,0.0012044678,0.000020746298,0.00023592095,9.143685e-8,0.0000013626646,0.9799409,0.009466726],"study_design_scores_gemma":[0.0013812538,0.00025079903,0.024019808,0.0051138937,0.0031131492,0.00010654855,0.00001533866,0.002444295,0.0000010364269,0.0000059452655,0.96187615,0.0016717977],"about_ca_topic_score_codex":0.0025196972,"about_ca_topic_score_gemma":0.013764163,"teacher_disagreement_score":0.87232405,"about_ca_system_score_codex":0.0010000154,"about_ca_system_score_gemma":0.045304965,"threshold_uncertainty_score":0.99903667},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7110890464","doi":"10.3389/feduc.2025.1610465.s005","title":"Table 6_Trends in neurology medical education: a bibliometric analysis (2000–2023).xlsx","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Impact factor; Neurology; Publishing; Table of contents; Web of science; MEDLINE; Trend analysis","score_opus":0.031406981771507476,"score_gpt":0.3426720156086018,"score_spread":0.3112650338370943,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7110890464","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000010890487,0.032787718,1.7481375e-7,0.0010451681,0.0012255744,0.0010004214,0.9608468,0.00016113347,0.0029221452],"genre_scores_gemma":[0.00013109023,0.0009133055,0.000015731353,0.0042760554,0.0011694746,0.004116614,0.9828054,0.00017396205,0.0063984203],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98194295,0.0022044473,0.0035092265,0.0045871977,0.0046706353,0.0030855332],"domain_scores_gemma":[0.98499835,0.0031666986,0.0021109118,0.005708829,0.0019335181,0.0020817015],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"category_scores_codex":[0.0012604985,0.0022377819,0.004165397,0.6600728,0.00041081797,0.000900167,0.007195771,0.005091857,0.9804599],"category_scores_gemma":[0.088501416,0.0026439296,0.001325604,0.90351224,0.00011389459,0.00077136897,0.0040874984,0.006464683,0.10820645],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017067319,0.0034047037,0.00083619443,0.0021280358,0.0024868953,0.0012045548,0.000021065505,0.00022798318,9.282518e-8,0.0000013657731,0.97990227,0.0096161505],"study_design_scores_gemma":[0.0013823493,0.00025094417,0.02320923,0.005095785,0.0031027147,0.00010671227,0.000015841606,0.002407124,0.0000010452908,0.000005937886,0.96274894,0.0016733439],"about_ca_topic_score_codex":0.0025173763,"about_ca_topic_score_gemma":0.013399967,"teacher_disagreement_score":0.8722535,"about_ca_system_score_codex":0.0010075314,"about_ca_system_score_gemma":0.045321006,"threshold_uncertainty_score":0.9990362},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7110935027","doi":"10.3389/fmed.2025.1523630.s004","title":"Table 4_Publications of systematic review and meta-analysis in the indexed anesthesia journals: a 10-year bibliometric analysis.xlsx","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Impact factor; Web of science; Scope (computer science); Bibliometrics; Medical literature; Citation database; Table (database)","score_opus":0.18282053133117146,"score_gpt":0.38226380325587594,"score_spread":0.19944327192470448,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7110935027","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[4.3765738e-7,0.40535277,9.079795e-7,0.0005492912,0.000002475781,0.004926326,0.5890849,0.000021832213,0.00006106152],"genre_scores_gemma":[0.00029057372,0.014085459,0.00004677467,0.0011086445,0.000023123326,0.010432963,0.9730575,0.0000842989,0.0008706367],"study_design_codex":"not_applicable","study_design_gemma":"meta_analysis","domain_scores_codex":[0.97335386,0.010175197,0.008271128,0.0023184353,0.004567619,0.0013137771],"domain_scores_gemma":[0.9612404,0.00736116,0.013851341,0.010634603,0.0063687004,0.000543796],"candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"category_scores_codex":[0.012792572,0.0019655793,0.021609576,0.45507026,0.0005279533,0.002756492,0.008085012,0.0010412888,0.6798733],"category_scores_gemma":[0.06542073,0.00132545,0.008906143,0.8850105,0.00008854467,0.0012976235,0.00095668813,0.0020655235,0.0035490713],"study_design_candidate":"meta_analysis","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004243492,0.00034130068,0.000014064872,0.25107238,0.2617159,0.000028277565,0.000021916194,0.00009253761,5.9461517e-8,0.000001512478,0.48670736,4.6946008e-7],"study_design_scores_gemma":[0.00026130446,0.000048930207,0.0011813556,0.044863958,0.83393866,0.000065511274,0.00012571285,0.00022293522,5.2789756e-7,0.000004023508,0.11843633,0.00085074577],"about_ca_topic_score_codex":0.0002473143,"about_ca_topic_score_gemma":0.00064496207,"teacher_disagreement_score":0.67632425,"about_ca_system_score_codex":0.0003062944,"about_ca_system_score_gemma":0.0014613749,"threshold_uncertainty_score":0.9993087},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7111018522","doi":"10.3389/feduc.2025.1610465.s001","title":"Table 1_Trends in neurology medical education: a bibliometric analysis (2000–2023).xlsx","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Impact factor; Neurology; Publishing; Table of contents; Web of science; MEDLINE; Trend analysis","score_opus":0.031406981771507476,"score_gpt":0.3426720156086018,"score_spread":0.3112650338370943,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7111018522","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000011267077,0.03304288,1.744456e-7,0.0009569125,0.0012256334,0.0010000918,0.96057916,0.00016107493,0.0030227988],"genre_scores_gemma":[0.00014113172,0.0009544068,0.000015744343,0.004326136,0.0011705472,0.004119195,0.9834337,0.00017393717,0.0056652008],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98193985,0.0022046482,0.003509484,0.0045898324,0.0046710144,0.0030851853],"domain_scores_gemma":[0.9849963,0.0031678828,0.002110782,0.0057092505,0.0019339203,0.0020818557],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"category_scores_codex":[0.0012616977,0.0022376815,0.004164944,0.65969944,0.0004110686,0.0009008322,0.007203888,0.005092543,0.98044986],"category_scores_gemma":[0.08851486,0.0026438169,0.0013254998,0.90335125,0.00011406998,0.0007719415,0.0040880903,0.0064652325,0.107423194],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017050204,0.0034040294,0.0008449018,0.002127538,0.0024817458,0.0012042186,0.000020972919,0.0002281524,9.269399e-8,0.0000014030448,0.9796921,0.009824335],"study_design_scores_gemma":[0.001381154,0.0002508837,0.023496313,0.0050872136,0.0031045037,0.00010668772,0.000015905089,0.0024040404,0.0000010449398,0.0000060348343,0.9624732,0.001673011],"about_ca_topic_score_codex":0.0025341392,"about_ca_topic_score_gemma":0.013955396,"teacher_disagreement_score":0.87302667,"about_ca_system_score_codex":0.0010005645,"about_ca_system_score_gemma":0.045348484,"threshold_uncertainty_score":0.99903625},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7111021763","doi":"10.3389/feduc.2025.1610465.s006","title":"Table 7_Trends in neurology medical education: a bibliometric analysis (2000–2023).xlsx","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Impact factor; Neurology; Publishing; Table of contents; Web of science; MEDLINE; Trend analysis","score_opus":0.031406981771507476,"score_gpt":0.3426720156086018,"score_spread":0.3112650338370943,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7111021763","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000010243398,0.031220352,1.763055e-7,0.0010072613,0.0011965153,0.0009944399,0.96224254,0.00016087483,0.0031675966],"genre_scores_gemma":[0.00013575473,0.00091653975,0.0000150311425,0.004409342,0.0011505787,0.0040457956,0.98306316,0.0001734735,0.006090313],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.982098,0.002115586,0.0034985216,0.004569787,0.004650766,0.003067381],"domain_scores_gemma":[0.98502225,0.0031488102,0.0021058293,0.0056850263,0.0019660261,0.002072053],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"category_scores_codex":[0.0012319296,0.002229904,0.004144171,0.6588987,0.0004015116,0.00093810423,0.0071738125,0.005025447,0.97977114],"category_scores_gemma":[0.08986474,0.0026345807,0.001320475,0.9031083,0.00011322127,0.00080138264,0.004093996,0.006357464,0.108224854],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016990937,0.0034044243,0.00081046857,0.0021061793,0.0024728049,0.0011935318,0.000020073245,0.00022752788,8.536878e-8,0.0000013534759,0.97962326,0.009970413],"study_design_scores_gemma":[0.0013709017,0.00024971337,0.022681471,0.005097619,0.0030207096,0.00010718968,0.000015514619,0.0023623302,0.0000010171755,0.0000061397277,0.9634192,0.0016681791],"about_ca_topic_score_codex":0.0024561337,"about_ca_topic_score_gemma":0.013268247,"teacher_disagreement_score":0.87154627,"about_ca_system_score_codex":0.0009915358,"about_ca_system_score_gemma":0.04495946,"threshold_uncertainty_score":0.99904406},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7111028582","doi":"10.3389/fonc.2025.1572725.s007","title":"Table 8_Quality of life in patients with pan-cancer undergoing concurrent chemoradiotherapy: a bibliometric analysis (1995-2024).docx","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Quality of life (healthcare); Cervical cancer; Bibliometrics; Chemoradiotherapy; Cancer; Clinical Oncology; Affect (linguistics); MEDLINE; Radiation oncology","score_opus":0.045196266891149466,"score_gpt":0.32961440079121024,"score_spread":0.28441813390006077,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7111028582","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0030241404,0.022470945,0.0000031652824,0.000018803361,0.00042857655,0.0032969534,0.9705937,0.00008966365,0.000074068026],"genre_scores_gemma":[0.022290038,0.0013988551,0.000037813403,0.00027003913,0.0001994884,0.0020204484,0.97326595,0.00025492377,0.0002624678],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9834974,0.0014519605,0.004393067,0.0037048105,0.0045443005,0.0024084728],"domain_scores_gemma":[0.9820423,0.0027923717,0.006931122,0.0040936377,0.0032349168,0.0009056282],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"category_scores_codex":[0.0009772005,0.0026381854,0.007152533,0.25809753,0.00031354363,0.000749633,0.003360323,0.0014256835,0.6310267],"category_scores_gemma":[0.01068989,0.0026375235,0.0012231249,0.6858607,0.00014089451,0.0011507471,0.0014034908,0.0024924967,0.0019978038],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00082721637,0.003408663,0.1146035,0.016031021,0.010962853,0.000017219258,0.00008313278,0.0019679822,9.2294937e-7,4.0456027e-7,0.85014683,0.0019502726],"study_design_scores_gemma":[0.013113919,0.0006800802,0.34232062,0.039229598,0.0057645403,3.679427e-7,0.00008997166,0.0010837516,0.000043389082,0.0000017502507,0.594444,0.0032280274],"about_ca_topic_score_codex":0.0028528802,"about_ca_topic_score_gemma":0.0060952087,"teacher_disagreement_score":0.62902886,"about_ca_system_score_codex":0.0028289754,"about_ca_system_score_gemma":0.0087476615,"threshold_uncertainty_score":0.99987066},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7111037167","doi":"10.3389/fmed.2025.1523630.s001","title":"Table 1_Publications of systematic review and meta-analysis in the indexed anesthesia journals: a 10-year bibliometric analysis.docx","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Citation; Impact factor; Web of science; Scope (computer science); Bibliometrics; Medical literature; Citation database; Table (database)","score_opus":0.16549425368778753,"score_gpt":0.37714146217170746,"score_spread":0.21164720848391994,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7111037167","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[4.4436115e-7,0.40421888,7.739879e-7,0.00052285165,0.0000024850572,0.0048719826,0.59029716,0.000021850625,0.00006360846],"genre_scores_gemma":[0.00029173418,0.014899607,0.000046774036,0.001131137,0.000022759372,0.01035084,0.9722932,0.00008353217,0.0008803894],"study_design_codex":"not_applicable","study_design_gemma":"meta_analysis","domain_scores_codex":[0.9734794,0.010076022,0.00824835,0.0023165348,0.0045592887,0.0013204159],"domain_scores_gemma":[0.96170545,0.0071250047,0.013718607,0.010573275,0.006335241,0.00054243533],"candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"category_scores_codex":[0.012492619,0.0019636524,0.02145949,0.45772702,0.00051914545,0.0027259125,0.008082715,0.0010070469,0.7591459],"category_scores_gemma":[0.061323166,0.0013247721,0.008883723,0.88375866,0.00008718918,0.0013195849,0.00097415,0.0020651724,0.003517841],"study_design_candidate":"meta_analysis","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000040112945,0.0003335447,0.0000140056745,0.25120464,0.2566081,0.000030603944,0.00002120943,0.00009122475,5.2609508e-8,0.0000016893636,0.49169046,4.7622393e-7],"study_design_scores_gemma":[0.00024206884,0.000047674665,0.001354496,0.04301358,0.8291037,0.00006870597,0.00012294001,0.00020569927,5.23638e-7,0.000004266386,0.12499142,0.00084491074],"about_ca_topic_score_codex":0.00024233735,"about_ca_topic_score_gemma":0.0005797365,"teacher_disagreement_score":0.75562805,"about_ca_system_score_codex":0.00030491853,"about_ca_system_score_gemma":0.0014369661,"threshold_uncertainty_score":0.9993107},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7111173119","doi":"10.3389/feduc.2025.1610465.s004","title":"Table 5_Trends in neurology medical education: a bibliometric analysis (2000–2023).xlsx","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Impact factor; Neurology; Publishing; Table of contents; Web of science; MEDLINE; Trend analysis","score_opus":0.031406981771507476,"score_gpt":0.3426720156086018,"score_spread":0.3112650338370943,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7111173119","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000012127283,0.03261718,1.535072e-7,0.0009436131,0.0012232485,0.000999857,0.9610648,0.00016111041,0.0029778732],"genre_scores_gemma":[0.00014178763,0.00095412007,0.000014481912,0.004286363,0.0011718207,0.0024821581,0.98507965,0.00017433707,0.005695313],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98195344,0.0022035744,0.003507457,0.004585933,0.0046666237,0.0030829874],"domain_scores_gemma":[0.98501277,0.0031634173,0.0021100356,0.005702436,0.0019327839,0.002078544],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"category_scores_codex":[0.0012598,0.0022367828,0.004162714,0.6599335,0.00041073872,0.00090014975,0.007201124,0.0050907424,0.9806677],"category_scores_gemma":[0.08841009,0.0026432557,0.0013253926,0.90329474,0.00011403191,0.0007714955,0.0040845317,0.0064626355,0.10834258],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017053697,0.0034076918,0.0008345386,0.0021285673,0.0024885992,0.001205176,0.00002079853,0.00031340178,9.291019e-8,0.0000013745668,0.98081243,0.008616794],"study_design_scores_gemma":[0.0013850613,0.00025190154,0.02030864,0.005102753,0.0031102747,0.00010686677,0.000015920306,0.0027732588,0.0000010388356,0.0000060574835,0.965261,0.001677219],"about_ca_topic_score_codex":0.0025139998,"about_ca_topic_score_gemma":0.011577271,"teacher_disagreement_score":0.8723252,"about_ca_system_score_codex":0.0010021658,"about_ca_system_score_gemma":0.04530805,"threshold_uncertainty_score":0.9990372},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7111188678","doi":"10.3389/fpsyt.2025.1715886.s001","title":"Supplementary file 1_Pathological fixation on shared beliefs: a review and bibliometric analysis of extreme, overvalued and delusion-like beliefs.docx","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Terminology; Consistency (knowledge bases); Bibliometrics; Credibility; Terrorism; Web of science","score_opus":0.07335323789238823,"score_gpt":0.3298037323781109,"score_spread":0.25645049448572266,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7111188678","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000627545,0.11324083,3.6668064e-7,0.00013220735,0.00008102293,0.0041624284,0.8821797,0.00006493451,0.00007580756],"genre_scores_gemma":[0.000058219357,0.057495177,0.0003349461,0.0028457274,0.00012499407,0.002204433,0.9365416,0.00012132589,0.00027354457],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9861147,0.0015972136,0.0038206687,0.004001311,0.002969553,0.0014965677],"domain_scores_gemma":[0.9857339,0.0040693926,0.004209813,0.0034280033,0.0017148879,0.0008440063],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"category_scores_codex":[0.0012840978,0.002333448,0.0054506105,0.1603098,0.0007719071,0.0005068554,0.0018924609,0.0014735818,0.9508723],"category_scores_gemma":[0.020694965,0.0022828006,0.0012712326,0.33918414,0.0001477908,0.0007055093,0.004865165,0.0016593408,0.0017523376],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001790303,0.0010668065,0.00016983865,0.026074665,0.003462805,0.0001984692,0.000045598146,0.000029496632,0.000017170327,6.6121464e-7,0.965357,0.0033984713],"study_design_scores_gemma":[0.0017221115,0.0010921805,0.07881948,0.073084295,0.019546108,0.000030834457,0.000025517702,0.0007590674,0.000008961017,0.000005323986,0.8232177,0.0016884056],"about_ca_topic_score_codex":0.00029760817,"about_ca_topic_score_gemma":0.00043812676,"teacher_disagreement_score":0.9491199,"about_ca_system_score_codex":0.0005739802,"about_ca_system_score_gemma":0.0005850189,"threshold_uncertainty_score":0.99982274},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7111383292","doi":"10.3389/fnut.2025.1680190.s002","title":"Data Sheet 2_Mapping the research trends and hotspots of exercise and nutrition in diabetes: a bibliometric and visual analysis (2005–2025).csv","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Fact sheet; Thematic analysis; Bibliometrics; Psychological intervention; Physical activity; Web of science","score_opus":0.16317129896041008,"score_gpt":0.4197868873408182,"score_spread":0.2566155883804081,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7111383292","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0020551274,0.106562264,1.4394617e-7,0.00014157881,0.000035107685,0.0018254308,0.8893103,0.00003227241,0.00003775354],"genre_scores_gemma":[0.013667095,0.025924986,0.000029283856,0.000028249886,0.00013712898,0.001262561,0.9587417,0.00009434439,0.0001146065],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9866965,0.0023210547,0.0021429232,0.003810421,0.002835783,0.0021933392],"domain_scores_gemma":[0.9860742,0.0067375763,0.0011599578,0.0041395607,0.0012064725,0.0006822508],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics"],"category_scores_codex":[0.0060253176,0.0013316484,0.0030039628,0.36981958,0.0008182556,0.00207286,0.0032290444,0.0012727504,0.051997088],"category_scores_gemma":[0.015734604,0.00123822,0.00023729549,0.47587577,0.000499569,0.0017458806,0.012133955,0.0028021168,0.00015304252],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003445097,0.0009532009,0.0011481052,0.015467574,0.0013600504,0.00007227597,0.000101778554,0.0000075757152,0.000019987005,2.6258672e-7,0.9513112,0.029213486],"study_design_scores_gemma":[0.0040803878,0.00061667897,0.17792225,0.09392373,0.005476701,0.000010423874,0.0010847778,0.041893374,0.00003409671,0.00002759391,0.67295706,0.0019729172],"about_ca_topic_score_codex":0.00045761603,"about_ca_topic_score_gemma":0.0016345731,"teacher_disagreement_score":0.27835414,"about_ca_system_score_codex":0.00030516007,"about_ca_system_score_gemma":0.00064514927,"threshold_uncertainty_score":0.9999435},"labels":[{"model":"gemma","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7111395803","doi":"10.3389/fonc.2025.1572725.s006","title":"Table 6_Quality of life in patients with pan-cancer undergoing concurrent chemoradiotherapy: a bibliometric analysis (1995-2024).docx","year":2025,"lang":"","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Quality of life (healthcare); Cervical cancer; Bibliometrics; Chemoradiotherapy; Cancer; Clinical Oncology; Affect (linguistics); MEDLINE; Radiation oncology","score_opus":0.045196266891149466,"score_gpt":0.32961440079121024,"score_spread":0.28441813390006077,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7111395803","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002954671,0.02256081,0.0000031594122,0.000018683855,0.00042894314,0.003287707,0.9705815,0.000089755806,0.00007477834],"genre_scores_gemma":[0.02248942,0.001384603,0.00003782345,0.00026923255,0.00019952834,0.0020134638,0.97309226,0.000254901,0.00025878227],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98349744,0.0014522474,0.004392624,0.003704648,0.0045438656,0.0024091804],"domain_scores_gemma":[0.9820257,0.0028059385,0.0069368966,0.0040916875,0.0032337816,0.0009060055],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"category_scores_codex":[0.00097615353,0.0026385833,0.0071531646,0.2580745,0.00031513147,0.00075039477,0.0033611155,0.0014259134,0.6296723],"category_scores_gemma":[0.010744286,0.0026382182,0.0012209833,0.6857167,0.00014101167,0.0011531058,0.0014012638,0.0024627086,0.0019633034],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008270884,0.0034064555,0.11459712,0.016019523,0.010970356,0.000017226736,0.00008403614,0.0019253854,9.172291e-7,4.1748842e-7,0.8501908,0.0019606773],"study_design_scores_gemma":[0.013114478,0.00068345416,0.34573817,0.039242115,0.005770128,3.6815052e-7,0.0000893974,0.0010499023,0.000043088377,0.0000017725741,0.5910371,0.0032300004],"about_ca_topic_score_codex":0.0028272162,"about_ca_topic_score_gemma":0.006329863,"teacher_disagreement_score":0.627709,"about_ca_system_score_codex":0.0028274679,"about_ca_system_score_gemma":0.008761638,"threshold_uncertainty_score":0.9998704},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7115034731","doi":"","title":"Research Trends of Health Professions Education Model from 2005 to 2024: A Bibliometric Analysis via CiteSpace","year":2025,"lang":"en","type":"article","venue":"DOAJ (DOAJ: Directory of Open Access Journals)","topic":"Innovations in Medical Education","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Health professions; Frontier; Publishing; Professional development; Yearbook; Bibliometrics; Continuing medical education; Web of science; Citation","score_opus":0.36820731279908686,"score_gpt":0.6832825968053261,"score_spread":0.31507528400623924,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7115034731","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9016752,0.019037567,0.03770227,0.02749399,0.0013669257,0.0012877453,0.00007471747,0.000035524103,0.011326054],"genre_scores_gemma":[0.96979433,0.003403695,0.008394502,0.0021078787,0.00019038873,0.00022755563,0.00018367758,0.000034795452,0.015663197],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9949805,0.0006085502,0.0015886935,0.00055580284,0.0017455902,0.00052087754],"domain_scores_gemma":[0.99468297,0.00052922167,0.0008382027,0.0010070686,0.002593735,0.00034880752],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0066479724,0.000227633,0.0011058169,0.20979513,0.00037088146,0.00027740299,0.0015515449,0.00017719633,0.010703951],"category_scores_gemma":[0.003363551,0.00020335967,0.00020707742,0.3481768,0.00015408536,0.0007751841,0.00063739694,0.00094948563,0.000013794052],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013252799,0.0013515553,0.1931893,0.00015047171,0.00057505287,7.999011e-7,0.00043041445,0.00043842968,0.0075002178,0.00014757422,0.5023783,0.29370537],"study_design_scores_gemma":[0.00045293127,0.00002404016,0.97781986,0.0013004966,0.0004434832,0.0000010563588,0.00089008594,0.004670846,0.0020410756,0.0036369185,0.008540344,0.00017884385],"about_ca_topic_score_codex":0.004320283,"about_ca_topic_score_gemma":0.00015099908,"teacher_disagreement_score":0.7846306,"about_ca_system_score_codex":0.00066094415,"about_ca_system_score_gemma":0.0035745464,"threshold_uncertainty_score":0.9902004},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7115717398","doi":"10.48448/1dme-6p77","title":"Retraction Prevalence and Gender Imbalance Among Highly-Cited Authors and Among All Authors Across Scientific Disciplines","year":2025,"lang":"","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Demographics; Gender gap; MEDLINE; Tracking (education); Propensity score matching","score_opus":0.030923282250168212,"score_gpt":0.34172130542734647,"score_spread":0.31079802317717825,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7115717398","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96275324,0.004040055,0.0023573318,0.0005530153,0.013968473,0.0042320723,0.0021360847,0.0009854629,0.008974274],"genre_scores_gemma":[0.75459707,0.00094343664,0.0038595796,0.0000615267,0.00062689226,0.00011158,0.00018662498,0.0005542084,0.23905909],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.977718,0.00086980057,0.0026834093,0.008983424,0.0052379323,0.0045074597],"domain_scores_gemma":[0.9882348,0.00062139955,0.0031461022,0.0034110963,0.0022032058,0.002383388],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","research_integrity"],"consensus_categories":["metaepi_narrow","sts","research_integrity"],"category_scores_codex":[0.016099006,0.0026207604,0.0019774341,0.00495946,0.0067598796,0.0090794945,0.0037054524,0.0024815698,0.0007472972],"category_scores_gemma":[0.0034666285,0.0025500364,0.0002879756,0.013239201,0.052134696,0.005978046,0.0041168653,0.0037326633,0.0004493552],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00045244434,0.0016980295,0.87808365,0.006416286,0.00079399394,0.00020788745,0.015378931,0.0014986935,0.049460456,0.00520785,0.023287212,0.01751456],"study_design_scores_gemma":[0.0017374097,0.00024106105,0.82202595,0.0031247924,0.0008459949,0.00008512374,0.0015287759,0.15098341,0.0012301913,0.0027936406,0.011917213,0.0034864105],"about_ca_topic_score_codex":0.0007324697,"about_ca_topic_score_gemma":0.00330804,"teacher_disagreement_score":0.2300848,"about_ca_system_score_codex":0.0014684665,"about_ca_system_score_gemma":0.0026568004,"threshold_uncertainty_score":0.9988134},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7116096669","doi":"10.82417/hqz4-8z48","title":"A bibliometric examination of the existing knowledge framework and advancements in vascular network engineering","year":2025,"lang":"en","type":"other","venue":"Espace ÉTS (ETS)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; York University","keywords":"Vascular network; Vascular tissue; Tissue engineering; Science and engineering; Knowledge engineering","score_opus":0.015869036701659914,"score_gpt":0.2804067219136181,"score_spread":0.2645376852119582,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7116096669","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012220875,0.17374513,0.021068634,0.000073262614,0.0044646794,0.0044835876,0.00019122823,0.0009037812,0.78284883],"genre_scores_gemma":[0.27437702,0.005170329,0.040119052,0.00006433237,0.00210979,0.00045693252,0.000084369705,0.0037499615,0.67386824],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99789715,0.00023543974,0.0003599448,0.00053883606,0.00048431536,0.0004843139],"domain_scores_gemma":[0.998233,0.00047990092,0.0004266103,0.00070799113,0.00008835803,0.00006414207],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0010858794,0.00042836304,0.0005756973,0.03103832,0.000051348878,0.00005146478,0.00041853936,0.0004489064,0.00018878697],"category_scores_gemma":[0.0019334217,0.0003907386,0.000105517924,0.08882964,0.00006086903,0.00008067759,0.00039871427,0.0006058946,0.00009602434],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014939865,0.0018385913,0.13076054,0.015400729,0.0034203238,0.0000739209,0.006951096,0.06528157,0.0014372282,0.03887083,0.39253253,0.34328327],"study_design_scores_gemma":[0.0013116278,0.00004231571,0.18972272,0.016944561,0.00022083851,0.0000014851173,0.00009949063,0.005448433,0.00007477209,0.00017595539,0.7850659,0.0008919413],"about_ca_topic_score_codex":0.00005961128,"about_ca_topic_score_gemma":0.00019249247,"teacher_disagreement_score":0.39253333,"about_ca_system_score_codex":0.00021413504,"about_ca_system_score_gemma":0.000083574145,"threshold_uncertainty_score":0.99985445},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7116920689","doi":"10.4103/joah.joah_101_25","title":"Global Overview of Von Willebrand Disease Research: A Bibliometric Study","year":2025,"lang":"en","type":"article","venue":"Journal of Applied Hematology","topic":"Platelet Disorders and Treatments","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scopus; Erasmus+; Haemophilia; Centrality; Citation impact; Von Willebrand disease; Bibliometrics; China","score_opus":0.08705347988076538,"score_gpt":0.4268997089711973,"score_spread":0.3398462290904319,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7116920689","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9799708,0.010795409,0.000024278632,0.0007120045,0.00009506294,0.00045986252,0.000005672497,0.0000055739783,0.0079313405],"genre_scores_gemma":[0.9986658,0.0009576249,0.00015049436,0.00013328766,0.000017485278,0.00000927719,0.0000016815892,0.000005937451,0.00005837721],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99852824,0.0000626463,0.000565251,0.00014491662,0.00046724916,0.00023167432],"domain_scores_gemma":[0.9988954,0.00018024238,0.0002297957,0.00021746887,0.00030388148,0.00017320523],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00057592575,0.00010913941,0.00079896726,0.00940691,0.000049324444,0.000009177321,0.00013933334,0.00006976697,0.000058055863],"category_scores_gemma":[0.0001836411,0.000078246696,0.00012650785,0.019185992,0.00009597248,0.000036533194,0.00007147527,0.00019712752,0.00000880494],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.008672882,0.0091411965,0.9087992,0.0010819137,0.001980358,0.005427792,0.00025420633,0.000009978974,0.00009032501,0.03079198,0.013834973,0.019915238],"study_design_scores_gemma":[0.025726745,0.0025716678,0.90697783,0.0006630822,0.001368882,0.0010473174,0.0020452687,0.000025209167,0.0003686872,0.0537024,0.005328369,0.00017453133],"about_ca_topic_score_codex":0.0000110318815,"about_ca_topic_score_gemma":0.0000073272545,"teacher_disagreement_score":0.02291042,"about_ca_system_score_codex":0.0000699308,"about_ca_system_score_gemma":0.0003395707,"threshold_uncertainty_score":0.92182326},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7117159665","doi":"10.69665/iss.v47i2.98","title":"Studying Baduanjin Qigong: A Bibliometric Analysis of Research Trends and Future Prospects","year":2025,"lang":"","type":"article","venue":"International Sports Studies","topic":"Biofield Effects and Biophysics","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Multidisciplinary approach; Scope (computer science); China; Mental health; Bibliometrics; Conceptual framework; Health care","score_opus":0.06344284276986398,"score_gpt":0.43244047367763616,"score_spread":0.36899763090777216,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7117159665","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9104502,0.05616192,0.000011743872,0.012240661,0.0025879268,0.00036890773,0.00006761081,0.000025731628,0.018085249],"genre_scores_gemma":[0.95994604,0.029622713,0.0000850965,0.00013668292,0.0009561459,0.000026398837,0.000024829264,0.000010231281,0.009191877],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99682605,0.00006916366,0.0006894047,0.00060546654,0.0014329734,0.00037691975],"domain_scores_gemma":[0.99704003,0.00051092356,0.00024276228,0.0003207303,0.0017970481,0.0000884772],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0013210019,0.00027137433,0.000963719,0.10269197,0.00023271677,0.000071856375,0.00022670954,0.00016626608,0.00029887867],"category_scores_gemma":[0.00036863371,0.00021131882,0.00027973513,0.17524244,0.0004913377,0.00011788693,0.0005119022,0.00047620817,0.0000013026998],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033389468,0.0007405525,0.6690327,0.0005788288,0.023121921,0.00014992105,0.0017927552,0.000007180357,0.00032388914,0.006693942,0.00491066,0.29231375],"study_design_scores_gemma":[0.00091688475,0.00032625065,0.97704786,0.00058939547,0.003095701,0.0000026306661,0.004284074,0.00055000704,0.0007670621,0.0003343777,0.011920811,0.00016493452],"about_ca_topic_score_codex":0.00014632828,"about_ca_topic_score_gemma":0.00006739586,"teacher_disagreement_score":0.30801517,"about_ca_system_score_codex":0.00016756047,"about_ca_system_score_gemma":0.00009308115,"threshold_uncertainty_score":0.9074783},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7117243073","doi":"10.5206/cjils-rcsib.v49i1.23107","title":"Citizen Science in Libraries: A Co-Citation Analysis","year":2025,"lang":"en","type":"article","venue":"Canadian Journal of Information and Library Science","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Citizen science; Field (mathematics); Citizen journalism; Thematic analysis; Relevance (law); Public awareness of science; Open science; Science communication; Sociology of scientific knowledge; Funding Agency","score_opus":0.007878302844152635,"score_gpt":0.21583594867534028,"score_spread":0.20795764583118764,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7117243073","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8155446,0.00003924909,0.00032062482,0.0017049959,0.00013473192,0.000058871075,0.000020231653,0.0000058871574,0.18217078],"genre_scores_gemma":[0.99779016,0.000026141133,0.00027986296,0.001740552,0.000003504739,9.0697284e-7,0.000008123102,8.233294e-7,0.0001499577],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99907494,0.00001247274,0.0003135046,0.00008233225,0.0002848177,0.00023191271],"domain_scores_gemma":[0.99935347,0.000023053723,0.0001355349,0.00009008756,0.000033195516,0.0003646415],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0006232184,0.00005722393,0.00009655821,0.0018705556,0.00033446506,0.00070703734,0.00046702122,0.000023786726,0.006862933],"category_scores_gemma":[0.00020606472,0.00005093612,0.000028443306,0.0076729385,0.0013763667,0.011495205,0.00004795755,0.0000858039,0.00004076677],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000015222299,0.000009025007,0.9181661,0.000010517544,0.000008403091,0.000006577307,0.0039337017,0.00039120408,0.00033439242,0.053276923,0.010467116,0.013380806],"study_design_scores_gemma":[0.00018115269,0.000019690693,0.9560236,0.00001269413,0.000007860009,0.0000058453425,0.0040610614,0.0009633454,0.0005757678,0.0005719718,0.037510466,0.00006657505],"about_ca_topic_score_codex":0.00055878214,"about_ca_topic_score_gemma":0.0009384137,"teacher_disagreement_score":0.1822455,"about_ca_system_score_codex":0.00027974375,"about_ca_system_score_gemma":0.0012043914,"threshold_uncertainty_score":0.9940449},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W7117252741","doi":"10.32038/ltrq.2025.52.03","title":"A Bibliometric Analysis of Practicum Research in Preservice Teacher Education: Trends, Themes, and Future Directions (2006–2024)","year":2025,"lang":"","type":"article","venue":"Language Teaching Research Quarterly","topic":"Teacher Education and Leadership Studies","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Practicum; Thematic analysis; Identity (music); Citation; Teacher education; Quality (philosophy); Scopus","score_opus":0.12267931180598009,"score_gpt":0.5151390836230951,"score_spread":0.39245977181711506,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7117252741","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5799218,0.117373325,0.000009804991,0.16075192,0.0013931191,0.0009905687,0.000045836605,0.00008348903,0.13943017],"genre_scores_gemma":[0.9266254,0.0023262165,0.00019652232,0.00009823418,0.0011038838,0.00026311376,0.000029435138,0.00002627639,0.06933096],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9794993,0.014486911,0.00088419893,0.0011405256,0.0024722505,0.0015168618],"domain_scores_gemma":[0.99264145,0.004608292,0.00023940996,0.0011113355,0.0010354063,0.00036408476],"candidate_categories":["metaepi_narrow","bibliometrics","sts","research_integrity","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.024937244,0.00033782015,0.0007806908,0.21557109,0.0017534508,0.00084315153,0.0011200968,0.0004205287,0.0017146055],"category_scores_gemma":[0.0026816137,0.00032719845,0.00023718369,0.3849334,0.0012058177,0.0008200697,0.00020961573,0.0048827194,0.000022607817],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004890681,0.0018517248,0.038435195,0.00018309703,0.0005845212,0.00000421785,0.49810243,4.8714173e-7,0.000029853536,0.003205356,0.010448132,0.4471061],"study_design_scores_gemma":[0.0002501308,0.00019214794,0.2338337,0.00015911003,0.00023782717,6.6253233e-7,0.650208,0.00009658251,0.0000018269972,0.00032143283,0.11452428,0.00017428005],"about_ca_topic_score_codex":0.11033073,"about_ca_topic_score_gemma":0.04945587,"teacher_disagreement_score":0.4469318,"about_ca_system_score_codex":0.00078581006,"about_ca_system_score_gemma":0.0020744645,"threshold_uncertainty_score":0.999918},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7118522923","doi":"10.30997/ijsr.v7i3.746","title":"Mapping the landscape of institutional village-owned enterprises: A bibliometric analysis of literature from 2005 to 2024","year":2025,"lang":"","type":"article","venue":"Indonesian Journal of Social Research (IJSR)","topic":"Local Governance and Development","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Government (linguistics); Scopus; Thematic analysis; Work (physics); Institutionalisation; Inclusion (mineral); Independence (probability theory); Meaning (existential)","score_opus":0.046999001968228905,"score_gpt":0.37517122031271033,"score_spread":0.3281722183444814,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7118522923","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9564768,0.012029216,0.0021727057,0.014543257,0.0013042674,0.0005840854,0.00040243714,0.0000055250975,0.012481675],"genre_scores_gemma":[0.9923899,0.0049134917,0.0002199622,0.00025156297,0.0008088827,0.000011746468,0.000008323219,0.000011406595,0.0013847048],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9900188,0.0014916385,0.0019342763,0.000417046,0.0051913545,0.0009469227],"domain_scores_gemma":[0.9926015,0.0018491182,0.0011463094,0.00030476993,0.0036812231,0.00041707986],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.009498207,0.00028783834,0.0012255888,0.06343087,0.0012559415,0.00051920663,0.0020001486,0.0004133077,0.0010287985],"category_scores_gemma":[0.0034438844,0.0002101346,0.0009939674,0.291906,0.0011819841,0.0005628431,0.00052833423,0.0013751892,0.00001683576],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.004156285,0.0020937838,0.42224798,0.0005100391,0.020835439,0.0005145444,0.26999268,0.00066540035,0.0029411411,0.017018246,0.082812384,0.17621209],"study_design_scores_gemma":[0.0014511744,0.00024139025,0.9052936,0.0018426493,0.0005494719,0.0000021902154,0.034777276,0.00008265105,0.00034579288,0.0017295004,0.05340598,0.0002783234],"about_ca_topic_score_codex":0.001004305,"about_ca_topic_score_gemma":0.0005112687,"teacher_disagreement_score":0.4830456,"about_ca_system_score_codex":0.0006844856,"about_ca_system_score_gemma":0.0056178044,"threshold_uncertainty_score":0.99988437},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7119239106","doi":"","title":"Práticas de pesquisas bibliométricas em tópicos de ciências farmacêuticas","year":2022,"lang":"pt","type":"book","venue":"LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)","topic":"Food composition and properties","field":"Nursing","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Order (exchange); Software; Identification (biology); Control (management)","score_opus":0.0366002389755584,"score_gpt":0.27236672541055273,"score_spread":0.23576648643499434,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7119239106","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.028155928,0.013044969,0.0035678397,0.07191602,0.013466247,0.0060979794,0.0017429405,0.0037411465,0.85826695],"genre_scores_gemma":[0.9613804,0.0011399054,0.0034546829,0.0033218078,0.003437982,0.0012251307,0.0018860772,0.0006288168,0.02352518],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9828828,0.0038798286,0.0028979226,0.002892382,0.0034477408,0.0039993105],"domain_scores_gemma":[0.99032867,0.0015096398,0.0018267866,0.0023537949,0.0013182295,0.0026628703],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","research_integrity"],"category_scores_codex":[0.00321255,0.0024263663,0.0023061663,0.005945829,0.0042620273,0.0018882627,0.0032064093,0.0027637796,0.0026472723],"category_scores_gemma":[0.001810213,0.00265794,0.0016312354,0.00497946,0.00017691293,0.000966745,0.0016357251,0.0058837463,0.00025499304],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.009552986,0.0048511326,0.00046519373,0.002194777,0.0025470266,0.004816726,0.022819597,0.0009195719,0.0052755577,0.8604234,0.066268094,0.01986593],"study_design_scores_gemma":[0.0032486161,0.0014869687,0.00076422864,0.00090392155,0.00124715,0.011892978,0.0021359455,0.003015467,0.001511241,0.0014589054,0.96957177,0.0027628215],"about_ca_topic_score_codex":0.000011360526,"about_ca_topic_score_gemma":0.00023216318,"teacher_disagreement_score":0.9332245,"about_ca_system_score_codex":0.011047758,"about_ca_system_score_gemma":0.0066468236,"threshold_uncertainty_score":0.9991479},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W7120538780","doi":"","title":"Estudos etnográficos na ciência da informação: análise bibliométrica na base de dados Web of Science","year":2023,"lang":"en","type":"other","venue":"LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Conselho Nacional de Desenvolvimento Científico e Tecnológico","keywords":"Ethnography; Resource (disambiguation); Information science; Web of science; Core (optical fiber); Information system","score_opus":0.028285439191153235,"score_gpt":0.27109660165992927,"score_spread":0.24281116246877604,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7120538780","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018779974,0.0022311339,0.0023028865,0.0026935034,0.0057326374,0.0052800504,0.0045434763,0.0104222335,0.9480141],"genre_scores_gemma":[0.979186,0.0007299721,0.0044359253,0.0002070227,0.0010147716,0.00088076293,0.00054317806,0.0025231855,0.010479206],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.986205,0.00074960024,0.0024367871,0.0025399667,0.004992461,0.0030762106],"domain_scores_gemma":[0.98948675,0.0006527871,0.0028674055,0.0030839047,0.001771663,0.0021374698],"candidate_categories":["metaepi_narrow","bibliometrics","sts","research_integrity"],"consensus_categories":["metaepi_narrow","bibliometrics"],"category_scores_codex":[0.0037902165,0.0019456353,0.0020740726,0.020761957,0.001402368,0.0009148098,0.0043286304,0.002139943,0.0005243994],"category_scores_gemma":[0.004725589,0.0019860382,0.0008470418,0.025741527,0.00060584856,0.0016352944,0.0017569067,0.0023005393,0.00076697697],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0013939322,0.004770277,0.006044398,0.0022117642,0.00197659,0.0021720896,0.004722984,0.0004927415,0.04455253,0.6532896,0.26748887,0.010884205],"study_design_scores_gemma":[0.0030226447,0.00020187067,0.0029053483,0.0014524864,0.00041740594,0.0012400771,0.00057466904,0.0035266357,0.0025192641,0.000073835996,0.9819912,0.002074577],"about_ca_topic_score_codex":0.0000371827,"about_ca_topic_score_gemma":0.00071762456,"teacher_disagreement_score":0.960406,"about_ca_system_score_codex":0.0032923545,"about_ca_system_score_gemma":0.010439313,"threshold_uncertainty_score":0.99989766},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W7127327690","doi":"10.58240/1829006x-2025.21.12-47","title":"EXPLORING THE IMPACT OF RACISM AND SOCIAL STRATIFICATION ON HEALTH DISPARITIES: A BIBLIOMETRIC ANALYSIS OF TRENDS, GAPS, AND GLOBAL CONTRIBUTION","year":2025,"lang":"","type":"article","venue":"BULLETIN OF STOMATOLOGY AND MAXILLOFACIAL SURGERY","topic":"Cultural Competency in Health Care","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Universitas Hasanuddin","keywords":"Racism; Health equity; Social determinants of health; Scopus; Social inequality; Inequality; Mental health","score_opus":0.06073644885692987,"score_gpt":0.3599248019868687,"score_spread":0.29918835312993886,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7127327690","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97694063,0.010372367,0.000011908708,0.011150133,0.00037346466,0.00030795162,0.0005336429,0.000015094643,0.00029477823],"genre_scores_gemma":[0.9798143,0.019954847,0.0000075508865,0.0001050357,0.000017666804,0.00002501383,0.000054395627,0.0000038946528,0.000017332253],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99686605,0.00072359527,0.0011853226,0.00040251852,0.00036788068,0.00045464656],"domain_scores_gemma":[0.99723226,0.0010231909,0.001060234,0.00016183195,0.00040711628,0.000115364805],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0018603455,0.00024777759,0.0014023025,0.00796212,0.00083680544,0.00004081265,0.00013460431,0.00023850787,0.00008792132],"category_scores_gemma":[0.0007433041,0.000197511,0.00036074454,0.02828517,0.0016128467,0.0000938122,0.00008325429,0.00018524117,3.0793007e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008518046,0.00024721588,0.5372863,0.0007534748,0.0014752136,0.000002601977,0.0052341204,0.000011891516,0.000021880898,0.07471777,0.0017948826,0.37760288],"study_design_scores_gemma":[0.00029895536,0.00018333308,0.99145275,0.00010749021,0.00040361597,0.0000031830934,0.0033229648,0.000032806318,0.0000212722,0.0006045954,0.0034283295,0.00014070187],"about_ca_topic_score_codex":0.026986515,"about_ca_topic_score_gemma":0.00061301625,"teacher_disagreement_score":0.45416647,"about_ca_system_score_codex":0.0001384967,"about_ca_system_score_gemma":0.00038135523,"threshold_uncertainty_score":0.9923691},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7134613536","doi":"","title":"The rise in subscription fees and the gold open access publishing, examined from financial reports of major publishers","year":2024,"lang":"ja","type":"article","venue":"Kyoto University Research Information Repository (Kyoto University)","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Payment; Financial plan; Quarter (Canadian coin); Work (physics); Financial services","score_opus":0.30010216013937585,"score_gpt":0.4569942950357946,"score_spread":0.15689213489641873,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7134613536","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.76416105,0.0015822063,0.00017037506,0.007486605,0.0018539369,0.0030445196,0.00013019665,0.00005727951,0.22151384],"genre_scores_gemma":[0.9388627,0.0019028779,0.000041859344,0.000055006174,0.000090958216,0.0000037784253,0.000032206422,0.000012530059,0.05899809],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98538876,0.0030823601,0.001218833,0.0010416836,0.008180087,0.0010882912],"domain_scores_gemma":[0.9778947,0.010479948,0.0009405225,0.0018155276,0.008032716,0.0008365524],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"category_scores_codex":[0.034248292,0.00032120195,0.0005587932,0.048108373,0.0025172473,0.07960398,0.013145881,0.00050400133,0.0002730589],"category_scores_gemma":[0.03221793,0.00023561661,0.00026169283,0.12425376,0.0028032702,0.07051289,0.012420703,0.001895324,0.000094265975],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0145887425,0.0006695271,0.1319952,0.00048407508,0.0007087655,0.006830854,0.026811624,0.00027788608,0.00041116937,0.18114449,0.5284851,0.10759255],"study_design_scores_gemma":[0.0034131354,0.00027004443,0.08361412,0.0002660428,0.000059751867,0.00004910272,0.044151172,0.0077712033,0.00031162793,0.0038128803,0.8558956,0.0003853631],"about_ca_topic_score_codex":0.07483634,"about_ca_topic_score_gemma":0.001994842,"teacher_disagreement_score":0.32741043,"about_ca_system_score_codex":0.0015331991,"about_ca_system_score_gemma":0.004045351,"threshold_uncertainty_score":0.99991053},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","scholarly_communication"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7135402466","doi":"10.47203/ijch.2025.v37i06.002","title":"Gender based Differences in Tobacco Cessation initiatives: A Bibliometric Review Analysis","year":2025,"lang":"en","type":"article","venue":"Indian Journal of Community Health","topic":"Smoking Behavior and Cessation","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Tobacco control; Smoking cessation; Varenicline; Citation; Gender disparity; Gender analysis; Intervention (counseling); Psychological intervention; Quit smoking","score_opus":0.1258588344925405,"score_gpt":0.42448982401073077,"score_spread":0.2986309895181902,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7135402466","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98910546,0.004855772,0.0012975197,0.0038819523,0.00008200972,0.00018721423,0.000005396411,0.000007774345,0.00057692215],"genre_scores_gemma":[0.98781353,0.006892457,0.00012935336,0.005105338,0.000017213313,0.000004023773,0.00002886565,0.0000043089685,0.0000049347054],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99662757,0.0020172636,0.0007600551,0.00005387889,0.00038360228,0.00015763326],"domain_scores_gemma":[0.9980592,0.00061580783,0.0005880982,0.00028797617,0.00032090698,0.00012801739],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.004556758,0.00009644918,0.00062478025,0.03015998,0.00020250873,0.00002550191,0.00019395242,0.00006356944,0.0001070061],"category_scores_gemma":[0.0008114097,0.00007988365,0.0001608158,0.047408838,0.00003824756,0.0001795437,0.000016938982,0.001317511,4.7359205e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000030003517,0.0003204993,0.97379935,0.0007915164,0.00012275483,0.000010834838,0.0015172806,0.000005053306,0.0000014585919,0.000017293889,0.00021434404,0.023169601],"study_design_scores_gemma":[0.00090488524,0.00025014003,0.9936623,0.0026842186,0.0003585896,0.00001144287,0.0018997961,0.000008913157,0.000015202697,0.00009153255,0.00006394359,0.000049025944],"about_ca_topic_score_codex":0.00057274144,"about_ca_topic_score_gemma":0.00025927238,"teacher_disagreement_score":0.023120575,"about_ca_system_score_codex":0.00033845202,"about_ca_system_score_gemma":0.0011345649,"threshold_uncertainty_score":0.98083234},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7135439329","doi":"","title":"Finding Hidden Patterns in ECtHR’s Case Law: On How Citation Network Analysis Can Improve Our Knowledge of ECtHR’s Article 14 Practice","year":2017,"lang":"en","type":"article","venue":"Research at the University of Copenhagen (University of Copenhagen)","topic":"Comparative and International Law Studies","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre for International Governance Innovation","funders":"","keywords":"Social network analysis; Network analysis; Citation; Key (lock); Citation analysis; Intelligence analysis","score_opus":0.11533383267224254,"score_gpt":0.39674785243476207,"score_spread":0.28141401976251956,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7135439329","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7730034,0.00038858035,0.00024013662,0.0044138734,0.0000727397,0.0004894644,0.00008354268,0.000008983744,0.22129929],"genre_scores_gemma":[0.8910049,0.0001747231,0.000116944255,0.000011615711,0.000039434904,2.2521435e-7,0.000007923875,0.0000056769504,0.108638555],"study_design_codex":"not_applicable","study_design_gemma":"qualitative","domain_scores_codex":[0.99704885,0.0013781628,0.00014912867,0.00032967297,0.00068432133,0.00040986537],"domain_scores_gemma":[0.996765,0.0010778666,0.00046162575,0.0004927905,0.0010735287,0.0001291555],"candidate_categories":["sts","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.002875872,0.00012652471,0.0004019686,0.00023143561,0.0032278292,0.0000747979,0.0010989975,0.00008902712,0.011328222],"category_scores_gemma":[0.00056046067,0.00013955904,0.00020727435,0.00074762455,0.0008244201,0.0008228127,0.000698374,0.00029315852,0.0009743772],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031318574,0.001122387,0.0023348203,0.00017547351,0.0039036344,0.001570351,0.19155651,0.0014647623,0.0021338346,0.048235465,0.73502886,0.009342027],"study_design_scores_gemma":[0.0029236283,0.00084736175,0.09547596,0.00025095826,0.0008051075,0.0000064195256,0.4680787,0.0023740504,0.0023812181,0.00014173727,0.4262041,0.0005107577],"about_ca_topic_score_codex":0.077156976,"about_ca_topic_score_gemma":0.4359673,"teacher_disagreement_score":0.3588103,"about_ca_system_score_codex":0.00047016214,"about_ca_system_score_gemma":0.00025794984,"threshold_uncertainty_score":0.9998035},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W7137410381","doi":"","title":"Escuelas de Negocios en el QS World University Rankings 2015. Un análisis bibliométrico (2006-2016)","year":2024,"lang":"es","type":"other","venue":"","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Product (mathematics); Productivity; Market research; Quarter (Canadian coin)","score_opus":0.007569067610510502,"score_gpt":0.2567603328408273,"score_spread":0.24919126523031676,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7137410381","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00159587,0.011631762,0.000875996,0.0020918008,0.0018310074,0.0016781061,0.003983845,0.0031456554,0.9731659],"genre_scores_gemma":[0.007262603,0.0048240726,0.003688017,0.00045701538,0.0016680375,0.000009081932,0.00022980598,0.0037678229,0.97809356],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9908894,0.0011990468,0.0008746039,0.0027693375,0.0017752011,0.0024924425],"domain_scores_gemma":[0.99490505,0.00063057063,0.00076493964,0.002195259,0.00042827838,0.0010758915],"candidate_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","bibliometrics","research_integrity","insufficient_payload"],"category_scores_codex":[0.0016311117,0.0021644593,0.0019350396,0.06294813,0.000335544,0.0009248434,0.0027645417,0.0019626282,0.118125595],"category_scores_gemma":[0.00020880536,0.002186967,0.0012991902,0.053809784,0.00066255993,0.00074606563,0.0016377679,0.0023783643,0.3820254],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002765384,0.00043833719,0.0039397143,0.00058327,0.0028142303,0.0012777695,0.0005316838,0.000015148888,0.0007146963,0.011113615,0.9763527,0.0019422796],"study_design_scores_gemma":[0.002615486,0.0000925434,0.0029545592,0.0015943872,0.0031205143,0.000078625555,0.00060245296,0.00017895727,0.0003271029,0.0015371576,0.9843546,0.0025435842],"about_ca_topic_score_codex":0.010712865,"about_ca_topic_score_gemma":0.0036569498,"teacher_disagreement_score":0.26389977,"about_ca_system_score_codex":0.0028384344,"about_ca_system_score_gemma":0.0014977875,"threshold_uncertainty_score":0.99992317},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7140158641","doi":"10.1109/iccist67338.2025.11438293","title":"A Bibliometric Analysis of AI-Driven Adaptive and Predictive Testing in Agile Software Quality Assurance","year":2025,"lang":"","type":"article","venue":"","topic":"Software Engineering Techniques and Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Agile software development; Software quality assurance; Quality assurance; Software; Quality (philosophy); Software quality; Software development","score_opus":0.04270072664259165,"score_gpt":0.3366957245250876,"score_spread":0.29399499788249595,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7140158641","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.063350394,0.0022373265,0.9327055,0.00021623418,0.00015638773,0.00033521146,0.00007031706,0.0002656487,0.0006629472],"genre_scores_gemma":[0.7226524,0.00022911308,0.2769114,0.00009682132,0.000008438926,0.000026301008,0.0000021471071,0.000006614322,0.00006680013],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9969735,0.00036217057,0.0009298538,0.00085683423,0.00047570848,0.00040190204],"domain_scores_gemma":[0.98837423,0.009596618,0.00045921322,0.0007433201,0.0007234976,0.00010314656],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0018363687,0.00030525727,0.00088801986,0.0980202,0.00009791579,0.00021810415,0.0007738177,0.00020427127,0.00002372071],"category_scores_gemma":[0.007955859,0.00031147228,0.00015946665,0.44403994,0.0001354341,0.0011968232,0.00071616087,0.00052379497,8.1824845e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005125656,0.00018698793,0.8393266,0.00021186445,0.0009762859,0.000010516837,0.0005279768,0.034664147,0.00004805758,0.002629487,0.0001609144,0.12120587],"study_design_scores_gemma":[0.00016295841,0.00016371759,0.55741155,0.00021243884,0.00027875684,7.7021895e-7,0.000029870245,0.44066313,0.00018319293,0.0006684618,0.000055552024,0.00016958627],"about_ca_topic_score_codex":0.002294286,"about_ca_topic_score_gemma":0.00018794094,"teacher_disagreement_score":0.65930194,"about_ca_system_score_codex":0.00011011478,"about_ca_system_score_gemma":0.00029753515,"threshold_uncertainty_score":0.9999337},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W800772465","doi":"","title":"Distribution of trace elements in lake sediments: a scientometric analysis","year":2009,"lang":"en","type":"article","venue":"Proceedings of ECOpole","topic":"Heavy metals in environment","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"China; Sediment; Environmental science; TRACE (psycholinguistics); Science Citation Index; Distribution (mathematics); Physical geography; Earth science; Geography; Environmental chemistry; Oceanography; Citation; Geology; Library science; Archaeology; Chemistry; Mathematics; Computer science","score_opus":0.010382072181345499,"score_gpt":0.2585788107640994,"score_spread":0.2481967385827539,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W800772465","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9926009,0.00003910307,0.000067470086,0.00009035085,0.000021817203,0.00018974299,0.000048081936,0.000009840181,0.006932663],"genre_scores_gemma":[0.9981907,0.000024840927,0.0015612249,0.00002881303,0.000006607972,0.000007718261,0.000012560156,0.000004459258,0.00016306079],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99829084,0.000006372793,0.00050889066,0.0003060292,0.00059742323,0.00029042878],"domain_scores_gemma":[0.9995109,0.000015066805,0.00027482712,0.00010072891,0.000014863405,0.000083647414],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00087163266,0.000115984745,0.00027272335,0.00043756602,0.000032660584,0.000011379798,0.0002775878,0.00005184909,0.0013849298],"category_scores_gemma":[0.00013752725,0.00011538515,0.00009413538,0.005026156,0.00014250015,0.00031827937,0.00009439106,0.000076388234,0.00002887687],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001761562,0.00069972035,0.9598922,0.000015068867,0.000039947943,3.7213678e-7,0.0002111266,0.00060503825,0.029314507,0.0001591153,0.0004375473,0.008607714],"study_design_scores_gemma":[0.00036306778,0.00016512303,0.94770575,0.0000135980545,0.00008499823,3.959567e-7,0.00011062783,0.0010601194,0.04758571,0.00045355025,0.0023402763,0.0001168107],"about_ca_topic_score_codex":0.000040592975,"about_ca_topic_score_gemma":0.000033345175,"teacher_disagreement_score":0.018271204,"about_ca_system_score_codex":0.00023041689,"about_ca_system_score_gemma":0.000004489426,"threshold_uncertainty_score":0.99952793},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W940306499","doi":"","title":"La educación para la ciudadanía en la producción científica: una aproximación bibliométrica","year":2013,"lang":"es","type":"article","venue":"Iber: Didáctica de las ciencias sociales, geografía e historia","topic":"Educational Practices and Policies","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Humanities; Political science; Philosophy","score_opus":0.03142435490289617,"score_gpt":0.3460621714942329,"score_spread":0.3146378165913367,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W940306499","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3096451,0.008925738,0.00009009502,0.08839953,0.011972061,0.0032191528,0.00024149394,0.00088829384,0.57661855],"genre_scores_gemma":[0.9644694,0.00560274,0.0022249746,0.0014996275,0.0076097767,0.0006119396,0.000054054693,0.00017924732,0.017748242],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.98673093,0.0053835474,0.0012890032,0.0016637823,0.0025809025,0.0023518212],"domain_scores_gemma":[0.98721826,0.007937943,0.0012597742,0.0011220941,0.0012262594,0.0012356914],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["sts","insufficient_payload"],"category_scores_codex":[0.007249767,0.0010120997,0.0010858938,0.0020379184,0.0030435997,0.0023136183,0.0021276642,0.0015405057,0.0017673401],"category_scores_gemma":[0.007124047,0.0010712416,0.0007247518,0.01001434,0.0046141963,0.0019944638,0.00046314727,0.002141891,0.0017993732],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006879409,0.0030943372,0.014424025,0.0003054825,0.0003121184,0.000044916476,0.067946665,0.000010428046,0.0005657779,0.47425678,0.42448497,0.014485709],"study_design_scores_gemma":[0.0006927857,0.0001419598,0.040240064,0.00016703494,0.0004051941,0.000058111586,0.010775991,0.000023598857,0.000040031995,0.018804023,0.92745405,0.0011971554],"about_ca_topic_score_codex":0.020981664,"about_ca_topic_score_gemma":0.00072075543,"teacher_disagreement_score":0.6548243,"about_ca_system_score_codex":0.002643536,"about_ca_system_score_gemma":0.0080092,"threshold_uncertainty_score":0.9997557},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"}]}