{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":923,"total_is_capped":false,"direct_labels_cover":923,"predictions_cover":923,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"query_hash":"72fb47ad8ea2","filters":{"category":"bibliometrics"}},"results":[{"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,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.05196923175334746,"gpt":0.3484261076968199,"spread":0.2964568759434724,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.01940332,0.0001085865,0.0003237092,0.002261807,0.001253985,0.0009835976,0.003130979,0.00003275443,0.000003290966],"category_scores_gemma":[0.02019862,0.00006406442,0.0003252192,0.03076893,0.003825553,0.005885516,0.001090073,0.0001813557,0.00001468475],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001522741,"about_ca_system_score_gemma":0.0007120204,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001136739,"about_ca_topic_score_gemma":7.47178e-7,"domain_scores_codex":[0.9956865,0.00001783388,0.0008506853,0.0002313574,0.002671872,0.0005417836],"domain_scores_gemma":[0.994259,0.0005955732,0.0020256,0.0008068256,0.002126514,0.000186479],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000004835402,0.0000452368,0.08801094,0.000008388066,0.0001500285,1.424548e-7,0.006928504,0.0004734667,0.001367213,0.008810039,0.1303992,0.763802],"study_design_scores_gemma":[0.0005843117,0.0001937111,0.1426622,0.00002744104,0.0002482357,0.0001180648,0.07462858,0.02250362,0.001643411,0.01109563,0.7461097,0.0001850314],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7817136,0.00007199423,0.2138005,0.003152357,0.0007336466,0.0001680681,0.00000909911,0.00003898368,0.0003117651],"genre_scores_gemma":[0.9304643,0.00002129438,0.06772479,0.001478792,0.00007412121,0.000002877725,4.595117e-7,0.000002543762,0.0002308691],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.763617,"threshold_uncertainty_score":0.9988855,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.652587060367001,"gpt":0.6967628458869507,"spread":0.04417578551994972,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.05948411,0.0001703509,0.0003682181,0.1369572,0.000413167,0.004798167,0.002418063,0.0003491818,0.002225628],"category_scores_gemma":[0.7677514,0.0001292189,0.0002238737,0.416792,0.0001774939,0.00162808,0.00049591,0.001073089,0.000106925],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004534263,"about_ca_system_score_gemma":0.004114283,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003250789,"about_ca_topic_score_gemma":0.000003341181,"domain_scores_codex":[0.9856709,0.0009725204,0.0008070684,0.001311657,0.01059632,0.000641596],"domain_scores_gemma":[0.9095557,0.06000977,0.0007112044,0.001913936,0.02511396,0.002695381],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00001709267,0.000630289,0.005164388,0.000007109139,0.0000195984,0.00000169737,0.002081413,0.000003215156,0.0002061208,0.0243053,0.05017921,0.9173846],"study_design_scores_gemma":[0.001984875,0.0005301767,0.09929708,0.0001258356,0.00003780989,0.00005040143,0.2043805,0.01754517,0.002924726,0.1837441,0.4886123,0.0007670138],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1275684,0.03384215,0.6953861,0.1058889,0.007056116,0.0009652635,0.00002620779,0.0001245586,0.02914232],"genre_scores_gemma":[0.8358486,0.006136074,0.1420518,0.006544131,0.002420725,0.0003218003,0.00004027083,0.00005396955,0.006582601],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9166176,"threshold_uncertainty_score":0.9986865,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.9531000786402971,"gpt":0.5564152615563063,"spread":0.3966848170839908,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.2668508,0.0001465039,0.004888489,0.001820847,0.00003723837,0.0002045148,0.0001654507,0.0000581372,0.003179322],"category_scores_gemma":[0.07878583,0.00006177694,0.002514345,0.001011101,0.00005143153,0.0007728412,0.00005354489,0.0002464357,0.00007655667],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002151029,"about_ca_system_score_gemma":0.0001820432,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000007342512,"about_ca_topic_score_gemma":0.00002887959,"domain_scores_codex":[0.9722267,0.003619225,0.01950309,0.0002344382,0.004233281,0.0001833069],"domain_scores_gemma":[0.9737484,0.008518032,0.01542074,0.0006922198,0.001426849,0.0001937674],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00004569821,0.0001337525,0.5793669,0.0001225566,0.0001545847,0.0001499682,0.0005337544,0.0001647275,0.002109779,0.00005762135,0.1445889,0.2725718],"study_design_scores_gemma":[0.0002702707,0.00006278293,0.9459771,0.002440015,0.0001678644,0.001154298,0.0005938788,0.0005970109,0.0003784989,0.002357909,0.04582328,0.0001771045],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9631046,0.0279148,0.002950586,0.005438004,0.0002356293,0.00008393815,0.000002650342,4.472583e-7,0.0002693213],"genre_scores_gemma":[0.9925882,0.003312625,0.002313992,0.0002623676,0.0002079353,7.107236e-7,6.935122e-8,0.000004722076,0.001309394],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3666103,"threshold_uncertainty_score":0.9977319,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.08917746577153815,"gpt":0.3523308093171318,"spread":0.2631533435455937,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.001941518,0.0002291014,0.000530223,0.005435857,0.0001811995,0.00006778599,0.0007594809,0.0001128365,0.0004043369],"category_scores_gemma":[0.0006767527,0.00007949929,0.0002540664,0.1408525,0.0003557618,0.0005201183,0.0001290358,0.0002141916,0.000007233739],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006309939,"about_ca_system_score_gemma":0.00002349021,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003475668,"about_ca_topic_score_gemma":0.0007659862,"domain_scores_codex":[0.9972717,0.0001953774,0.0003876414,0.000541679,0.001114981,0.0004886687],"domain_scores_gemma":[0.9982044,0.0003849267,0.0003256952,0.0003150523,0.0005288555,0.0002411195],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","study_design_scores_codex":[0.0001209215,0.0006414018,0.06354304,0.0001381313,0.0000358012,0.000001757711,0.00006194856,0.00003421339,0.5509889,0.0004205109,0.00009326562,0.3839201],"study_design_scores_gemma":[0.0002323424,0.0007888868,0.9583662,0.00009354153,0.0001430122,0.000001150714,0.00007121229,0.02743992,0.01200082,0.00002161687,0.0006001662,0.0002411325],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.996933,0.0006825763,8.825533e-7,0.0005332876,0.0001605669,0.0002579668,0.0009032481,0.00005736832,0.0004711075],"genre_scores_gemma":[0.9983675,0.0005307064,0.0001519135,0.0004868027,0.0001494136,0.00001287319,0.0002950955,0.000001419523,0.000004252835],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8948231,"threshold_uncertainty_score":0.8774091,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.7241108159152894,"gpt":0.6179719659443282,"spread":0.1061388499709611,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.3040244,0.0003927602,0.003108634,0.3693186,0.0003526997,0.001346627,0.004739739,0.0002328276,0.0001548714],"category_scores_gemma":[0.6806818,0.0001656883,0.001664211,0.8988605,0.0001831617,0.0005614391,0.001268242,0.0002945683,0.0006429006],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009489898,"about_ca_system_score_gemma":0.0002189743,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000012878,"about_ca_topic_score_gemma":0.000001853323,"domain_scores_codex":[0.9593017,0.003498162,0.02162185,0.001339024,0.01316512,0.001074178],"domain_scores_gemma":[0.9459808,0.01643913,0.02419077,0.005348907,0.007377742,0.0006626272],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0000120688,0.000114052,0.002424926,0.03509367,0.00003390304,0.00001167434,0.0002709839,0.00000176771,0.000920639,0.0001008676,0.9187132,0.04230223],"study_design_scores_gemma":[0.0006258288,0.000421873,0.004296261,0.01503196,0.0001129182,0.00007117442,0.0003022782,0.006154674,0.0001841271,0.00348189,0.968953,0.0003640869],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"review","genre_gemma":"empirical","genre_scores_codex":[0.07901409,0.6124431,0.081806,0.02064036,0.02229428,0.1700873,0.0009310783,0.0004809448,0.01230281],"genre_scores_gemma":[0.4747724,0.2505498,0.1046699,0.006810553,0.005130322,0.03773705,0.001107106,0.0006802449,0.1185426],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.5295419,"threshold_uncertainty_score":0.9996901,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.2802135795098643,"gpt":0.4847581285676076,"spread":0.2045445490577433,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["sts"],"category_scores_codex":[0.02117882,0.00007076069,0.0002323258,0.005429512,0.001389506,0.00147905,0.001206029,0.00002790762,2.377291e-7],"category_scores_gemma":[0.009161904,0.00003278094,0.00003626956,0.0419622,0.007014023,0.002738237,0.0003946512,0.0001952862,7.257341e-8],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005978278,"about_ca_system_score_gemma":0.0004433976,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001018022,"about_ca_topic_score_gemma":0.00003930776,"domain_scores_codex":[0.9966777,0.00002705425,0.0005481834,0.0001352605,0.00233248,0.0002793426],"domain_scores_gemma":[0.9942899,0.001960979,0.001977154,0.00009659831,0.001650492,0.00002485512],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0002569174,0.00009866897,0.551029,0.0001214073,0.00008087121,0.00000123283,0.01170226,0.001858135,0.03741667,0.1973272,0.003427963,0.1966797],"study_design_scores_gemma":[0.002141481,0.002501498,0.6369707,0.0001661279,0.00004123221,0.0004478701,0.2087575,0.09776947,0.008233329,0.01351365,0.02896735,0.0004897672],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9953043,0.0004004287,0.0001653765,0.003739407,0.00007384374,0.0001135893,0.000001856142,0.000003987117,0.0001971972],"genre_scores_gemma":[0.9968065,0.0001924588,0.002888852,0.00008264338,0.00001743036,0.000002274214,3.497029e-8,0.000001173749,0.000008609317],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1970553,"threshold_uncertainty_score":0.9999105,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.7862779829075962,"gpt":0.7031571443127773,"spread":0.08312083859481889,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.1844882,0.000165856,0.0004970072,0.2421684,0.001109686,0.002588179,0.003694294,0.0002490385,0.0001507489],"category_scores_gemma":[0.127095,0.0001342473,0.00005949626,0.5092384,0.002072211,0.0009045127,0.002558496,0.002082496,0.0001681883],"about_ca_system_candidate":true,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001143691,"about_ca_system_score_gemma":0.007747794,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03315038,"about_ca_topic_score_gemma":0.005316909,"domain_scores_codex":[0.9454972,0.008164673,0.000955415,0.001719316,0.0409476,0.002715793],"domain_scores_gemma":[0.9568757,0.01761192,0.0001544177,0.001776589,0.02113022,0.00245111],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00115748,0.005328098,0.4738782,0.00002671007,0.00003041158,0.0006778577,0.01315452,0.00001837002,0.002064446,0.02000117,0.2722166,0.2114461],"study_design_scores_gemma":[0.002865368,0.0047742,0.887785,0.00004687319,0.000001531385,0.00001109558,0.01468072,0.000906811,0.0001327432,0.02636861,0.06226656,0.0001605351],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9792498,0.0002926729,0.0000349394,0.01353765,0.00006502938,0.001311289,0.0000355659,0.00001262631,0.00546038],"genre_scores_gemma":[0.9807648,0.0002889131,0.000184821,0.00001235308,0.0002277506,0.000004704167,0.000003319155,0.00001012233,0.01850321],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4139068,"threshold_uncertainty_score":0.9984472,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.6759460167148517,"gpt":0.6428787603658095,"spread":0.03306725634904217,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.1105867,0.0001252791,0.0003975341,0.1961017,0.001480091,0.003022602,0.003578217,0.0002107702,0.00001788467],"category_scores_gemma":[0.07944057,0.0000759355,0.0000612592,0.6007964,0.005534285,0.001059885,0.001483366,0.0009892581,0.000006216451],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001908275,"about_ca_system_score_gemma":0.001213069,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002149599,"about_ca_topic_score_gemma":0.00006073781,"domain_scores_codex":[0.98425,0.000626305,0.0005848394,0.0009559014,0.01238464,0.0011983],"domain_scores_gemma":[0.9856061,0.003685956,0.0001801565,0.001352742,0.008299977,0.0008750276],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.000009533296,0.0001033095,0.2988846,0.000004157993,0.00005753013,0.000006226197,0.001127208,0.00009716838,0.003146597,0.01702386,0.001259559,0.6782803],"study_design_scores_gemma":[0.0003077975,0.0006630869,0.8852993,0.0000174604,0.00003865564,0.000004226462,0.002682596,0.06876666,0.002269824,0.0366415,0.003104863,0.000204091],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9895003,0.001145746,0.0005756292,0.007412505,0.00004070182,0.0002601793,0.00001515751,0.00001940069,0.001030381],"genre_scores_gemma":[0.9991848,0.0002647186,0.0001556129,0.0001119951,0.0001203875,0.000006543683,0.000001033604,0.000005074159,0.0001498178],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6780762,"threshold_uncertainty_score":0.9998199,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.6756483836744561,"gpt":0.4639262424223971,"spread":0.211722141252059,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004359345,0.0001641958,0.0003131358,0.003089755,0.0001916437,0.0002798409,0.0005323717,0.0001557484,0.00009235202],"category_scores_gemma":[0.006037281,0.0001810383,0.00001112853,0.008497719,0.0001809535,0.0009881628,0.0009497867,0.0003951282,0.0007106015],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002227656,"about_ca_system_score_gemma":0.0004090706,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001743417,"about_ca_topic_score_gemma":0.0001011961,"domain_scores_codex":[0.9978738,0.00009865796,0.0007723781,0.0006748414,0.0001381105,0.0004422388],"domain_scores_gemma":[0.9975904,0.0002536775,0.000198373,0.0008108311,0.00100438,0.0001423886],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00006818076,0.0001964355,0.08947204,0.0001339797,0.00006749904,0.000004063927,0.0040013,0.000001324836,0.0005884174,0.8238314,0.05442255,0.02721277],"study_design_scores_gemma":[0.003104143,0.0002882645,0.2702588,0.000281659,0.000006257429,0.00001265003,0.0007205748,0.02058598,0.0003969621,0.3053985,0.3981752,0.0007710076],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9661533,0.005118477,0.0007973618,0.006890927,0.0002518698,0.0005077638,0.0005106116,0.00006196247,0.01970776],"genre_scores_gemma":[0.9943405,0.0003977368,0.0007742619,0.0002545765,0.0006471721,0.00004285382,0.0003131141,0.00003107115,0.003198763],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.518433,"threshold_uncertainty_score":0.9133581,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"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":"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,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.872966831153213,"gpt":0.6254438142378042,"spread":0.2475230169154089,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.09664137,0.0001753771,0.0006941858,0.4405308,0.0003897616,0.0007019419,0.003181933,0.0001542178,0.0004866706],"category_scores_gemma":[0.05953104,0.0001256271,0.0001734338,0.766765,0.001225225,0.001425368,0.001410186,0.0005116217,0.00001835959],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005574236,"about_ca_system_score_gemma":0.0008836918,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.04696523,"about_ca_topic_score_gemma":0.01873166,"domain_scores_codex":[0.9827235,0.001359591,0.0006469616,0.00138475,0.01278458,0.001100574],"domain_scores_gemma":[0.9875354,0.00497048,0.0003451366,0.001278139,0.004956141,0.000914686],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00003939861,0.0005245767,0.9679577,0.000005277978,0.00005193958,0.000009488515,0.003533811,0.00001156052,0.0001515115,0.0005357205,0.0001065642,0.0270724],"study_design_scores_gemma":[0.0002270595,0.0003935306,0.9861859,0.000003647233,0.00002919876,8.675518e-7,0.004411916,0.007311497,0.0003616364,0.0007324847,0.0001932674,0.0001489997],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9955975,0.0003410447,0.0003625681,0.00009150092,0.0001520823,0.0003088051,0.00005928618,0.00001566487,0.003071569],"genre_scores_gemma":[0.9969999,0.00009431289,0.0003836482,0.000003084328,0.00001358111,0.000001911296,0.000002390196,0.000006587633,0.002494563],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3262342,"threshold_uncertainty_score":0.9991739,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal; Polytechnique Montréal","funders":"","keywords":"Gender gap; Psychology; Sociology; Demographic economics; Economics","retraction":null,"screen_n_in":null,"score":{"opus":0.6838478894611898,"gpt":0.6503567219247052,"spread":0.03349116753648462,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.05528185,0.0001998701,0.0004190041,0.005378654,0.0008224125,0.0001566266,0.0004699895,0.000158725,0.0004484158],"category_scores_gemma":[0.6148646,0.0001019866,0.00003940639,0.01055684,0.006009784,0.0004874444,0.0004883672,0.00134692,0.0001134741],"about_ca_system_candidate":true,"about_ca_system_consensus":true,"about_ca_system_score_codex":0.003964731,"about_ca_system_score_gemma":0.05164779,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006891842,"about_ca_topic_score_gemma":0.0003894531,"domain_scores_codex":[0.9863351,0.0008122507,0.0003963663,0.0008306471,0.006881052,0.004744577],"domain_scores_gemma":[0.9521697,0.008423207,0.00009873187,0.001766484,0.006448668,0.03109326],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","study_design_scores_codex":[0.0007832059,0.0004150591,0.3378479,0.0009900542,0.00004595448,0.0002434792,0.01160254,3.088789e-7,0.4503441,0.003687255,0.05518405,0.1388561],"study_design_scores_gemma":[0.00484758,0.002326857,0.9666799,0.001587759,0.000004782795,0.00009055457,0.002195869,0.0001270857,0.007512067,0.003180834,0.01122627,0.0002204502],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.913521,0.0003473114,0.0000108232,0.06914721,0.00005935441,0.001003136,0.000008927449,0.00003933584,0.01586293],"genre_scores_gemma":[0.9970373,0.0004590301,0.00003603563,0.0004390957,0.0005159189,0.00002189816,0.000002605633,0.00003033019,0.001457787],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.628832,"threshold_uncertainty_score":0.9998589,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.8303977243332353,"gpt":0.6710300330829925,"spread":0.1593676912502429,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.3731061,0.000193056,0.001184311,0.0004303302,0.0004181263,0.0004345586,0.003742714,0.000101187,0.004156955],"category_scores_gemma":[0.3992323,0.00008170154,0.0003017064,0.007155955,0.00319495,0.0004016326,0.0002400073,0.0007256275,0.0003058846],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004235767,"about_ca_system_score_gemma":0.00009722513,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001543104,"about_ca_topic_score_gemma":0.00001307397,"domain_scores_codex":[0.9646918,0.01839627,0.004332855,0.00643433,0.005604085,0.0005406974],"domain_scores_gemma":[0.9702287,0.0122155,0.003299846,0.01145486,0.002620866,0.0001801872],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0002685632,0.005461874,0.6356961,0.0003300307,0.0001309211,0.000191086,0.06401878,0.0006903328,0.03464501,0.08822694,0.01190653,0.1584339],"study_design_scores_gemma":[0.0001957317,0.0002981776,0.9122776,0.00005804512,0.00002147954,0.00009668678,0.005852564,0.002069968,0.00007956713,0.0784025,0.0005227869,0.0001248653],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9479204,0.004228794,0.0005178048,0.01094838,0.0004474581,0.0005919368,0.000005176702,0.00000567789,0.03533441],"genre_scores_gemma":[0.9923798,0.0004333655,0.005687431,0.0001418952,0.00002843924,0.00003278854,1.462613e-7,0.000003458522,0.00129265],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2765816,"threshold_uncertainty_score":0.9995178,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.3265772538046295,"gpt":0.5119929688617303,"spread":0.1854157150571007,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["sts"],"category_scores_codex":[0.01517532,0.000113504,0.0002225426,0.005488127,0.003502267,0.00118903,0.003476512,0.00005840816,0.00001703442],"category_scores_gemma":[0.02085039,0.00004212226,0.0003188631,0.06675902,0.00397756,0.004047459,0.0003780067,0.0006843707,0.000006488919],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002915002,"about_ca_system_score_gemma":0.001017892,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001700896,"about_ca_topic_score_gemma":5.537988e-7,"domain_scores_codex":[0.9922488,0.00007500657,0.0008125613,0.0001282671,0.006304288,0.0004310769],"domain_scores_gemma":[0.9867939,0.002684447,0.00253445,0.0003343049,0.007501177,0.0001516907],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.00009780682,0.00008249054,0.1050041,0.000004960155,0.0001287138,7.043571e-7,0.008819265,0.0004573914,0.01914751,0.001048284,0.1834876,0.6817212],"study_design_scores_gemma":[0.001551607,0.001826382,0.526252,0.00002487265,0.00004332952,0.003159626,0.0603164,0.05702914,0.006229599,0.01365121,0.3294467,0.0004690734],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9520446,0.0002219826,0.02939676,0.01770973,0.0003274615,0.0001809829,0.00002988109,0.000007634373,0.00008096347],"genre_scores_gemma":[0.9970457,0.0009196606,0.001010278,0.0009105585,0.00006081862,0.000003001913,1.897327e-7,0.000002766558,0.00004701877],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6812521,"threshold_uncertainty_score":0.9998478,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.1955078746505247,"gpt":0.4623797083122703,"spread":0.2668718336617457,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.02030293,0.0001063433,0.0004570007,0.3879982,0.001042778,0.001625596,0.00341921,0.00007028617,0.00001220074],"category_scores_gemma":[0.03729713,0.00006118925,0.0004377193,0.8834214,0.006742647,0.00787826,0.000707602,0.0002585347,0.00001057392],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001160325,"about_ca_system_score_gemma":0.001387902,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000264137,"about_ca_topic_score_gemma":5.160722e-7,"domain_scores_codex":[0.9902678,0.00001416645,0.001268296,0.0001738131,0.007887334,0.0003886229],"domain_scores_gemma":[0.9630436,0.000609794,0.003511141,0.0003966988,0.03224511,0.0001936573],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.00007316556,0.00009902123,0.3403982,0.00002902375,0.0002601891,2.841863e-7,0.005985707,0.0001058225,0.003419667,0.007811286,0.02899052,0.6128271],"study_design_scores_gemma":[0.001771479,0.001958616,0.5000706,0.00003059775,0.0002407674,0.0005887469,0.05346389,0.04823322,0.0138195,0.009815167,0.3695249,0.0004824856],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9842134,0.00006186569,0.01176674,0.003056912,0.0001541136,0.0001997077,0.00001095474,0.00001134054,0.0005250382],"genre_scores_gemma":[0.9951681,0.0002244475,0.004161401,0.0003327184,0.00001141745,0.000005184791,4.385744e-7,0.000001623189,0.00009464654],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6123446,"threshold_uncertainty_score":0.9994108,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":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":"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.02386674333689827,"gpt":0.2543653634252096,"spread":0.2304986200883113,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007972515,0.00006006131,0.00009388247,0.0002875361,0.0002581223,0.00006382904,0.0005328443,0.00008333671,0.000007570384],"category_scores_gemma":[0.000633768,0.00004182437,0.00006445764,0.001732676,0.0005951077,0.002399083,0.0002253075,0.000338378,7.023477e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003351673,"about_ca_system_score_gemma":0.0001477037,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001033414,"about_ca_topic_score_gemma":0.000003054178,"domain_scores_codex":[0.9988476,0.000004755106,0.0002561871,0.00006887512,0.0007249579,0.00009762246],"domain_scores_gemma":[0.9981208,0.00004942371,0.0009504773,0.00008161923,0.0007880792,0.000009561675],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0004425397,0.0002171689,0.1461739,0.00001651511,0.0001904001,4.90529e-7,0.001008252,0.002654309,0.003727416,0.7684973,0.03585563,0.04121599],"study_design_scores_gemma":[0.007022057,0.0002244663,0.3596581,0.0001613919,0.000140968,0.0000402816,0.05768771,0.06901142,0.0006257758,0.0139138,0.4911005,0.0004135093],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.945246,0.000003475776,0.0004583988,0.05085815,0.0004009778,0.0001429487,0.00001431665,0.000008377137,0.002867373],"genre_scores_gemma":[0.9979131,0.00001482455,0.0006096022,0.001239531,0.00004488561,4.224107e-7,9.098535e-7,0.000001383321,0.0001753771],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7545835,"threshold_uncertainty_score":0.2192699,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":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":"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,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":true,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.3334433946008996,"gpt":0.4961553382486549,"spread":0.1627119436477553,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.08107378,0.0001390894,0.0003110713,0.03402918,0.001081469,0.01064864,0.002797762,0.0005501052,0.01038845],"category_scores_gemma":[0.3063317,0.00009917778,0.0002036654,0.06545956,0.0002076342,0.001307828,0.0001640679,0.001497262,0.001171165],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001971778,"about_ca_system_score_gemma":0.005237541,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007693563,"about_ca_topic_score_gemma":0.03113579,"domain_scores_codex":[0.9711182,0.0007254727,0.0008256675,0.0003913632,0.02564573,0.001293546],"domain_scores_gemma":[0.9762717,0.002183804,0.0006431378,0.0003524112,0.01287234,0.007676636],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000002213984,0.00001248318,0.1015512,5.924e-7,0.00002489375,0.0001193744,0.0001234219,2.134901e-7,0.00000590086,0.0001070481,0.4496251,0.4484275],"study_design_scores_gemma":[0.0003753556,0.00006154756,0.1490853,0.00001317011,0.000003368051,0.0001801671,0.0005085059,0.0004152021,0.00003005905,0.005552176,0.8436528,0.0001223041],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.41811,0.004612911,0.009230076,0.4997303,0.0150188,0.0004934419,0.00009336133,0.00005080419,0.0526603],"genre_scores_gemma":[0.9742511,0.000494603,0.0002422246,0.006159651,0.00583498,0.000007096136,0.000001337181,0.00001413167,0.01299491],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.556141,"threshold_uncertainty_score":0.9996065,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.2681295025211727,"gpt":0.4824654141450026,"spread":0.2143359116238299,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["bibliometrics","sts"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.00701017,0.000204343,0.0004308519,0.02126955,0.001517379,0.0000369934,0.0007843348,0.0001821242,0.0002543287],"category_scores_gemma":[0.0005057551,0.00009055939,0.00008530278,0.09628533,0.0003224619,0.00005803564,0.0003274043,0.001954307,0.00004150143],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008887538,"about_ca_system_score_gemma":0.002116648,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.9561101,"about_ca_topic_score_gemma":0.8448243,"domain_scores_codex":[0.9924088,0.004689184,0.000735611,0.0003128829,0.0009176559,0.0009359236],"domain_scores_gemma":[0.989983,0.008052072,0.0002543883,0.001014962,0.0005303863,0.0001651432],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"observational","study_design_scores_codex":[0.0009850351,0.00004581529,0.4336362,0.000330573,0.0003768099,0.00005350217,0.03041011,0.0001373406,4.404662e-7,0.00215015,0.5302569,0.001617156],"study_design_scores_gemma":[0.001110081,0.00005954195,0.7430903,0.00004721414,0.0001037316,9.100403e-7,0.009044307,0.0001278886,7.820084e-7,0.0002117196,0.2460959,0.0001076605],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9476455,0.0006305022,0.000006711599,0.0381513,0.0003066118,0.002059489,0.0001004079,0.00003225447,0.01106715],"genre_scores_gemma":[0.9505448,0.0006256992,0.00001779912,0.04308137,0.0002240748,0.0003557203,0.000284264,0.00002867139,0.004837554],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3094541,"threshold_uncertainty_score":0.9997825,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.7558819682205911,"gpt":0.6277056305859157,"spread":0.1281763376346754,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.09974855,0.0002088223,0.0006025062,0.4767476,0.0009271588,0.007461565,0.002574105,0.0001240696,0.0003443215],"category_scores_gemma":[0.3991574,0.0001175394,0.0001643002,0.7982325,0.00181283,0.001135526,0.001701463,0.000261525,0.00009876736],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004343511,"about_ca_system_score_gemma":0.0003677282,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001203945,"about_ca_topic_score_gemma":0.00009319081,"domain_scores_codex":[0.9693173,0.0005634705,0.000879407,0.002003178,0.02569866,0.001538015],"domain_scores_gemma":[0.9652255,0.01815615,0.0003588853,0.001794067,0.0127262,0.001739262],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00001667726,0.0001081464,0.9263775,0.000008778137,0.00008360518,0.000002911884,0.0007880807,0.00002786013,0.01989581,0.0045163,0.004624322,0.04355003],"study_design_scores_gemma":[0.0002493096,0.0002517567,0.9469564,0.0000267177,0.00005707133,0.00000189696,0.002226408,0.0006100925,0.009076896,0.002316291,0.0379461,0.0002810913],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9753835,0.0002480292,0.02024381,0.001788594,0.0009827956,0.0004143104,0.0003321317,0.00002790379,0.0005789458],"genre_scores_gemma":[0.9891614,0.0000179198,0.002309162,0.00002737837,0.00005451293,0.00001044032,0.000002150611,0.00001488962,0.008402177],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3214849,"threshold_uncertainty_score":0.9935688,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.09078199573774924,"gpt":0.4091928979380296,"spread":0.3184109022002803,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01218653,0.00009163265,0.000297848,0.001201348,0.0003419147,0.0006052165,0.001148215,0.00004997226,0.000008541886],"category_scores_gemma":[0.00420164,0.00004900637,0.0002171935,0.009250639,0.001419402,0.005489805,0.0002173999,0.0001737817,0.000004899618],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001254322,"about_ca_system_score_gemma":0.0001692373,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001101173,"about_ca_topic_score_gemma":0.00002599965,"domain_scores_codex":[0.997117,0.00006428365,0.0008877515,0.0001273518,0.001525796,0.0002778135],"domain_scores_gemma":[0.9963351,0.0004701557,0.001737698,0.0003898343,0.0009873743,0.00007980366],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.00005161586,0.0001120362,0.04618249,0.0000191877,0.0001143131,1.293369e-7,0.01156454,0.0004473299,0.0005035694,0.03128979,0.004732374,0.9049826],"study_design_scores_gemma":[0.001077252,0.0003516331,0.3976319,0.00007015421,0.0003020225,0.00001352903,0.3310764,0.04945956,0.001191119,0.03537738,0.183068,0.0003809643],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8012107,0.00005475375,0.10133,0.09650245,0.0005437273,0.0002559965,0.00001742504,0.00001588989,0.00006904471],"genre_scores_gemma":[0.9905065,0.00004233454,0.006253454,0.003064326,0.00002424999,0.0000120378,0.00000227845,0.000001592795,0.00009318952],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9046016,"threshold_uncertainty_score":0.5836117,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.8178627288745213,"gpt":0.5691392540613007,"spread":0.2487234748132205,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.005769523,0.0001064121,0.0004358964,0.1001214,0.001202996,0.0001810399,0.001032251,0.00006403039,0.0003732873],"category_scores_gemma":[0.04071479,0.00008326264,0.0002173837,0.1215166,0.002003598,0.0003092969,0.0004173456,0.0004152429,0.00008935257],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003101507,"about_ca_system_score_gemma":0.0001250677,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004962005,"about_ca_topic_score_gemma":0.0001533685,"domain_scores_codex":[0.9948865,0.001398894,0.0003639027,0.0004842224,0.00230322,0.0005632223],"domain_scores_gemma":[0.9855291,0.01192271,0.0001862526,0.001203806,0.001040087,0.0001180497],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0002243364,0.0002439467,0.4209071,0.00006546422,0.0001490973,0.000006727329,0.0004854559,0.00008895498,0.5712587,0.002703479,0.000456279,0.003410425],"study_design_scores_gemma":[0.0003872821,0.0002416762,0.4770129,0.00001942917,0.00002866559,6.485485e-7,0.0001965973,0.0006732449,0.5189373,0.001207489,0.001216599,0.00007806552],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9940803,0.0002915371,0.00009850263,0.0001532548,0.00008138554,0.0002302567,0.00004431243,0.00001772711,0.005002707],"genre_scores_gemma":[0.997363,0.0006323482,0.0001964732,0.000007205649,0.00005832476,0.00001415574,6.633866e-7,0.00001167262,0.001716195],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.05610583,"threshold_uncertainty_score":0.9673657,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.8768176192761763,"gpt":0.8135927768787335,"spread":0.06322484239744275,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","research_integrity"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.3466973,0.000515479,0.007231952,0.009849807,0.001413736,0.00006458577,0.001720078,0.0004915367,0.0005264744],"category_scores_gemma":[0.01669248,0.0003807912,0.0004647653,0.02399792,0.0002249563,0.0003829674,0.001224063,0.002803407,0.0004746551],"about_ca_system_candidate":true,"about_ca_system_consensus":true,"about_ca_system_score_codex":0.004550843,"about_ca_system_score_gemma":0.01294925,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.02588356,"about_ca_topic_score_gemma":0.04100468,"domain_scores_codex":[0.7348617,0.2435455,0.01220722,0.001769797,0.004094049,0.003521719],"domain_scores_gemma":[0.9340963,0.05355746,0.00472767,0.002319767,0.003353944,0.001944802],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","study_design_scores_codex":[0.00001645033,0.00004361364,0.0001933321,0.8664023,0.00007888093,5.479517e-7,0.009494971,1.30948e-7,1.537012e-7,0.0001306714,0.0003839339,0.123255],"study_design_scores_gemma":[0.0003793933,0.0003123984,0.0008594386,0.7930966,0.0001111927,0.000001138931,0.004791769,0.00003645942,1.178766e-7,0.0001038149,0.2001076,0.0002001146],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.00003798263,0.9600897,0.002010124,0.004482548,0.0001780843,0.03248675,0.0005274392,0.00004756052,0.0001397523],"genre_scores_gemma":[0.00002838835,0.9795943,0.006354507,0.003053722,0.0001189454,0.01016313,0.0001462272,0.00007546199,0.0004652904],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.3300048,"threshold_uncertainty_score":0.9998863,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.7446807243754998,"gpt":0.7166696462433838,"spread":0.02801107813211601,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.1110541,0.0001452651,0.0002770204,0.1449599,0.001836017,0.004066691,0.006555762,0.00009848775,0.0007072539],"category_scores_gemma":[0.1398584,0.00008727179,0.00006876844,0.649469,0.01472262,0.00342216,0.0009533037,0.0006162042,0.0004134382],"about_ca_system_candidate":true,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007039648,"about_ca_system_score_gemma":0.02144049,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00123317,"about_ca_topic_score_gemma":0.00000792122,"domain_scores_codex":[0.9689531,0.0003060286,0.0005745196,0.001219144,0.02703029,0.001916955],"domain_scores_gemma":[0.9721659,0.003736732,0.0001906603,0.0013432,0.02058616,0.001977392],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0000398802,0.0001529086,0.06106233,0.000001716957,0.000004073504,0.000002304863,0.0007125039,0.00002544049,0.01055144,0.07187016,0.008114133,0.8474631],"study_design_scores_gemma":[0.0004820411,0.001576618,0.9268859,0.00001458668,9.619396e-7,0.00001401453,0.001089487,0.003245034,0.003406226,0.04104366,0.02199176,0.0002497009],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8832105,0.0001906876,0.000004783341,0.003469484,0.00007329052,0.0002159547,0.00004289124,0.00001168671,0.1127807],"genre_scores_gemma":[0.9973704,0.000579678,0.00007398598,0.0001658549,0.0001161199,0.000008252528,3.018669e-7,0.000004894948,0.001680552],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8658236,"threshold_uncertainty_score":0.9994634,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.2321609182054428,"gpt":0.5205154533318411,"spread":0.2883545351263983,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.1563829,0.00007295605,0.0002229463,0.1205403,0.002183084,0.003194276,0.001711683,0.0001203672,0.000001757956],"category_scores_gemma":[0.1357491,0.00003920052,0.00006032514,0.2378983,0.001895122,0.01006718,0.0004871218,0.000305009,0.000002487516],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001915284,"about_ca_system_score_gemma":0.0005925051,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002364792,"about_ca_topic_score_gemma":0.0000313437,"domain_scores_codex":[0.9917275,0.0001171021,0.001056519,0.0001626187,0.00651302,0.0004232283],"domain_scores_gemma":[0.9625207,0.003804151,0.001946386,0.0003855405,0.03114583,0.0001973722],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00006565619,0.00006275449,0.5927094,0.00002207323,0.00007412658,3.99706e-8,0.003422791,0.000420019,0.003386487,0.07882194,0.002977151,0.3180375],"study_design_scores_gemma":[0.0007305701,0.0005058637,0.6841904,0.00001483755,0.00005266448,0.000006887899,0.006948883,0.2558177,0.002183531,0.03082073,0.01861867,0.0001092897],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9861547,0.0002085948,0.0069395,0.005485698,0.0003723175,0.0002899472,0.0000167347,0.000007701477,0.0005248531],"genre_scores_gemma":[0.9987474,0.0003614008,0.0007787516,0.00003493373,0.00001584554,0.000005799609,6.804995e-7,0.00000126116,0.00005390917],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3179283,"threshold_uncertainty_score":0.9991159,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.9881890202619653,"gpt":0.7469333192546113,"spread":0.2412557010073541,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts","insufficient_payload"],"consensus_categories":["metaresearch","insufficient_payload"],"category_scores_codex":[0.8078911,0.0002594265,0.003340085,0.00598592,0.0004685068,0.0005160672,0.003795607,0.0001944614,0.0449817],"category_scores_gemma":[0.3223713,0.0001497856,0.0005041219,0.01554193,0.003433848,0.001121509,0.0005694111,0.001375669,0.002436257],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000256491,"about_ca_system_score_gemma":0.001042109,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001807971,"about_ca_topic_score_gemma":0.000599915,"domain_scores_codex":[0.5424928,0.3158225,0.02142779,0.003721924,0.1144018,0.002133145],"domain_scores_gemma":[0.7953251,0.1346063,0.007049821,0.005489219,0.05630734,0.001222262],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.00009480126,0.002527922,0.006166672,0.0002947228,0.0005432023,0.00003689345,0.9139432,2.890382e-7,0.003520758,0.06650325,0.004433029,0.001935241],"study_design_scores_gemma":[0.0006814597,0.00110683,0.07232341,0.001544904,0.00008883067,0.000004511339,0.6453928,0.004064035,0.01096329,0.2633777,0.0001351412,0.0003170489],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9424556,0.0008785365,0.0006837029,0.001944837,0.0000439568,0.00103595,0.00004102412,0.000002928739,0.0529134],"genre_scores_gemma":[0.9849542,0.00005745765,0.01061139,0.00005977202,0.00003928522,0.0001171124,0.000001992283,0.00001386578,0.004144961],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4855198,"threshold_uncertainty_score":0.9992782,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.838152280754456,"gpt":0.6455667087672443,"spread":0.1925855719872117,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.1082317,0.0002111957,0.0006495731,0.2690909,0.0003236766,0.007054362,0.0114311,0.0001994283,0.0004386413],"category_scores_gemma":[0.04319692,0.0001450322,0.000194003,0.6152785,0.0009399091,0.005927231,0.003343812,0.0009296827,0.00002469403],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009683496,"about_ca_system_score_gemma":0.001984661,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004475062,"about_ca_topic_score_gemma":0.0009897391,"domain_scores_codex":[0.9737656,0.002431782,0.001498649,0.001011109,0.02021208,0.001080846],"domain_scores_gemma":[0.9783849,0.008512352,0.001221544,0.001751825,0.009483315,0.0006460279],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0000714087,0.0004190861,0.9661016,0.00002050067,0.000007404975,0.000005940251,0.001483423,0.003067941,0.007809614,0.0001659838,0.0003029797,0.0205441],"study_design_scores_gemma":[0.001086377,0.0002137827,0.9348478,0.0001193549,0.000004348318,0.000001658204,0.008116978,0.01944543,0.02996376,0.005993111,0.00004977768,0.0001576139],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9955313,0.0008126932,0.0002076451,0.001303783,0.0004433122,0.0009106401,0.00006028094,0.000006311641,0.0007239901],"genre_scores_gemma":[0.9989226,0.0001418545,0.0003432622,0.00006284908,0.0000231433,0.000008554824,0.000001501654,0.00001389719,0.0004822791],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3461877,"threshold_uncertainty_score":0.9939764,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Calgary","funders":"","keywords":"Randomized controlled trial; Evidence-based medicine; Medicine; Placebo; Plastic surgery; Surgery; Alternative medicine","retraction":null,"screen_n_in":null,"score":{"opus":0.9182824885556872,"gpt":0.572964135707008,"spread":0.3453183528486792,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","insufficient_payload"],"consensus_categories":["metaresearch","insufficient_payload"],"category_scores_codex":[0.1479968,0.001089383,0.02483101,0.008924003,0.0001584674,0.0004300977,0.002433293,0.0006775048,0.01396271],"category_scores_gemma":[0.6369861,0.0005937038,0.008649891,0.01022407,0.001146067,0.0006672803,0.0005037706,0.001562405,0.004293304],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000396931,"about_ca_system_score_gemma":0.004037193,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005314026,"about_ca_topic_score_gemma":0.00007081865,"domain_scores_codex":[0.9114874,0.04465953,0.02704288,0.003155317,0.01236431,0.001290554],"domain_scores_gemma":[0.4535389,0.5319249,0.009204293,0.003246885,0.001775566,0.0003094747],"domain_codex":null,"domain_gemma":"methods","domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00001693896,0.00006773239,0.02739477,0.004418702,0.0006659739,0.0002417051,0.00009181951,0.000008670196,4.683581e-7,0.00006081455,0.0364504,0.930582],"study_design_scores_gemma":[0.0001191321,0.00003754983,0.04601054,0.0697675,0.0009972451,0.000869097,0.0002457093,0.0001762239,0.000001409338,0.001169908,0.87919,0.001415729],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.007730691,0.9852973,0.0003022363,0.000006846473,0.004557755,0.001161068,0.0004314426,0.000009133867,0.0005035118],"genre_scores_gemma":[0.003680189,0.9945496,0.0002842241,0.000005180692,0.0002968217,0.0002038075,0.00002154835,0.0000778796,0.0008807336],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.9291663,"threshold_uncertainty_score":0.9996514,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Library science; Political science; Sociology; Computer science","retraction":null,"screen_n_in":null,"score":{"opus":0.9310062741247978,"gpt":0.7398197713321439,"spread":0.1911865027926538,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.3787386,0.00009932149,0.0001594555,0.03890661,0.00262834,0.006547365,0.002153058,0.0001484505,0.001876795],"category_scores_gemma":[0.4854091,0.00006166334,0.00005475197,0.2369975,0.002114334,0.0009025843,0.0005463376,0.001440555,0.000449815],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001556715,"about_ca_system_score_gemma":0.001654821,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009542593,"about_ca_topic_score_gemma":0.0004928461,"domain_scores_codex":[0.9422621,0.00590698,0.0009780321,0.001253549,0.04843822,0.001161079],"domain_scores_gemma":[0.8053127,0.07102977,0.0003242964,0.002953659,0.1197017,0.0006778389],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.00005226105,0.0005543532,0.05216272,0.00001577684,0.00002269403,3.181317e-7,0.001643517,0.00008468857,0.1486854,0.4982465,0.1535962,0.1449356],"study_design_scores_gemma":[0.0005898558,0.0003797115,0.5472916,0.00002646997,0.000006831246,0.000002831291,0.003006627,0.1401683,0.0107034,0.2571795,0.04049361,0.0001512516],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.932269,0.0002174226,0.000217279,0.03912688,0.0003973407,0.001601754,0.00004117568,0.00001572947,0.02611341],"genre_scores_gemma":[0.9974484,0.00006989692,0.0002783833,0.00002578484,0.0001163321,0.0002458047,0.00003444543,0.000009335355,0.001771596],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4951289,"threshold_uncertainty_score":0.9990356,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.01484112560340749,"gpt":0.3143350200127951,"spread":0.2994938944093876,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.001098274,0.0001719278,0.0002994731,0.01090065,0.00009514178,0.0003234973,0.000934427,0.00005750515,0.00001076958],"category_scores_gemma":[0.0003679678,0.0001382727,0.000125952,0.02500297,0.00002225663,0.001033843,0.0002330069,0.0004117702,0.00001351341],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001843642,"about_ca_system_score_gemma":0.00008092297,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005291308,"about_ca_topic_score_gemma":0.0000147729,"domain_scores_codex":[0.9978135,0.00062302,0.0004438435,0.0004705895,0.0004312878,0.0002178124],"domain_scores_gemma":[0.9981641,0.0004056917,0.0004013331,0.0008630939,0.0001044392,0.00006140777],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.000003895831,0.0004129925,0.1473175,0.0001083839,0.0001739869,0.000004046204,0.09039904,0.02818277,0.0009806391,0.002580442,0.00002553958,0.7298108],"study_design_scores_gemma":[0.0001702865,0.0001043014,0.8559509,0.000199134,0.00002647916,0.000005519863,0.008730582,0.1343261,0.0002243136,0.00003217583,0.00006379862,0.0001664357],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.962738,0.0003706042,0.03186206,0.00004812648,0.0004870489,0.0002798129,9.868721e-7,0.0001072261,0.004106131],"genre_scores_gemma":[0.9812225,0.00000131246,0.01790286,0.0002162059,0.0001210912,0.00002295819,0.00002678906,0.00001293303,0.000473423],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7296444,"threshold_uncertainty_score":0.995721,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.4131373920694704,"gpt":0.4947529833668562,"spread":0.08161559129738588,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","sts"],"consensus_categories":[],"category_scores_codex":[0.01112648,0.0004498558,0.001906986,0.002003773,0.001379189,0.0006141363,0.00165349,0.0002960261,0.00008678757],"category_scores_gemma":[0.04069273,0.0003017741,0.0007207623,0.006777404,0.001082182,0.0002723176,0.0001956956,0.001929884,0.00002995275],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003836448,"about_ca_system_score_gemma":0.001954194,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001701645,"about_ca_topic_score_gemma":0.006112929,"domain_scores_codex":[0.9891219,0.005077949,0.002499805,0.0006366553,0.001360876,0.001302875],"domain_scores_gemma":[0.9434631,0.05267152,0.002417709,0.001081338,0.00008517458,0.0002811122],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0000164944,0.0001560857,0.04751999,0.0005016867,0.0000109047,0.00008779835,0.0008503961,0.000001763152,1.64292e-7,0.001644994,0.002096913,0.9471128],"study_design_scores_gemma":[0.00007796227,0.000011146,0.03832578,0.001829655,0.00005106505,0.00000607369,0.00007225524,0.00000202203,7.4909e-7,0.0003904096,0.9589963,0.0002366116],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.004219275,0.9585164,2.28687e-7,0.01123419,0.02240023,0.002147876,0.00001683648,0.0000452414,0.001419708],"genre_scores_gemma":[0.01051392,0.9865433,0.000002094019,0.0002170748,0.001685351,0.0005074731,0.000003483117,0.0000734612,0.0004538512],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.9568993,"threshold_uncertainty_score":0.9999434,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.5681965202413667,"gpt":0.575281995445556,"spread":0.00708547520418934,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.1320133,0.0001846183,0.0009974277,0.505332,0.0003130617,0.0002190296,0.001156311,0.0001317246,0.0003684885],"category_scores_gemma":[0.055937,0.0001014415,0.0001051613,0.710571,0.001044191,0.0004971984,0.0003227755,0.0003447937,0.000003998525],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002059156,"about_ca_system_score_gemma":0.0009180653,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01289675,"about_ca_topic_score_gemma":0.002666965,"domain_scores_codex":[0.9849613,0.004650559,0.002572722,0.0009934698,0.005589648,0.001232341],"domain_scores_gemma":[0.9731256,0.02392163,0.001107604,0.0006741067,0.0004087842,0.0007623082],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.000004810743,0.0001658453,0.8950847,0.0000184103,0.00002090739,6.256835e-7,0.0004353852,0.000006507757,8.579258e-7,0.003511562,0.01688755,0.08386283],"study_design_scores_gemma":[0.0006068487,0.0004452606,0.9790449,0.000007198271,0.000006339485,0.000008834256,0.00270815,0.003997907,1.261855e-7,0.004403072,0.008692787,0.00007855061],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.832119,0.02280031,0.04052914,0.1024223,0.0001363002,0.0006649686,0.0001253394,0.00001176209,0.001190918],"genre_scores_gemma":[0.9933293,0.00289985,0.0009440565,0.002478968,0.0001010969,0.000021872,0.00008669411,0.000005454859,0.0001326834],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.205239,"threshold_uncertainty_score":0.9936764,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.03018464284592223,"gpt":0.3520499505746199,"spread":0.3218653077286976,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0002604248,0.0001109124,0.0003260988,0.0278868,0.00005438256,0.0000187593,0.0002303244,0.0001302725,0.0003660999],"category_scores_gemma":[0.00007173356,0.000123523,0.00009059884,0.2014352,0.00006122238,0.00004723212,0.00002837167,0.0002747946,0.000007962923],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002755231,"about_ca_system_score_gemma":0.00008914519,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003586786,"about_ca_topic_score_gemma":0.0001034553,"domain_scores_codex":[0.9988654,0.0001038576,0.0002751215,0.0002397687,0.0001521925,0.0003636375],"domain_scores_gemma":[0.9992057,0.0004126834,0.00002640361,0.0001775552,0.0001162158,0.00006140934],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"observational","study_design_scores_codex":[0.00008907481,0.0004014036,0.3163919,0.0001110367,0.003868846,0.00007347583,0.0004620591,0.02278758,0.1451975,0.001739842,0.3652075,0.1436698],"study_design_scores_gemma":[0.002634053,0.0001826089,0.6421359,0.00004641209,0.001303572,0.00003717294,0.0008853532,0.01354625,0.1594193,0.001986785,0.1769356,0.000886954],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.971692,0.002922677,0.01334724,0.0004682327,0.0009759101,0.00008730296,0.0000743218,0.00006753938,0.01036475],"genre_scores_gemma":[0.9920067,0.001204295,0.006192102,0.0002548154,0.00006113321,0.00004486427,0.00004241993,0.00001302098,0.0001806542],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.325744,"threshold_uncertainty_score":0.9831313,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Science North","funders":"","keywords":"Computer science; Citation; Science Citation Index; Information retrieval; Basic research; Data science; Library science","retraction":null,"screen_n_in":null,"score":{"opus":0.9272708597111224,"gpt":0.7093757539334359,"spread":0.2178951057776866,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.2057137,0.000154572,0.0003276869,0.1777989,0.001536793,0.004603872,0.007405369,0.0002101508,0.000207557],"category_scores_gemma":[0.3083346,0.00008479669,0.0002015976,0.7157363,0.002311401,0.0006741668,0.001195092,0.000835082,0.0006112434],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005187356,"about_ca_system_score_gemma":0.001055703,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001161045,"about_ca_topic_score_gemma":0.00001303946,"domain_scores_codex":[0.9392418,0.0007403781,0.001669267,0.0009770918,0.05584848,0.00152296],"domain_scores_gemma":[0.9560933,0.0199717,0.0006784089,0.001840533,0.02019403,0.001222094],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00003475098,0.0008469789,0.02419321,0.000007270738,0.00002577331,0.000007267863,0.0003623342,0.00003383488,0.05391213,0.01981578,0.2646721,0.6360886],"study_design_scores_gemma":[0.001374111,0.0006347309,0.4007756,0.00002622899,0.00000609612,0.00001834893,0.002229139,0.001320611,0.01171471,0.08020875,0.5013958,0.0002959356],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8835268,0.01285592,0.06419225,0.02068269,0.003084413,0.0008100205,0.0003111181,0.00004413114,0.01449268],"genre_scores_gemma":[0.9937416,0.001260756,0.001223136,0.00004754034,0.0001021749,0.00001281361,0.000005308968,0.00001266674,0.003593993],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6357926,"threshold_uncertainty_score":0.9997631,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.07319567428049205,"gpt":0.3701037668851918,"spread":0.2969080926046997,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","bibliometrics","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","bibliometrics","research_integrity"],"category_scores_codex":[0.04268253,0.001217651,0.0120346,0.1384439,0.0002079659,0.00225541,0.008980818,0.001298132,0.0007676836],"category_scores_gemma":[0.02119201,0.0007729624,0.02071055,0.3258841,0.0004566945,0.002088543,0.001004882,0.002528215,0.0003047783],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000598367,"about_ca_system_score_gemma":0.001378812,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001756963,"about_ca_topic_score_gemma":0.00001137841,"domain_scores_codex":[0.9715447,0.005283297,0.006717949,0.002101639,0.01287947,0.001472912],"domain_scores_gemma":[0.9858853,0.003793191,0.002202082,0.00497991,0.002397126,0.0007424452],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00003518858,0.0000664714,0.0008731426,0.0003043532,0.002827369,0.0009957249,0.000066675,0.0001731299,5.656618e-8,0.00001960523,0.004724405,0.9899139],"study_design_scores_gemma":[0.001466165,0.0002082609,0.01313324,0.005164119,0.01053896,0.0007108742,0.00005548212,0.0001121775,8.151576e-7,0.0005587186,0.9671428,0.00090834],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.0005714177,0.9949289,0.0008326631,0.00009449187,0.001090172,0.0008819717,0.00009133458,0.00003323421,0.001475831],"genre_scores_gemma":[0.0003822884,0.9952545,0.0008102618,0.0000685058,0.001171305,0.00008016746,0.000004719374,0.0001180585,0.002110183],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.9890056,"threshold_uncertainty_score":0.9999984,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Impact; OpenAlex","funders":"","keywords":"Altmetrics; Citation; Social media; Computer science; Data science; World Wide Web","retraction":null,"screen_n_in":null,"score":{"opus":0.1260122823245402,"gpt":0.4456347829134141,"spread":0.3196225005888739,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.01265662,0.00007769094,0.000201619,0.007162035,0.0006609843,0.0004769817,0.003048559,0.00005126933,0.00002821903],"category_scores_gemma":[0.04292988,0.00003601714,0.0001650493,0.1185164,0.006011039,0.0003109757,0.0007805343,0.00014545,0.00005126737],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004896835,"about_ca_system_score_gemma":0.0002077043,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00007433892,"about_ca_topic_score_gemma":3.235553e-7,"domain_scores_codex":[0.9947464,0.00002273687,0.0005795477,0.0001908659,0.004103722,0.0003566625],"domain_scores_gemma":[0.9883602,0.002946296,0.00099154,0.0007659908,0.006858831,0.00007717804],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0000163051,0.00004657764,0.006585426,0.000006464713,0.00001745575,1.217617e-8,0.0003659671,0.00001143922,0.00144955,0.03773147,0.1996507,0.7541186],"study_design_scores_gemma":[0.0004072957,0.0008349729,0.06514032,0.000008471907,0.00000532742,0.000004289449,0.01588363,0.002124711,0.008306456,0.01330125,0.8938459,0.0001374221],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9379225,0.0001204607,0.001040615,0.05484475,0.0002138291,0.0007350143,0.0000245451,0.00002084219,0.00507744],"genre_scores_gemma":[0.9962651,0.0001995271,0.002163837,0.001040637,0.000004491798,0.00003525451,1.758055e-7,0.000002184204,0.0002887659],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7539812,"threshold_uncertainty_score":0.996694,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.5726813817650799,"gpt":0.5567807477511177,"spread":0.01590063401396213,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.06158305,0.0001795445,0.0003808583,0.2483297,0.0002090949,0.003611983,0.001883733,0.0002070862,0.0002956545],"category_scores_gemma":[0.1921769,0.0001440193,0.00009545132,0.5391781,0.0002872671,0.003999108,0.0009399208,0.0003335925,0.0001327346],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003777984,"about_ca_system_score_gemma":0.0004307377,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008616144,"about_ca_topic_score_gemma":0.0006853588,"domain_scores_codex":[0.9797834,0.0002406213,0.001294367,0.00108498,0.01684294,0.0007536688],"domain_scores_gemma":[0.9724427,0.02094738,0.0006786906,0.0007354334,0.004372299,0.0008234766],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00005946088,0.00004929365,0.8260859,0.000006706121,0.000004549861,0.000005882124,0.0008868973,0.00003081033,0.01218201,0.0002786964,0.001110754,0.1592991],"study_design_scores_gemma":[0.0007069465,0.0001980155,0.9824387,0.0000535971,0.000005681039,0.000001966001,0.0008031926,0.001849006,0.005668396,0.0061341,0.001961292,0.0001791179],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9728765,0.00912236,0.01575318,0.0004166511,0.0005281832,0.0002774931,0.00002254843,0.00001815375,0.0009849267],"genre_scores_gemma":[0.9885912,0.0006787465,0.009100274,0.0000812181,0.00006168662,0.000002291321,0.000003677136,0.000009933621,0.001471007],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2908484,"threshold_uncertainty_score":0.9974223,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.1136474693055103,"gpt":0.4694544594801149,"spread":0.3558069901746047,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.01540162,0.000130408,0.0003457652,0.05063199,0.0001753469,0.000285279,0.001128622,0.0001924777,0.00009330339],"category_scores_gemma":[0.03425495,0.00006740615,0.00003072491,0.4314651,0.0005514974,0.000243503,0.0004486114,0.0003088981,0.000007501409],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003298815,"about_ca_system_score_gemma":0.0005209459,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.05370288,"about_ca_topic_score_gemma":0.07347152,"domain_scores_codex":[0.9953133,0.000356166,0.0006361364,0.0004599351,0.002798569,0.0004358876],"domain_scores_gemma":[0.9895739,0.00773305,0.0003446456,0.0006007021,0.001624832,0.0001228694],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.00002286156,0.00003863651,0.2841277,0.00002394552,0.000005109296,0.00001150405,0.00006510794,0.0009922805,0.0009183518,0.001664325,0.001864912,0.7102652],"study_design_scores_gemma":[0.001534187,0.002950004,0.740337,0.0000499169,0.00001660216,0.0001227433,0.001751652,0.150799,0.09668067,0.00259415,0.002805426,0.0003586424],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9970747,0.0008125238,0.00008144907,0.0005073469,0.0003035494,0.0005588941,0.00002506945,0.000007662398,0.0006287817],"genre_scores_gemma":[0.9995803,0.0001863151,0.0001270948,0.00001459261,0.00001331321,0.00001261571,0.000001521015,0.000004052707,0.00006016437],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7099066,"threshold_uncertainty_score":0.9738799,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Autism; Scopus; Citation; Bibliometrics; Library science; MEDLINE; Psychology; Medicine; Political science; Psychiatry; Computer science","retraction":null,"screen_n_in":null,"score":{"opus":0.4784403492772248,"gpt":0.5618977958917807,"spread":0.08345744661455584,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.03899785,0.000180673,0.0003083669,0.3939531,0.0008376112,0.001906356,0.006065951,0.0001414123,0.001307297],"category_scores_gemma":[0.03707615,0.00008099148,0.0002886065,0.8815393,0.001591571,0.0004464787,0.00247573,0.0003585799,0.002252445],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005054498,"about_ca_system_score_gemma":0.000639227,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000173269,"about_ca_topic_score_gemma":0.0001276565,"domain_scores_codex":[0.9763728,0.0006556888,0.0007387757,0.00120013,0.01976923,0.001263347],"domain_scores_gemma":[0.9901704,0.005407088,0.0002986104,0.002527643,0.001157677,0.0004385597],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00005557661,0.0003429249,0.4506551,0.000008921988,0.00001747219,0.000004505324,0.00003886722,0.00001290126,0.00152348,0.105715,0.01915295,0.4224723],"study_design_scores_gemma":[0.0004621007,0.0002183973,0.8180109,0.0000357479,0.000002111298,0.000002061734,0.00004368976,0.0001724499,0.00533789,0.1190238,0.05655034,0.0001405248],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9342546,0.0006031579,0.0006913149,0.01473945,0.002806951,0.0006004771,0.0001227881,0.00003554707,0.04614574],"genre_scores_gemma":[0.989234,0.0001124456,0.000137496,0.00002322455,0.00006275977,0.0000181185,3.467463e-7,0.00001269869,0.01039887],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4875862,"threshold_uncertainty_score":0.9996057,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":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":"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,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.3203970563219408,"gpt":0.5196480824640306,"spread":0.1992510261420898,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","research_integrity"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.04110247,0.0002156074,0.001152638,0.1064811,0.0005732693,0.0003613488,0.00128925,0.0003667845,0.0001502543],"category_scores_gemma":[0.0291508,0.0001627095,0.0005655728,0.3508556,0.0005455956,0.001370893,0.0002860553,0.002417597,0.000003105943],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000985107,"about_ca_system_score_gemma":0.0008525269,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001123867,"about_ca_topic_score_gemma":0.00002528011,"domain_scores_codex":[0.9883673,0.001370294,0.002349667,0.0008597886,0.006182096,0.0008709243],"domain_scores_gemma":[0.9912344,0.001433869,0.001256221,0.0005797195,0.005189528,0.0003063221],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","study_design_scores_codex":[0.02848316,0.003108722,0.2966217,0.0009525804,0.003731144,0.0004004899,0.05103545,0.3203127,0.001257799,0.004140698,0.009231633,0.280724],"study_design_scores_gemma":[0.003850276,0.003542077,0.4315703,0.0009194021,0.001747253,0.0001867953,0.3829671,0.06722584,0.008901374,0.09052077,0.007530911,0.001037944],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9345188,0.0006522113,0.002657295,0.06078568,0.000103792,0.0006583066,0.00001117703,0.00001247647,0.0006003116],"genre_scores_gemma":[0.9980719,0.0003744643,0.0006469823,0.00002365743,0.000522287,0.00001464898,0.000004570065,0.00002562179,0.0003158404],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3319317,"threshold_uncertainty_score":0.9998839,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.50896915729909,"gpt":0.496626232536117,"spread":0.01234292476297294,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002881206,0.00009728035,0.000231788,0.0007436593,0.0005672697,0.0004536173,0.001341408,0.000009673142,0.0001286192],"category_scores_gemma":[0.00003649474,0.00007493598,0.0000300079,0.002437967,0.001170432,0.0008551654,0.001680405,0.0001352011,3.589174e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001150339,"about_ca_system_score_gemma":0.0001372697,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005020816,"about_ca_topic_score_gemma":0.001261922,"domain_scores_codex":[0.9978277,0.0001657172,0.0003713213,0.0003587165,0.0009465833,0.0003299293],"domain_scores_gemma":[0.9984764,0.0001961748,0.0002840087,0.0004961622,0.0004572796,0.00009000579],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"qualitative","study_design_scores_codex":[0.00006484748,0.00308025,0.9437094,0.00003387287,0.00008888314,0.00001453184,0.01517381,0.00002697074,0.01169206,0.02102828,0.00006047397,0.005026561],"study_design_scores_gemma":[0.002666369,0.004180921,0.08483757,0.0008741533,0.0001098891,0.00001165853,0.7495735,0.0130032,0.1070031,0.03728257,0.00001963811,0.0004374775],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.967214,0.000001843534,0.03166658,0.000146102,0.00001781885,0.0004348508,0.000005407892,0.00001310858,0.0005002314],"genre_scores_gemma":[0.9992802,7.697605e-7,0.0006444073,0.000006315426,0.00003245316,0.00001299439,0.000001083108,0.000007875936,0.00001390155],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8588719,"threshold_uncertainty_score":0.7590005,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":true},"ca_institutions":"","funders":"","keywords":"The Internet; Health care; Medical education; Public relations; Psychology; Political science; Medicine; Computer science; World Wide Web","retraction":null,"screen_n_in":null,"score":{"opus":0.02382220927478649,"gpt":0.3358496085904542,"spread":0.3120273993156677,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0003997959,0.0001344711,0.0003505488,0.01720912,0.00009814298,0.000282092,0.0005544354,0.00006726233,0.00002330116],"category_scores_gemma":[0.000113897,0.000133596,0.0001334274,0.1999624,0.00003192812,0.0003498138,0.0002473659,0.0001491038,0.00003461628],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002045248,"about_ca_system_score_gemma":0.0004466067,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001153018,"about_ca_topic_score_gemma":0.00161077,"domain_scores_codex":[0.998121,0.000183979,0.0004087994,0.000515863,0.0003694253,0.0004009823],"domain_scores_gemma":[0.9983596,0.00009285994,0.0001349917,0.001040688,0.0001479977,0.0002239171],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.000005564642,0.0002737535,0.004391067,0.0000431374,0.0001011598,0.00003046668,0.003429202,0.001036822,0.0001538561,0.02288178,0.003755168,0.963898],"study_design_scores_gemma":[0.0004020096,0.0001980301,0.9141215,0.00001732193,0.00001568433,0.00001330989,0.0005651318,0.006647089,0.0004843673,0.0006257985,0.07662349,0.0002862894],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6873318,0.005078935,0.2810203,0.01975564,0.0002795986,0.0007343678,0.00006941456,0.0004423831,0.005287582],"genre_scores_gemma":[0.9956751,0.0004312768,0.002469245,0.001044261,0.00002932921,0.00005607043,0.00003597022,0.000006865563,0.0002518538],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9636117,"threshold_uncertainty_score":0.99393,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.06917925277951521,"gpt":0.2601733931358076,"spread":0.1909941403562924,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0003681551,0.0001660633,0.000236797,0.003211268,0.0005011857,0.00016423,0.0004007736,0.0000781518,0.003433934],"category_scores_gemma":[0.00009996556,0.00007729125,0.0001076842,0.0464882,0.0001205794,0.000041362,0.0004093731,0.0002891849,0.00006041222],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000481114,"about_ca_system_score_gemma":0.0000161218,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002268656,"about_ca_topic_score_gemma":0.0002229432,"domain_scores_codex":[0.9977115,0.0002588572,0.0002321549,0.0005828984,0.000781408,0.0004332217],"domain_scores_gemma":[0.9992512,0.0002303694,0.00005204246,0.0001446135,0.0001044068,0.0002173344],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0002001655,0.0005482693,0.02918538,0.00001219315,0.000366497,0.000008624265,0.000194445,0.0005486761,0.4165229,0.0009934427,0.0003937434,0.5510257],"study_design_scores_gemma":[0.0001954452,0.002405072,0.9925503,0.000001379183,0.0000974358,0.000005755137,0.0001950639,0.0006328278,0.001116138,0.0004268719,0.002156661,0.0002170845],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9973544,0.0002454665,0.00006965765,0.0006700786,0.00003831536,0.0002754508,0.0001225761,0.00005762563,0.001166388],"genre_scores_gemma":[0.998988,0.0001833858,0.0002669342,0.000212081,0.00004474976,0.00007111789,0.00006240212,0.000002184008,0.000169195],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9633649,"threshold_uncertainty_score":0.9974771,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.1017841299482366,"gpt":0.4854889908538,"spread":0.3837048609055633,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.002640399,0.0005867099,0.002343807,0.03713749,0.001100358,0.0002152784,0.001375619,0.0004921703,0.00007196912],"category_scores_gemma":[0.00009991803,0.0004221059,0.0008121946,0.2687443,0.0001709894,0.00009748698,0.0002304117,0.00111995,0.0004929064],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003214654,"about_ca_system_score_gemma":0.0007222631,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005819532,"about_ca_topic_score_gemma":0.0004696405,"domain_scores_codex":[0.9921075,0.002383159,0.001514773,0.00129828,0.0009916001,0.001704684],"domain_scores_gemma":[0.9971878,0.0002035759,0.0006695057,0.001113535,0.00009752504,0.000728045],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001046539,0.0001724919,0.005086704,0.0005217001,0.0007724696,0.00001262777,0.00003826597,7.489074e-8,8.967885e-11,0.001152302,0.4775974,0.5145413],"study_design_scores_gemma":[0.0007802282,0.0004057036,0.003307063,0.001370203,0.000486119,0.00002248571,0.0003587951,0.000005274717,2.169181e-9,0.0006750018,0.992275,0.0003141054],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.00001702305,0.9554495,0.00104963,0.01003031,0.02732607,0.00420606,0.0006012734,0.00006729892,0.001252803],"genre_scores_gemma":[0.00005586935,0.9946618,0.001269746,0.0007832319,0.0007151371,0.0008837787,0.0006593635,0.00004958612,0.0009214786],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.5146776,"threshold_uncertainty_score":0.9998231,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.9556517327164997,"gpt":0.775290194182347,"spread":0.1803615385341527,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.7603914,0.0002651678,0.004306586,0.01549014,0.000739342,0.0006502793,0.003303014,0.0002614071,0.00267452],"category_scores_gemma":[0.01374474,0.0001564081,0.0004174438,0.03444181,0.0009398926,0.0006705663,0.001319388,0.00144384,0.00009174114],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002758441,"about_ca_system_score_gemma":0.003603311,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.09955664,"about_ca_topic_score_gemma":0.08406115,"domain_scores_codex":[0.6852011,0.2546909,0.01789029,0.002529951,0.03725348,0.002434235],"domain_scores_gemma":[0.9145306,0.04247639,0.006662241,0.005610083,0.02789819,0.002822514],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0001472882,0.0005117475,0.7854776,0.008146445,0.0004968208,0.000003619002,0.1692538,0.000007783848,0.00003862888,0.01315886,0.0008042309,0.02195315],"study_design_scores_gemma":[0.0003744807,0.000427282,0.8713043,0.0007982214,0.00002304416,0.000004368301,0.07939598,0.02915843,0.00002758278,0.01737408,0.001000798,0.0001113694],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.970503,0.01570574,0.000222782,0.01011242,0.00002638004,0.0018064,0.00007647666,0.000003501179,0.001543299],"genre_scores_gemma":[0.9896191,0.005526877,0.00368284,0.0004316442,0.0001044479,0.0001166938,0.00002642119,0.00001726286,0.0004746533],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7466466,"threshold_uncertainty_score":0.9982372,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.02074527027169123,"gpt":0.3570799097212898,"spread":0.3363346394495986,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005380032,0.0001139127,0.000457509,0.005640474,0.00001945641,0.00002267104,0.00003715817,0.00004141489,0.000001988056],"category_scores_gemma":[0.0007586628,0.0001147794,0.00007116684,0.009006096,0.00004710604,0.000155056,0.00001288581,0.00009151249,2.796351e-8],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001188902,"about_ca_system_score_gemma":0.00001575422,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00000117968,"about_ca_topic_score_gemma":0.000001225298,"domain_scores_codex":[0.9988799,0.000196772,0.0006053554,0.0001353549,0.0001091307,0.00007347235],"domain_scores_gemma":[0.9974486,0.002133936,0.0001383138,0.00007623198,0.0001746415,0.0000282345],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00001536596,0.00003453597,0.0003004888,0.170435,0.001121092,0.000001337991,0.0115928,0.4525576,0.04261892,0.1439582,0.000005327972,0.1773593],"study_design_scores_gemma":[0.0002722334,0.00002791996,0.005494426,0.01571053,0.0002465266,0.00003406019,0.0008536051,0.9572991,0.004603999,0.01532417,0.000003704349,0.0001297966],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01385481,0.03190194,0.9537891,0.00002792705,0.00009061061,0.0002325552,0.00002406237,0.00003496818,0.0000440107],"genre_scores_gemma":[0.2208469,0.003025054,0.7760086,0.000004969195,0.000006782459,0.00004546762,0.00005241323,0.000008329405,0.000001532398],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.5047414,"threshold_uncertainty_score":0.5032936,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.02514335089884866,"gpt":0.2437797157616595,"spread":0.2186363648628108,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001519732,0.0002832051,0.000573695,0.003226763,0.0004040634,0.0002176717,0.001101595,0.00006802844,0.000624358],"category_scores_gemma":[0.001411377,0.0001222814,0.0007037119,0.01666921,0.000557454,0.0005113013,0.0002760888,0.0004036151,0.000005205779],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006241378,"about_ca_system_score_gemma":0.00004396886,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003242604,"about_ca_topic_score_gemma":0.00003937726,"domain_scores_codex":[0.9972329,0.00002090768,0.001168237,0.000229889,0.00104121,0.0003068276],"domain_scores_gemma":[0.9955868,0.0007388997,0.001542264,0.0002508588,0.001864831,0.0000163282],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.01500214,0.005121025,0.06216661,0.003410588,0.04373755,0.00001303527,0.1815277,0.007344487,0.03188865,0.3161114,0.2070595,0.1266173],"study_design_scores_gemma":[0.009141067,0.006609961,0.3506894,0.003965783,0.03835298,0.000192059,0.2495855,0.1425439,0.1325788,0.01465353,0.04844442,0.003242631],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9939413,0.0004839922,0.00009238772,0.001665671,0.0004222798,0.0005029438,0.00001033361,0.000005345714,0.002875732],"genre_scores_gemma":[0.9973821,0.00009089664,0.0000314825,0.0006783759,0.0002448067,0.000008567539,0.000002077641,0.0000153562,0.001546344],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3014578,"threshold_uncertainty_score":0.8009002,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":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":"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,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.02511664582110678,"gpt":0.2281311560538719,"spread":0.2030145102327651,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.002089577,0.0002639125,0.0008704378,0.01184082,0.0001115637,0.0002036348,0.0003018386,0.00005463489,0.00001660567],"category_scores_gemma":[0.0006974782,0.0002221601,0.0001089928,0.01618482,0.00004753171,0.001013886,0.0004569833,0.000253632,0.000008529476],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006184382,"about_ca_system_score_gemma":0.0000122802,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004221642,"about_ca_topic_score_gemma":0.00001388331,"domain_scores_codex":[0.9978995,0.00005384437,0.0008776168,0.0003166584,0.000441164,0.0004112502],"domain_scores_gemma":[0.9981534,0.0000560202,0.001316802,0.0001878315,0.0002471408,0.00003875201],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000189862,0.0001897926,0.002778692,0.5228189,0.00008789528,0.002766162,0.0005239741,0.0001132307,3.535915e-7,0.1435121,0.009053453,0.3179656],"study_design_scores_gemma":[0.01010961,0.0007217634,0.3966501,0.0715977,0.004601655,0.00007503261,0.0232074,0.002346513,0.000001657436,0.04991242,0.4386271,0.00214912],"study_design_candidate":"systematic_review","study_design_consensus":null,"genre_codex":"review","genre_gemma":"empirical","genre_scores_codex":[0.06427904,0.890828,0.004222556,0.01568619,0.0005661429,0.003944526,0.000004413536,0.0001143603,0.02035474],"genre_scores_gemma":[0.7038795,0.2922593,0.0002622688,0.002795586,0.0003588514,0.00004552207,0.000002153596,0.00002771427,0.0003691037],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.6396005,"threshold_uncertainty_score":0.9993591,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.2288541884787154,"gpt":0.414867894221078,"spread":0.1860137057423626,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.00168868,0.0001221811,0.0002831788,0.009312577,0.0005436088,0.00008561133,0.0001179145,0.0001126255,0.0001343952],"category_scores_gemma":[0.02116254,0.00009759446,0.00006454562,0.0271212,0.0005078531,0.0003318294,0.00004994674,0.0001710369,0.000007358668],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005894629,"about_ca_system_score_gemma":0.0002088982,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00007729436,"about_ca_topic_score_gemma":0.0001089044,"domain_scores_codex":[0.9974291,0.0006821209,0.0004951618,0.0002787593,0.0007548959,0.0003599648],"domain_scores_gemma":[0.9776748,0.02139642,0.0003821083,0.0001059303,0.0001693175,0.0002714007],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00001617298,0.00005133623,0.9503936,0.00002279337,0.000005277935,0.000004072523,0.002545135,7.417694e-9,0.00007332437,0.001814473,0.003165202,0.04190862],"study_design_scores_gemma":[0.0002073464,0.00001318747,0.9660642,0.00003373876,0.00001882531,0.000002111896,0.0001767836,7.43548e-7,0.0001194876,0.001781686,0.03147894,0.0001029173],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9824025,0.002080314,0.000005151064,0.01245017,0.002004493,0.0003803418,0.00002082273,0.0000430386,0.0006131802],"genre_scores_gemma":[0.9952462,0.003595469,0.00004900122,0.0002285765,0.0006401542,0.0000811298,8.770802e-7,0.00001913568,0.0001394316],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.04180571,"threshold_uncertainty_score":0.9935578,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.02573323452045045,"gpt":0.3215061265589551,"spread":0.2957728920385046,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.003438634,0.0004384904,0.001630144,0.05847432,0.000150587,0.0002366697,0.0005605132,0.0004250808,0.0001884839],"category_scores_gemma":[0.001342221,0.0003032245,0.001462874,0.1742324,0.0002349188,0.0004335475,0.00003124724,0.0009126042,0.00003135129],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007674842,"about_ca_system_score_gemma":0.0002455827,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002255991,"about_ca_topic_score_gemma":0.00006621662,"domain_scores_codex":[0.9947191,0.0006431704,0.00123351,0.0006447286,0.002180091,0.0005794124],"domain_scores_gemma":[0.9967256,0.0007292223,0.0004665597,0.0009918525,0.0008048266,0.0002818917],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0003235916,0.0005848004,0.9401734,0.00004394752,0.0009388709,0.001198346,0.001817182,0.0005955116,0.000499629,0.01173164,0.0009543332,0.04113874],"study_design_scores_gemma":[0.001235856,0.0002739415,0.9781672,0.0000739312,0.001277155,0.000405314,0.0002649428,0.002956693,0.0001277092,0.0003097529,0.01459809,0.0003094019],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9864461,0.001762378,0.00004508545,0.007072296,0.00003955865,0.0005934238,0.00001374151,0.00006446766,0.003962895],"genre_scores_gemma":[0.9945927,0.001244191,0.000936535,0.002492719,0.0001967187,0.0000587781,0.0001759621,0.00004471833,0.000257695],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.115758,"threshold_uncertainty_score":0.999942,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":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":"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,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.0310087613957104,"gpt":0.2318401739998235,"spread":0.2008314126041131,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001072792,0.00006005938,0.0002571358,0.0002859612,0.0008732048,0.0001560613,0.0003550833,0.00003453953,0.000004275299],"category_scores_gemma":[0.0003524128,0.00004792362,0.0002128551,0.001812958,0.0008726004,0.000965997,0.00006146256,0.00008812766,0.000001464148],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008569341,"about_ca_system_score_gemma":0.0001165635,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009359526,"about_ca_topic_score_gemma":0.000003922845,"domain_scores_codex":[0.9991305,0.000002446994,0.0005136212,0.000105148,0.00008876406,0.0001595034],"domain_scores_gemma":[0.9978786,0.00005644158,0.001487792,0.0001552064,0.0003848953,0.00003708077],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003051458,0.0001312238,0.06967558,0.0001372493,0.001119571,3.138792e-7,0.03386844,0.01983586,0.002303296,0.3422149,0.1094637,0.4209447],"study_design_scores_gemma":[0.002054409,0.000470893,0.0150989,0.0000282179,0.00007378309,0.00009097741,0.03319187,0.6019155,0.0007169732,0.04247516,0.3034185,0.0004648231],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3088748,0.0002827444,0.6887603,0.001230868,0.0003232236,0.0002206609,0.00001678866,0.00001259398,0.0002779219],"genre_scores_gemma":[0.9555224,0.00006551385,0.04399395,0.0001791787,0.000117361,0.000007219822,0.000002183571,0.000003231841,0.0001090114],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.6466475,"threshold_uncertainty_score":0.6716074,"prediction_status":"machine_predicted_unvalidated"},"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":"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,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"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","retraction":null,"screen_n_in":null,"score":{"opus":0.02572536010457595,"gpt":0.2882131049806921,"spread":0.2624877448761162,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0003885244,0.0005075192,0.004415213,0.003041571,0.00004094044,0.00002697742,0.0004433546,0.0002411441,0.0002997922],"category_scores_gemma":[0.0004311411,0.0004206847,0.0008753425,0.002899694,0.0002263087,0.00005989675,0.0002507453,0.0002513865,0.000003911387],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000905303,"about_ca_system_score_gemma":0.00004910347,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004906073,"about_ca_topic_score_gemma":0.000001768387,"domain_scores_codex":[0.99756,0.0001612464,0.001196168,0.0004081957,0.0003368164,0.0003375343],"domain_scores_gemma":[0.9981073,0.0004061248,0.0007226646,0.00063629,0.00005082888,0.00007676092],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","study_design_scores_codex":[2.591509e-7,0.000009627563,3.213681e-7,0.7483926,0.002428979,0.00000479765,0.00001807359,0.000001834793,0.000001851098,0.000006790389,0.00001376001,0.249121],"study_design_scores_gemma":[0.0001167141,0.00002582778,0.000010822,0.9396553,0.02363137,0.00002233399,0.0001290061,0.00003372183,0.02843518,0.000001268332,0.007195744,0.0007427023],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.00005613932,0.9976858,0.0003713086,0.000001678256,0.0002198644,0.0005797304,0.000690732,0.0001924825,0.0002022417],"genre_scores_gemma":[0.001031702,0.9981333,0.0004017982,0.000004319309,0.00001739573,0.00008149126,0.0001946438,0.00005536892,0.00007999926],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.2483783,"threshold_uncertainty_score":0.9998245,"prediction_status":"machine_predicted_unvalidated"},"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"}]}