{"meta":{"query_hash":"ac8d0f485ba1","filters":{"venue":"Scientometrics"},"cohort_total":176,"direct_labels_cover":44,"predictions_cover":176,"exported":176,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/ac8d0f485ba1","api":"https://metacan.xera.ac/api/v1/cohort?venue=Scientometrics"},"results":[{"id":"W1172433429","doi":"10.1007/s11192-015-1660-0","title":"Global food allergy research trend: a bibliometric analysis","year":2015,"lang":"en","type":"article","venue":"Scientometrics","topic":"Food Allergy and Anaphylaxis Research","field":"Medicine","cited_by":23,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Science Citation Index; Social Sciences Citation Index; Bibliometrics; Web of science; Scientometrics; Citation; Food allergy; Scientific literature; Citation analysis; Population; Citation index; Environmental health; Social science; Library science; Medicine; MEDLINE; Political science; Allergy; Sociology; Computer science","score_opus":0.25255445713104036,"score_gpt":0.46594745442060165,"score_spread":0.2133929972895613,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1172433429","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9082082,0.007080885,0.0006249791,0.0009984091,0.00032991284,0.000271483,0.000060601946,0.00012205864,0.08230347],"genre_scores_gemma":[0.99472904,0.00031237633,0.00086905505,0.00011318116,0.0001726985,0.00001693616,0.000060590934,0.000014761343,0.003711387],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9912878,0.00022093201,0.00035349716,0.0007010906,0.0062698307,0.0011667988],"domain_scores_gemma":[0.9952663,0.00027158982,0.00006308222,0.00078726927,0.0019631637,0.0016486023],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0064672697,0.00017961602,0.0004364953,0.47173247,0.00022236872,0.0002747119,0.00062217016,0.0001775466,0.00028283813],"category_scores_gemma":[0.007369693,0.00014819793,0.0003091,0.9206145,0.0003478532,0.00020550258,0.00041053855,0.0004184983,0.00031397148],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011924028,0.004121839,0.4486575,0.0001429263,0.0065085175,0.00050717226,0.00078774855,0.001093538,0.00010290104,0.013865805,0.1824174,0.34060222],"study_design_scores_gemma":[0.0033355118,0.004319129,0.6523522,0.000017200715,0.000052122497,0.00003720994,0.000631116,0.0032411062,0.00013398015,0.0013682744,0.33419216,0.00031995686],"about_ca_topic_score_codex":0.00028493695,"about_ca_topic_score_gemma":0.0005944821,"teacher_disagreement_score":0.448882,"about_ca_system_score_codex":0.000626227,"about_ca_system_score_gemma":0.0006566505,"threshold_uncertainty_score":0.8822743},"labels":[],"label_agreement":null},{"id":"W1724869758","doi":"10.1023/a:1014256102041","title":"Collaboration and Cognitive Structures in Social Science Research Fields. Towards Socio-Cognitive Analysis in Information Systems","year":2001,"lang":"en","type":"article","venue":"Scientometrics","topic":"Knowledge Management and Sharing","field":"Social Sciences","cited_by":43,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Mainstream; Consolidation (business); Cognition; Field (mathematics); Sociology; Value (mathematics); Social network analysis; Knowledge management; Epistemology; Social science; Computer science; Psychology; Political science; Business; Social capital","score_opus":0.09573227246461083,"score_gpt":0.44943950812195993,"score_spread":0.3537072356573491,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1724869758","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.88329846,0.0001846134,0.0005126857,0.00019873948,0.000229792,0.00045383457,0.000008257317,0.000018917255,0.11509472],"genre_scores_gemma":[0.9993936,0.000112290916,0.000017639848,0.000029698713,0.00008288825,0.00002116191,0.000010905481,0.0000021323945,0.00032972186],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9969449,0.00026432282,0.00026032238,0.0002679544,0.0017488993,0.0005135717],"domain_scores_gemma":[0.9980844,0.00031250942,0.000087659326,0.000055787084,0.0013671606,0.00009247985],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.009587261,0.00007555395,0.00016276387,0.013945316,0.0011210191,0.0013179001,0.00027218673,0.00010289104,0.000023031142],"category_scores_gemma":[0.005288403,0.00007779912,0.00002486976,0.088501714,0.00080137316,0.0019861222,0.00017664299,0.00022150583,0.000010791246],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000037105674,0.00005849356,0.6752026,0.00003205007,0.00003249327,0.0000073024967,0.18902326,0.00005790975,0.0000058819314,0.04560138,0.00010158243,0.08983993],"study_design_scores_gemma":[0.00058834575,0.000032815424,0.79225373,0.000027577642,0.000038033973,1.2501476e-7,0.20112053,0.0031734267,0.000015249943,0.0014381902,0.0011476729,0.00016428217],"about_ca_topic_score_codex":0.0021985674,"about_ca_topic_score_gemma":0.0035517197,"teacher_disagreement_score":0.11705114,"about_ca_system_score_codex":0.00045410387,"about_ca_system_score_gemma":0.00037472718,"threshold_uncertainty_score":0.99971884},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1766971555","doi":"10.1023/a:1016013912188","title":"Informetric analysis of a music database","year":2002,"lang":"en","type":"article","venue":"Scientometrics","topic":"Scientific Research and Discoveries","field":"Physics and Astronomy","cited_by":10,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Zipf's law; Computer science; Poisson distribution; Information retrieval; Equivalence (formal languages); Informetrics; Database; Data mining; Statistics; Mathematics; Bibliometrics","score_opus":0.08002575222830237,"score_gpt":0.32166463378928656,"score_spread":0.24163888156098418,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1766971555","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8791011,0.00026618483,0.0076649785,0.000025487869,0.0002602391,0.0001228896,0.00048787627,0.000016521615,0.112054735],"genre_scores_gemma":[0.99565965,0.000006578233,0.00041660128,0.0000119877795,0.000028428667,0.000004374548,0.00008341619,0.000003113548,0.003785873],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9981932,0.000017212462,0.0002206262,0.00022307686,0.0010259687,0.0003199233],"domain_scores_gemma":[0.9988667,0.00015245045,0.00010182027,0.00043943856,0.00025826285,0.0001813795],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00066544546,0.00007611402,0.00018290356,0.010690344,0.00009747974,0.00014636696,0.00036930788,0.0000131430415,0.008052825],"category_scores_gemma":[0.00036095895,0.000062784726,0.00014518373,0.106951036,0.0001541643,0.00040686165,0.00014260874,0.000067036264,0.00013254356],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000063912757,0.0006249488,0.8541872,0.00002305439,0.00069139304,0.0000020278746,0.0010058586,0.00037981145,0.0009634753,0.019211112,0.042918857,0.079985894],"study_design_scores_gemma":[0.0037151824,0.000492286,0.5039988,0.00004130023,0.0023049251,0.0000010045399,0.0067263176,0.33950132,0.027062243,0.002156284,0.11224534,0.0017549691],"about_ca_topic_score_codex":0.0001474459,"about_ca_topic_score_gemma":0.000005130658,"teacher_disagreement_score":0.35018837,"about_ca_system_score_codex":0.00001742804,"about_ca_system_score_gemma":0.000032850854,"threshold_uncertainty_score":0.99285394},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W1767272795","doi":"10.1007/s11192-015-1765-5","title":"The journal coverage of Web of Science and Scopus: a comparative analysis","year":2015,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":4652,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Scopus; Web of science; Directory; Computer science; Scientometrics; Bibliometrics; Data science; Library science; Information retrieval; Political science; MEDLINE","score_opus":0.6780420299185497,"score_gpt":0.6098436100295971,"score_spread":0.06819841988895259,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1767272795","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97595584,0.003999752,0.0046122884,0.00020417503,0.0007329189,0.0001741945,0.000037939775,0.0000053967206,0.01427749],"genre_scores_gemma":[0.9978505,0.0007961409,0.0008597731,0.000021692464,0.000021083526,0.0000010195303,3.463934e-7,0.0000028411425,0.000446605],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.96428096,0.00041218387,0.0011759534,0.00051766087,0.032911114,0.0007021165],"domain_scores_gemma":[0.94693005,0.0100336075,0.0012141656,0.001055681,0.03948574,0.0012807787],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.17926545,0.00013699303,0.0005580933,0.25405735,0.0007816073,0.002532158,0.004579442,0.000058656515,0.000034977926],"category_scores_gemma":[0.19043386,0.0000739299,0.00015185864,0.8145463,0.005190757,0.0010767195,0.0013767644,0.00027966313,0.000018335248],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001534978,0.0004063345,0.8360892,0.00000975065,0.00023411625,0.0000068952463,0.0038662988,0.0018332454,0.0045858664,0.0050961617,0.0175405,0.13017815],"study_design_scores_gemma":[0.002708204,0.0017921032,0.766818,0.000021351834,0.00026941564,0.00005672234,0.016396187,0.1313935,0.013746417,0.012000176,0.05426313,0.00053478964],"about_ca_topic_score_codex":0.000054865723,"about_ca_topic_score_gemma":0.000024595498,"teacher_disagreement_score":0.56048894,"about_ca_system_score_codex":0.00024674425,"about_ca_system_score_gemma":0.003039995,"threshold_uncertainty_score":0.9985033},"labels":[],"label_agreement":null},{"id":"W1774927437","doi":"10.1023/a:1012761903614","title":"Investigating the concentration within a research community using joint publications and co-authorship via intermediaries","year":2001,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":11,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Quebec Statistical Institute; Université Laval","funders":"","keywords":"Wonder; Field (mathematics); Period (music); Political science; Sociology; Computer science; Operations research; Epistemology; Mathematics; Philosophy","score_opus":0.9143610746626762,"score_gpt":0.6578600446631518,"score_spread":0.2565010299995244,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1774927437","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9726645,0.00052262517,0.017817054,0.0034463578,0.00062736904,0.00056147017,0.000022608447,0.00005157914,0.004286471],"genre_scores_gemma":[0.99563587,0.00009657643,0.002963396,0.0002551333,0.00011031246,0.000021703223,0.000008446481,0.000014578344,0.0008939884],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9741909,0.004210167,0.0012878188,0.0008727388,0.018108575,0.0013297987],"domain_scores_gemma":[0.9336367,0.042077243,0.0010344805,0.0031642043,0.018022709,0.0020646104],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.20395179,0.00019774341,0.0003197948,0.09063404,0.004677109,0.012246966,0.0036195558,0.00018760307,0.0001540988],"category_scores_gemma":[0.3803259,0.00012482706,0.00009228703,0.5311368,0.0045192726,0.0015484529,0.001782836,0.0019430174,0.00011283209],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011161773,0.0002788132,0.8948023,0.000015592826,0.000021555321,0.0000047153185,0.012732099,0.00019895787,0.0070301443,0.012337531,0.008636611,0.06393056],"study_design_scores_gemma":[0.00077392976,0.00036570337,0.5864945,0.00004516,0.000016353495,0.0001360675,0.014907818,0.25502226,0.0039103036,0.1230517,0.014759676,0.00051652745],"about_ca_topic_score_codex":0.0005938893,"about_ca_topic_score_gemma":0.000114592156,"teacher_disagreement_score":0.4405028,"about_ca_system_score_codex":0.00040056213,"about_ca_system_score_gemma":0.0006439393,"threshold_uncertainty_score":0.99818987},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1953706973","doi":"10.1007/s11192-016-2157-1","title":"Neophilia ranking of scientific journals","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":18,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Institute on Aging; University of Illinois at Chicago","keywords":"Ranking (information retrieval); Journal ranking; Information retrieval; Scientific literature; Computer science; Library science; Data science; Political science; Citation; Geology","score_opus":0.7322136034264928,"score_gpt":0.6133548832507849,"score_spread":0.11885872017570787,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1953706973","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95462674,0.0021876637,0.02599094,0.0015964389,0.0042207,0.00028231234,0.000101821825,0.000043757842,0.01094963],"genre_scores_gemma":[0.9864747,0.00018074313,0.0018732687,0.00005303145,0.00011506751,0.000004282975,8.5659644e-7,0.000015671732,0.011282366],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9605085,0.00035390694,0.0017099311,0.0013636119,0.034763012,0.0013010231],"domain_scores_gemma":[0.96909523,0.011056822,0.0010530971,0.0021534045,0.01551424,0.001127233],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.10989885,0.00022311913,0.0005636096,0.3311329,0.00060516875,0.0046070055,0.0061568823,0.00015952627,0.0029112697],"category_scores_gemma":[0.23368925,0.000117548196,0.00038906085,0.6913339,0.00126198,0.0017193832,0.0014450144,0.00022139132,0.0012319804],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016902004,0.00016456137,0.27150992,0.000005237749,0.000013393643,0.000009699492,0.00011183134,0.000008658365,0.04910531,0.0023077675,0.015953306,0.6607934],"study_design_scores_gemma":[0.002352449,0.00039873007,0.68733615,0.000114950424,0.000019723577,0.00003949886,0.00025834888,0.000977283,0.06320393,0.06730294,0.17730926,0.00068673596],"about_ca_topic_score_codex":0.000013861546,"about_ca_topic_score_gemma":0.000004497225,"teacher_disagreement_score":0.66010666,"about_ca_system_score_codex":0.00023739987,"about_ca_system_score_gemma":0.0005451105,"threshold_uncertainty_score":0.9995457},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1965490711","doi":"10.1007/s11192-006-0100-6","title":"Why are hyperlinks to business Websites created? A content analysis","year":2006,"lang":"en","type":"article","venue":"Scientometrics","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":53,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Hyperlink; Competitor analysis; Link analysis; Context (archaeology); Link (geometry); Content analysis; Business information; Business; Sample (material); Computer science; World Wide Web; Marketing; Advertising; Web page; Sociology; Geography","score_opus":0.059049478563344755,"score_gpt":0.2712085113588207,"score_spread":0.21215903279547593,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1965490711","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2898587,0.00033093468,0.7007824,0.0019413006,0.0006153736,0.00026330186,0.000025279547,0.00027260397,0.0059100613],"genre_scores_gemma":[0.9467893,0.000019876508,0.046094794,0.0042563863,0.0001218293,0.000016579159,0.00002286419,0.0000121542835,0.0026661989],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99647886,0.00004412284,0.0005171584,0.00076284294,0.0015486259,0.000648408],"domain_scores_gemma":[0.99638057,0.0002814756,0.00020548394,0.0011401598,0.001673225,0.00031910886],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0015025585,0.00022077186,0.0004001511,0.014996822,0.00026804995,0.001101478,0.0017992944,0.00012276134,0.00005547744],"category_scores_gemma":[0.0021025052,0.00019184001,0.0002148172,0.19440645,0.000080212616,0.0010088377,0.0005689597,0.00015640372,0.0003438093],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017328462,0.00092411233,0.8901268,0.000063740255,0.00020853273,0.000048316055,0.00033503474,0.017305953,0.00086106575,0.016011167,0.055371918,0.018726015],"study_design_scores_gemma":[0.00031082184,0.00008395342,0.90693545,0.00001053038,0.00007965986,0.000004255188,0.000044413362,0.027168388,0.0010207373,0.00028813301,0.063674636,0.0003790033],"about_ca_topic_score_codex":0.000939315,"about_ca_topic_score_gemma":0.00019858913,"teacher_disagreement_score":0.6569306,"about_ca_system_score_codex":0.00022003264,"about_ca_system_score_gemma":0.000094761985,"threshold_uncertainty_score":0.99993545},"labels":[],"label_agreement":null},{"id":"W1967374088","doi":"10.1023/b:scie.0000013298.22207.2b","title":"Modelling the characteristics of Web page outlinks","year":2004,"lang":"en","type":"article","venue":"Scientometrics","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":9,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Web page; Negative binomial distribution; Poisson distribution; Computer science; Mandelbrot set; Distribution (mathematics); Frequency distribution; Gaussian; Mathematics; Information retrieval; World Wide Web; Statistics; Physics","score_opus":0.03194412944867789,"score_gpt":0.28198450038848505,"score_spread":0.25004037093980713,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1967374088","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.64549387,0.00005701884,0.34925666,0.000090270936,0.00011175625,0.00010756705,0.000025949206,0.00003697641,0.004819912],"genre_scores_gemma":[0.9952945,0.000010137116,0.004372962,0.000020431045,0.00014071612,0.000004917564,0.000012359941,0.000008836175,0.0001351235],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99892247,0.000018226048,0.00026086648,0.0001674891,0.000419,0.0002119487],"domain_scores_gemma":[0.99915165,0.00007838372,0.00016961842,0.0003677337,0.00017886884,0.000053777367],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006182035,0.00009136832,0.0001674985,0.00064944295,0.00011743217,0.00006549761,0.0003996999,0.000022763097,0.00007490122],"category_scores_gemma":[0.000020421176,0.000065663924,0.00012369787,0.0055641737,0.00009097743,0.000061705425,0.000113608956,0.00014692346,0.000015675732],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000020017142,0.0012562867,0.2051464,0.000049147297,0.00035738727,0.000004179621,0.0026711666,0.23529044,0.004493071,0.42758366,0.0019833394,0.12114492],"study_design_scores_gemma":[0.0010700291,0.00014507285,0.0060646916,0.00008831337,0.00025376532,0.000001095546,0.00047134524,0.8811399,0.007627387,0.08101626,0.02137701,0.0007451544],"about_ca_topic_score_codex":0.000118185,"about_ca_topic_score_gemma":8.0793666e-7,"teacher_disagreement_score":0.64584947,"about_ca_system_score_codex":0.0000309482,"about_ca_system_score_gemma":0.000059241745,"threshold_uncertainty_score":0.26776972},"labels":[],"label_agreement":null},{"id":"W1975616173","doi":"10.1007/s11192-010-0317-2","title":"Intellectual structure of stem cell research: a comprehensive author co-citation analysis of a highly collaborative and multidisciplinary field","year":2010,"lang":"en","type":"article","venue":"Scientometrics","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":37,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Genome Canada","keywords":"Identification (biology); Citation; Multidisciplinary approach; Field (mathematics); Scopus; Citation analysis; Stem cell; Medical research; Data science; Sociology; Engineering ethics; Computer science; Library science; Social science; MEDLINE; Political science; Biology; Biotechnology","score_opus":0.2730697765223793,"score_gpt":0.4911885556865035,"score_spread":0.2181187791641242,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1975616173","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99325454,0.00010858487,0.004354933,0.00016157722,0.00059232843,0.0002634631,0.00034744618,0.000012219527,0.00090490934],"genre_scores_gemma":[0.99378437,0.000005661771,0.005396063,0.000014263083,0.000014947567,0.0000014390781,0.000025039208,0.000004010235,0.0007542183],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9942007,0.00039112387,0.0007072118,0.0007811189,0.003635369,0.00028447286],"domain_scores_gemma":[0.98154736,0.01308195,0.0004397243,0.0010054731,0.0037518826,0.00017359343],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.008066761,0.00011692341,0.00040421975,0.017125035,0.00028771182,0.00035126452,0.0010259302,0.00009218339,0.00021126283],"category_scores_gemma":[0.008546376,0.00008727885,0.000091491704,0.08827024,0.0006098738,0.00023493824,0.0008251443,0.00029690732,0.000013397403],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004563472,0.00094072043,0.18189694,0.0001768078,0.0006688563,0.000016996655,0.13548715,0.0067337085,0.28442174,0.00631011,0.115941845,0.26694876],"study_design_scores_gemma":[0.0015939209,0.0015472247,0.42480582,0.00003686996,0.0004489695,0.0000026402129,0.11214725,0.31271726,0.1224713,0.0052608363,0.018322062,0.00064584654],"about_ca_topic_score_codex":0.000045865825,"about_ca_topic_score_gemma":0.00012848558,"teacher_disagreement_score":0.30598354,"about_ca_system_score_codex":0.00002699884,"about_ca_system_score_gemma":0.00013174745,"threshold_uncertainty_score":0.99980503},"labels":[],"label_agreement":null},{"id":"W1979812464","doi":"10.1007/s11192-010-0249-x","title":"Mathematics 1868–2008: a bibliometric analysis","year":2010,"lang":"en","type":"article","venue":"Scientometrics","topic":"Complex Systems and Time Series Analysis","field":"Economics, Econometrics and Finance","cited_by":43,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Productivity; Distribution (mathematics); Range (aeronautics); Mathematics; Bibliometrics; Quarter (Canadian coin); Statistics; Mathematical economics; Econometrics; History; Library science; Economics; Computer science; Economic growth","score_opus":0.040057207972988534,"score_gpt":0.25840684375512885,"score_spread":0.2183496357821403,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1979812464","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9140246,0.0018326128,0.027156983,0.00018858047,0.0013135353,0.00021397544,0.00016953796,0.00013198027,0.054968186],"genre_scores_gemma":[0.9838277,0.000079920406,0.010160224,0.00007116372,0.0001453048,0.000015267347,0.000021576303,0.000027737155,0.0056510884],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9972323,0.000010455689,0.0010696267,0.0007236893,0.00034513907,0.0006187649],"domain_scores_gemma":[0.99739295,0.00019287682,0.00065319816,0.0011489186,0.00026616923,0.00034587234],"candidate_categories":["metaepi_narrow","bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.002787555,0.00023873821,0.00084277894,0.35937294,0.00027297597,0.000619211,0.0008242433,0.00015718208,0.0049384646],"category_scores_gemma":[0.0014642386,0.00024934788,0.0006385334,0.7178043,0.0001283555,0.0003612175,0.0002095707,0.0003133886,0.0020299936],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000022488691,0.0004435472,0.81135297,0.00004496218,0.0010807603,0.000013981279,0.0005758913,0.00019851445,0.00013491839,0.17545484,0.0026541925,0.008043173],"study_design_scores_gemma":[0.0007733971,0.00011396879,0.5448301,0.000006138839,0.00043726133,0.000029346105,0.0004539177,0.059397835,0.00008568605,0.012847179,0.379807,0.00121818],"about_ca_topic_score_codex":0.0012071561,"about_ca_topic_score_gemma":0.00030495,"teacher_disagreement_score":0.3771528,"about_ca_system_score_codex":0.00007385346,"about_ca_system_score_gemma":0.000029264069,"threshold_uncertainty_score":0.9999959},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W1982657965","doi":"10.1007/s11192-007-1731-y","title":"Direct interactions medical school faculty members have with professionals and managers working in public and private sector organizations: A cross-sectional study","year":2007,"lang":"en","type":"article","venue":"Scientometrics","topic":"Primary Care and Health Outcomes","field":"Health Professions","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Santé Montérégie; Université Laval","funders":"","keywords":"Multivariate analysis; Work (physics); Cross-sectional study; Psychology; Private sector; Medical school; Probit model; Public sector; Multivariate statistics; Medical education; Public relations; Medicine; Political science; Computer science; Engineering","score_opus":0.12241647269355602,"score_gpt":0.4864101305550571,"score_spread":0.36399365786150106,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1982657965","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9884573,0.00017728,0.00033336275,0.001993762,0.0010330784,0.0012811532,0.0000146216535,0.0001047867,0.0066046207],"genre_scores_gemma":[0.9928729,0.000055509194,0.0007574721,0.003759909,0.00014290639,0.000074791635,0.00003504232,0.00002566857,0.0022758283],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9962897,0.00028389614,0.0007430769,0.0005524918,0.0014008089,0.0007300402],"domain_scores_gemma":[0.9964879,0.0016651885,0.0002612573,0.00022612233,0.0005344892,0.0008250256],"candidate_categories":["sts","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0060429373,0.0001843008,0.00029935097,0.0035448177,0.0013246218,0.00013510518,0.00023285784,0.00018599414,0.0009828042],"category_scores_gemma":[0.0063253595,0.00014095615,0.00002036985,0.01004637,0.0001626981,0.0005193994,0.0003899424,0.0009695842,0.000024014187],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004406972,0.00019929363,0.9957612,0.00006335214,0.000029083629,0.00002764873,0.0018222482,0.0000014297141,0.0000037906889,0.00016984645,0.0009949174,0.00088309497],"study_design_scores_gemma":[0.0014803747,0.00006886303,0.9808996,0.00008200567,0.0000103955235,0.000010278465,0.0041164085,0.000058464702,0.0000026668847,0.000024539951,0.013090245,0.0001561921],"about_ca_topic_score_codex":0.0003053958,"about_ca_topic_score_gemma":0.005422342,"teacher_disagreement_score":0.014861659,"about_ca_system_score_codex":0.00094923715,"about_ca_system_score_gemma":0.0015143944,"threshold_uncertainty_score":0.9999755},"labels":[],"label_agreement":null},{"id":"W1987570886","doi":"10.1007/s11192-010-0162-3","title":"The discipline dependence of citation statistics","year":2010,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":60,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"National Academy of Sciences; National Science Council","keywords":"Citation; Field (mathematics); Citation analysis; Value (mathematics); Index (typography); Citation index; Scientometrics; Computer science; Data science; Statistics; Library science; Mathematics; World Wide Web","score_opus":0.5142649586072647,"score_gpt":0.6001517069820625,"score_spread":0.08588674837479782,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1987570886","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8341965,0.00050621736,0.14780019,0.0011288591,0.0054419166,0.0003842601,0.00030748936,0.00003372031,0.010200854],"genre_scores_gemma":[0.9823861,0.00013040716,0.014079803,0.000025296124,0.00007393467,0.000005067968,0.0000058215337,0.000009736238,0.0032838332],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.97771823,0.00015070154,0.0010190462,0.0006342519,0.0197705,0.00070729],"domain_scores_gemma":[0.96844697,0.018998379,0.00063867145,0.001449353,0.009935386,0.0005312515],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.055365648,0.00013448713,0.00023314428,0.07803998,0.00079306547,0.002939895,0.0046700407,0.000109475885,0.0003254991],"category_scores_gemma":[0.27158225,0.00007410849,0.000109825225,0.47606227,0.0008723497,0.0005462879,0.000974177,0.00049582694,0.00029599795],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000020514288,0.00017640238,0.13462837,0.000005681876,0.000010817009,0.000006222821,0.00021416764,0.00006139318,0.011551148,0.064567596,0.02087903,0.76787865],"study_design_scores_gemma":[0.0005558886,0.00025569618,0.78751874,0.0000034288405,0.000010905592,0.000016929742,0.0006101981,0.03280526,0.0072140093,0.09064214,0.080055565,0.00031121864],"about_ca_topic_score_codex":0.00003567485,"about_ca_topic_score_gemma":0.00012241805,"teacher_disagreement_score":0.76756746,"about_ca_system_score_codex":0.000048531107,"about_ca_system_score_gemma":0.00029333602,"threshold_uncertainty_score":0.99809515},"labels":[],"label_agreement":null},{"id":"W1990436954","doi":"10.1007/s11192-014-1386-4","title":"Forty years of gender disparities in Russian science: a historical bibliometric analysis","year":2014,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":50,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"National Science Foundation","keywords":"Bibliometrics; Scientometrics; Political science; Sociology; Library science; Social science; Computer science","score_opus":0.4898121462066615,"score_gpt":0.5481804217127789,"score_spread":0.058368275506117395,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1990436954","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.961884,0.0013173057,0.020420766,0.00017838643,0.001292742,0.0002601983,0.000030842046,0.000036367997,0.014579358],"genre_scores_gemma":[0.99530274,0.0001948203,0.0032134058,0.000079445446,0.000058700727,0.000010883985,0.0000036281226,0.00001527516,0.0011210808],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.96010697,0.00036187234,0.0018934307,0.0017686631,0.034118418,0.0017506549],"domain_scores_gemma":[0.9827145,0.008040538,0.0008120106,0.0021791484,0.0048983623,0.0013554548],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.10636593,0.0002678087,0.0010580055,0.9673125,0.0003187365,0.0027875814,0.006775371,0.00021298365,0.00033314273],"category_scores_gemma":[0.19493154,0.0002168584,0.0005343935,0.994532,0.00115386,0.001603256,0.0017179346,0.0004808527,0.00012607941],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009892534,0.00029458714,0.8801878,0.000008443941,0.000025095718,0.0000052184264,0.00011798891,0.00035992338,0.0001566562,0.004329752,0.00094581215,0.11355882],"study_design_scores_gemma":[0.00047813993,0.00014897622,0.9695664,0.0000032480923,0.000038108225,0.0000020944447,0.00006832961,0.017478969,0.00022425748,0.005777599,0.0059449254,0.00026891922],"about_ca_topic_score_codex":0.0009924469,"about_ca_topic_score_gemma":0.0000649958,"teacher_disagreement_score":0.1132899,"about_ca_system_score_codex":0.0015348381,"about_ca_system_score_gemma":0.00067945936,"threshold_uncertainty_score":0.99859846},"labels":[],"label_agreement":null},{"id":"W1993018298","doi":"10.1007/s11192-008-2064-1","title":"The relationship between reviewers’ quality-scores and number of citations for papers published in the journal Physics in Medicine and Biology from 2003–2005","year":2009,"lang":"en","type":"article","venue":"Scientometrics","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":44,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University; Juravinski Cancer Centre","funders":"","keywords":"Impact factor; Citation; Quality (philosophy); Statistics; Mathematics; Medicine; Psychology; Computer science; Library science; Biology; Philosophy; Epistemology","score_opus":0.8825900474161539,"score_gpt":0.626060316233719,"score_spread":0.25652973118243483,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1993018298","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93279535,0.022403868,0.015575385,0.024529325,0.0005049249,0.0014203861,0.00006921892,0.0000021769342,0.0026993812],"genre_scores_gemma":[0.9948215,0.0008311295,0.0036315103,0.00034847416,0.000100015226,0.000008058087,0.000010497686,0.0000024268725,0.00024637132],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9853525,0.0069032977,0.0047371374,0.00040562724,0.0023499893,0.00025143492],"domain_scores_gemma":[0.92024773,0.07382497,0.0031335768,0.0012159444,0.0014527831,0.00012500474],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.21014048,0.00011563543,0.0011525834,0.0011737384,0.00024226717,0.0004643724,0.0009572013,0.000056438126,0.00013853508],"category_scores_gemma":[0.39309895,0.00004203422,0.0001512442,0.017710894,0.00025102805,0.0002647469,0.000036254823,0.00019467779,0.00000705657],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000200282,0.000012483998,0.9491445,0.0000072659977,0.000011109779,6.61503e-8,0.0015886085,0.000003219115,0.0000047095614,0.01467165,0.012749166,0.021805221],"study_design_scores_gemma":[0.0002570283,0.000027233018,0.86742884,0.000031803993,0.000057326895,0.000001112706,0.001480971,0.000121841455,2.6945068e-7,0.10883276,0.021715134,0.000045703106],"about_ca_topic_score_codex":0.00003503761,"about_ca_topic_score_gemma":0.000100824735,"teacher_disagreement_score":0.18295848,"about_ca_system_score_codex":0.000022892667,"about_ca_system_score_gemma":0.00005842571,"threshold_uncertainty_score":0.85094976},"labels":[],"label_agreement":null},{"id":"W1993126323","doi":"10.1556/scient.67.2006.3.7","title":"Indirect patent citations","year":2006,"lang":"en","type":"article","venue":"Scientometrics","topic":"Intellectual Property and Patents","field":"Business, Management and Accounting","cited_by":50,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Center for Interuniversity Research and Analysis on Organizations; University of Ottawa; Statistics Canada","funders":"","keywords":"Index (typography); Quality (philosophy); Citation; Measure (data warehouse); Variable (mathematics); Citation analysis; Computer science; Information retrieval; Mathematics; Data mining; Library science; Epistemology; World Wide Web; Philosophy","score_opus":0.2538094299036781,"score_gpt":0.25225322252204485,"score_spread":0.0015562073816332567,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1993126323","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.65749043,0.00017323902,0.0016621964,0.00035312562,0.0015614459,0.00020071633,0.0000039486076,0.00019504427,0.33835986],"genre_scores_gemma":[0.9935892,0.0000023164127,0.00018866209,0.00086963543,0.00051426445,0.000006574587,0.000037794587,0.000010852067,0.004780687],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99885416,0.0000049536993,0.00016593169,0.00019761099,0.0004992402,0.00027811292],"domain_scores_gemma":[0.99945766,0.000043348642,0.000077597215,0.00013030213,0.00028147266,0.000009628807],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0004973737,0.000090743066,0.00008620462,0.0027732493,0.0002672363,0.00052003236,0.00022334601,0.00003914983,0.0005256935],"category_scores_gemma":[0.0004841019,0.000072895076,0.000053705193,0.011634976,0.00004641608,0.0005555399,0.00009036941,0.00007825599,0.0030629076],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000051666084,0.0010064172,0.35918784,0.00021875136,0.000044850167,0.000039355607,0.00023108744,0.0016751662,0.0021916542,0.098125115,0.47813192,0.059096165],"study_design_scores_gemma":[0.00073713966,0.000026030088,0.2533346,0.000029691262,0.000048262602,0.0000025965946,0.00011084218,0.011467942,0.00078436657,0.011244012,0.72166157,0.000552959],"about_ca_topic_score_codex":0.0005490694,"about_ca_topic_score_gemma":0.000031676766,"teacher_disagreement_score":0.3360988,"about_ca_system_score_codex":0.00003257796,"about_ca_system_score_gemma":0.000012616178,"threshold_uncertainty_score":0.9977133},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W1998258485","doi":"10.1023/b:scie.0000045122.93018.2a","title":"Exploring website features for business information","year":2004,"lang":"en","type":"article","venue":"Scientometrics","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":64,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Webometrics; Ranking (information retrieval); Data collection; Business; China; Marketing; Computer science; Information retrieval; World Wide Web; Statistics; Political science; Mathematics","score_opus":0.12208833343293496,"score_gpt":0.283840460132473,"score_spread":0.16175212669953803,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1998258485","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.051034153,0.0000795457,0.94331294,0.0007295134,0.001854828,0.00030397618,0.000010443522,0.00021229935,0.0024622867],"genre_scores_gemma":[0.7692963,0.00012483481,0.22925895,0.0009416692,0.00013998071,0.00007281841,0.000024313993,0.0000082353945,0.00013291856],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99826765,0.0000067782903,0.00028546157,0.00023291228,0.00080510945,0.00040210492],"domain_scores_gemma":[0.99825656,0.00013709863,0.00011981049,0.0005024792,0.0008417224,0.0001423128],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0012214399,0.00011853445,0.00012502071,0.0047240243,0.00026663745,0.0008469886,0.0010316637,0.000054746433,0.0000019241288],"category_scores_gemma":[0.0028087657,0.000107077474,0.00007024513,0.0349849,0.00004297145,0.009716222,0.0002507407,0.00009948926,0.00011236625],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017618213,0.00016629229,0.001312844,0.00022192055,0.00001596967,0.0000017853162,0.0032340593,0.011645254,0.00012678694,0.43097955,0.0021619545,0.55011594],"study_design_scores_gemma":[0.0052979295,0.00068072,0.4864403,0.000108405424,0.000024411625,0.000033055425,0.0004914824,0.013927449,0.023187218,0.02732241,0.44095802,0.0015286143],"about_ca_topic_score_codex":0.00002878189,"about_ca_topic_score_gemma":0.0000036618164,"teacher_disagreement_score":0.71826214,"about_ca_system_score_codex":0.00020911405,"about_ca_system_score_gemma":0.00019818093,"threshold_uncertainty_score":0.985527},"labels":[],"label_agreement":null},{"id":"W1999584965","doi":"10.1007/s11192-014-1258-y","title":"On the quality evaluation of scientific entities in Poland supported by consistency-driven pairwise comparisons method","year":2014,"lang":"en","type":"article","venue":"Scientometrics","topic":"Multi-Criteria Decision Making","field":"Decision Sciences","cited_by":37,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Laurentian University","funders":"Dalhousie University; Akademia Górniczo-Hutnicza im. Stanislawa Staszica; Politechnika Wrocławska; Fayetteville State University; Rensselaer Polytechnic Institute","keywords":"Pairwise comparison; Ranking (information retrieval); Consistency (knowledge bases); Focus (optics); Computer science; Domain (mathematical analysis); Parametric statistics; Quality (philosophy); Event (particle physics); Information retrieval; Operations research; Management science; Data mining; Mathematics; Statistics; Artificial intelligence; Epistemology; Economics","score_opus":0.49126161277195085,"score_gpt":0.5458934524210635,"score_spread":0.05463183964911261,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1999584965","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95749176,0.00012342274,0.034427453,0.00052468263,0.0016090367,0.00054599316,0.00013416712,0.000023594186,0.0051199035],"genre_scores_gemma":[0.9937315,0.000001399746,0.005281521,0.0001506174,0.00001633525,0.000016219843,0.000015941285,0.000009464148,0.0007769921],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.97865915,0.0051880335,0.0017861543,0.0010412255,0.012817711,0.00050770584],"domain_scores_gemma":[0.96494836,0.02702789,0.001296949,0.0023101703,0.0041741105,0.00024252429],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.1432846,0.0001960841,0.00054223277,0.006494879,0.00043796634,0.0011395134,0.0017743977,0.00010333249,0.0012427135],"category_scores_gemma":[0.13953464,0.00012542662,0.00017351376,0.023135124,0.000710582,0.00034869972,0.00033962663,0.00023490204,0.00017739531],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013816619,0.0014017577,0.53961676,0.000030997217,0.000042153915,0.0000020205246,0.005575716,0.00575669,0.028588714,0.058735475,0.21237168,0.14773987],"study_design_scores_gemma":[0.0019602699,0.00012991039,0.40392196,0.00006194839,0.000044872235,0.0000023212715,0.002456371,0.52433676,0.003973226,0.04380644,0.018918071,0.00038786387],"about_ca_topic_score_codex":0.0001281927,"about_ca_topic_score_gemma":0.0002149756,"teacher_disagreement_score":0.51858,"about_ca_system_score_codex":0.00017371922,"about_ca_system_score_gemma":0.00031577362,"threshold_uncertainty_score":0.9998974},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2001213194","doi":"10.1007/s11192-010-0295-4","title":"Reasons for and developments in international scientific collaboration: does an Asia–Pacific research area exist from a bibliometric point of view?","year":2010,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":35,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Forschungszentrum Jülich; Bundesministerium für Bildung und Forschung","keywords":"China; Regional science; Strengths and weaknesses; Citation; Library science; Citation impact; Citation analysis; Quarter (Canadian coin); Bibliometrics; Scientometrics; Political science; Point (geometry); Geography; Computer science; Law; Psychology; Mathematics","score_opus":0.5650881049249804,"score_gpt":0.5968205699851685,"score_spread":0.03173246506018812,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2001213194","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9833241,0.0005808174,0.0014479689,0.0011715897,0.0051320046,0.0010336011,0.00081985816,0.000027412581,0.006462665],"genre_scores_gemma":[0.96881086,0.00022279513,0.028839646,0.000019878415,0.00011991809,0.00008564748,0.00009371714,0.000022370878,0.0017851877],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9738749,0.00045442168,0.0015780524,0.002153784,0.020705216,0.0012335865],"domain_scores_gemma":[0.959282,0.013308792,0.0006086323,0.0017252322,0.023922266,0.0011530975],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.10445518,0.00027000444,0.0005653087,0.75205237,0.0008316996,0.010614284,0.004762477,0.0002945154,0.00038835325],"category_scores_gemma":[0.19600609,0.00019275086,0.00011686228,0.90742224,0.0014585081,0.0025958137,0.0014709235,0.0007297994,0.000058443784],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000104961255,0.0009815342,0.77835584,0.000017766657,0.000024058038,0.000011200485,0.0013997176,0.0000057024963,0.022391524,0.0071590734,0.008865267,0.18068333],"study_design_scores_gemma":[0.0018494697,0.0003344311,0.82489455,0.000033791544,0.0000071375925,0.0000067617993,0.0050660954,0.00395097,0.0074052797,0.043177497,0.112805925,0.00046809696],"about_ca_topic_score_codex":0.00020445864,"about_ca_topic_score_gemma":0.0010398943,"teacher_disagreement_score":0.18021524,"about_ca_system_score_codex":0.00029817995,"about_ca_system_score_gemma":0.0012464792,"threshold_uncertainty_score":0.9904128},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2002166221","doi":"10.1007/s11192-014-1432-2","title":"The role of public funding in nanotechnology scientific production: Where Canada stands in comparison to the United States","year":2014,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":26,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Concordia University; Polytechnique Montréal","funders":"","keywords":"Government (linguistics); Position (finance); Quality (philosophy); Societal impact of nanotechnology; Political science; Order (exchange); Production (economics); Business; Economics; Nanotechnology; Physics; Finance","score_opus":0.3818642461742465,"score_gpt":0.49254159140057985,"score_spread":0.11067734522633332,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2002166221","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9788197,0.002534238,0.0014822212,0.013594344,0.0024526354,0.0004928039,0.00002901905,0.000016684047,0.00057832774],"genre_scores_gemma":[0.9985254,0.00012993813,0.00018028186,0.00006269413,0.000036306585,0.000022515074,0.00000717613,0.000009703729,0.0010260092],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9804521,0.0008089694,0.0014116334,0.0011484269,0.014596118,0.0015827271],"domain_scores_gemma":[0.9816681,0.009484132,0.0005485455,0.002112684,0.005655602,0.00053094456],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.10047732,0.00017970991,0.00038638143,0.17397282,0.00094775046,0.003646784,0.0058815535,0.00011147395,0.00004983498],"category_scores_gemma":[0.16267467,0.00009889924,0.00006184533,0.75051457,0.00074204174,0.00039214277,0.0013338703,0.0005529416,0.000033291948],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012552317,0.00012439626,0.72753334,0.000004821248,0.0000050101316,0.0000013615087,0.0005351916,0.0065490208,0.00025813532,0.0042599034,0.016334828,0.24438141],"study_design_scores_gemma":[0.00025157566,0.000116481635,0.086358584,0.000013552526,0.0000017191693,0.000002089018,0.0075999144,0.060755022,0.002283325,0.0035446684,0.838903,0.00017008197],"about_ca_topic_score_codex":0.07146944,"about_ca_topic_score_gemma":0.64787954,"teacher_disagreement_score":0.8225682,"about_ca_system_score_codex":0.0008641114,"about_ca_system_score_gemma":0.0011056543,"threshold_uncertainty_score":0.9994971},"labels":[],"label_agreement":null},{"id":"W2004098455","doi":"10.1007/s11192-013-1069-6","title":"Effect of collaboration network structure on knowledge creation and technological performance: the case of biotechnology in Canada","year":2013,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":84,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Concordia University","funders":"","keywords":"Network structure; Productivity; Quality (philosophy); Production (economics); Knowledge management; Field (mathematics); Network analysis; Biotechnology; Business; Computer science; Data science; Engineering; Biology; Economics","score_opus":0.11364746930551028,"score_gpt":0.46945445948011494,"score_spread":0.35580699017460465,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2004098455","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9970747,0.00081252377,0.00008144907,0.00050734694,0.00030354943,0.0005588941,0.000025069447,0.000007662398,0.0006287817],"genre_scores_gemma":[0.9995803,0.00018631513,0.00012709483,0.000014592612,0.000013313213,0.000012615712,0.000001521015,0.0000040527066,0.000060164373],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9953133,0.00035616598,0.0006361364,0.00045993508,0.0027985687,0.00043588757],"domain_scores_gemma":[0.9895739,0.0077330503,0.00034464564,0.00060070213,0.0016248319,0.00012286937],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.01540162,0.00013040802,0.00034576523,0.050631985,0.00017534694,0.000285279,0.0011286217,0.00019247769,0.00009330339],"category_scores_gemma":[0.034254946,0.00006740615,0.000030724907,0.43146506,0.00055149745,0.00024350305,0.00044861136,0.00030889813,0.000007501409],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000022861559,0.00003863651,0.2841277,0.00002394552,0.0000051092957,0.000011504054,0.00006510794,0.0009922805,0.0009183518,0.0016643254,0.0018649116,0.7102652],"study_design_scores_gemma":[0.001534187,0.0029500036,0.74033695,0.000049916896,0.000016602158,0.0001227433,0.0017516516,0.15079904,0.09668067,0.0025941497,0.0028054256,0.0003586424],"about_ca_topic_score_codex":0.053702876,"about_ca_topic_score_gemma":0.07347152,"teacher_disagreement_score":0.7099066,"about_ca_system_score_codex":0.00032988153,"about_ca_system_score_gemma":0.0005209459,"threshold_uncertainty_score":0.97387993},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2017043167","doi":"10.1007/s11192-009-0069-z","title":"Applying the h-index in exploring bibliometric properties of elite marketing scholars","year":2009,"lang":"en","type":"article","venue":"Scientometrics","topic":"Management and Marketing Education","field":"Business, Management and Accounting","cited_by":36,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Elite; Prestige; Index (typography); Metric (unit); Sociology; Bibliometrics; Library science; Marketing; Social science; Political science; Computer science; Business; Philosophy; Law","score_opus":0.089740442885697,"score_gpt":0.2748107175528747,"score_spread":0.18507027466717768,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2017043167","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9644911,0.0010463112,0.00013664857,0.00053705263,0.0007469446,0.0007487284,2.1126796e-7,0.00007856502,0.0322144],"genre_scores_gemma":[0.99844146,0.00019140731,0.0001460383,0.0005211003,0.00027937113,0.00012443725,0.0000021683652,0.000011816406,0.00028221676],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99783224,0.000037653925,0.00040260333,0.0003033496,0.0010184761,0.0004056692],"domain_scores_gemma":[0.9989232,0.00014573598,0.00029892666,0.00031993046,0.00029740375,0.000014801503],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.009333398,0.00014236358,0.0001654866,0.08039055,0.00021748125,0.00070736546,0.00059127045,0.00003669304,0.00003633398],"category_scores_gemma":[0.0058032214,0.00010600907,0.000062379135,0.24279997,0.000048829228,0.0019953228,0.00020633907,0.00018711983,0.000029478788],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005180074,0.00016839894,0.31747058,0.00033166705,0.0000083913155,0.0000015131877,0.00010828139,0.00019541575,0.0015806842,0.001387331,0.0011261087,0.6775698],"study_design_scores_gemma":[0.00033738677,0.000005506839,0.94795185,0.00023489776,0.000018586681,3.2205605e-7,0.0007611414,0.0042936383,0.00015601012,0.00028640658,0.04574656,0.0002077005],"about_ca_topic_score_codex":0.00011609996,"about_ca_topic_score_gemma":0.00000674531,"teacher_disagreement_score":0.67736214,"about_ca_system_score_codex":0.000059841397,"about_ca_system_score_gemma":0.000021209413,"threshold_uncertainty_score":0.93003243},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2017301591","doi":"10.1007/s11192-014-1274-y","title":"Extending citer analysis to journal impact evaluation","year":2014,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Informetrics; Scientometrics; Impact factor; Ranking (information retrieval); Citation; Measure (data warehouse); Library science; Bibliometrics; Psychology; Information retrieval; Computer science; Sociology; Political science; Data mining","score_opus":0.7601598960034236,"score_gpt":0.7056378674837388,"score_spread":0.054522028519684795,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2017301591","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.767471,0.00037595988,0.2202187,0.0005489162,0.0016291613,0.00028688824,0.000020132225,0.000026743823,0.009422479],"genre_scores_gemma":[0.9943737,0.000032312993,0.0034286114,0.00029727924,0.000285818,0.000008771955,0.0000058856936,0.00001376486,0.0015538674],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9526649,0.00096714275,0.0012019225,0.0011959639,0.04262906,0.0013410222],"domain_scores_gemma":[0.96919125,0.0067912093,0.00063861394,0.0016940971,0.019049495,0.0026353227],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.22129287,0.0002464309,0.00057242514,0.5435105,0.00078729034,0.012955809,0.004299265,0.0001318257,0.004085271],"category_scores_gemma":[0.3109278,0.00015952822,0.00074695185,0.85212326,0.00011818832,0.0013021724,0.0008898209,0.00042614908,0.001649343],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012722144,0.00010420107,0.3140946,8.1168747e-7,0.00012539465,0.0000033650667,0.00017929112,0.0137634305,0.000345959,0.00022168357,0.019117672,0.6520309],"study_design_scores_gemma":[0.0004173702,0.00030360947,0.76743525,0.0000027060487,0.00012616227,0.000014166719,0.00010484914,0.19967523,0.00015971845,0.004045449,0.027452677,0.00026282275],"about_ca_topic_score_codex":0.00004039085,"about_ca_topic_score_gemma":0.000011384209,"teacher_disagreement_score":0.651768,"about_ca_system_score_codex":0.0009269925,"about_ca_system_score_gemma":0.0003741275,"threshold_uncertainty_score":0.999128},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2020164772","doi":"10.1007/s11192-006-0047-7","title":"Key Labs and Open Labs in the Chinese scientific research system: Their role in the national and international scientific arena","year":2006,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":9,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"China; Science Citation Index; Citation; Key (lock); Library science; Quarter (Canadian coin); Chinese academy of sciences; Resource (disambiguation); Political science; Computer science; Data science; World Wide Web; Public relations; History; Law","score_opus":0.5945123837319141,"score_gpt":0.593806149860948,"score_spread":0.0007062338709660221,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2020164772","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9509983,0.0029052193,0.000052442043,0.0037106278,0.0012437458,0.0010476724,0.00016419026,0.000014192637,0.039863598],"genre_scores_gemma":[0.9951642,0.000044835433,0.00023932436,0.00012848998,0.00015072334,0.00008031988,0.000028633081,0.000011178877,0.0041522626],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9640816,0.0020847549,0.0012719936,0.0021770175,0.028954996,0.0014296323],"domain_scores_gemma":[0.97729456,0.0141298175,0.00030019623,0.0012071647,0.0067909774,0.0002772922],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.27039412,0.00024684524,0.00035170827,0.16124608,0.002049613,0.0528011,0.011581393,0.00013899885,0.00009072709],"category_scores_gemma":[0.050815355,0.00012271648,0.00007877634,0.5613865,0.002106619,0.0019925833,0.0038017046,0.0008250338,0.0001606389],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004910176,0.00092479656,0.85550433,0.000027699401,0.000011020989,0.000040698662,0.0072135213,0.00016969391,0.00113616,0.03542786,0.06537916,0.034115948],"study_design_scores_gemma":[0.00090382586,0.00007145682,0.8559312,0.000027064274,0.0000018033272,0.000050993287,0.008599245,0.018287303,0.000056334957,0.03324895,0.08260381,0.00021798928],"about_ca_topic_score_codex":0.000543033,"about_ca_topic_score_gemma":0.0012826539,"teacher_disagreement_score":0.40014046,"about_ca_system_score_codex":0.00043094708,"about_ca_system_score_gemma":0.0005685454,"threshold_uncertainty_score":0.9992496},"labels":[],"label_agreement":null},{"id":"W2027998681","doi":"10.1007/s11192-009-0424-0","title":"Citation structure of an emerging research area on the verge of application","year":2008,"lang":"en","type":"article","venue":"Scientometrics","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":57,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thomson Reuters (Canada)","funders":"","keywords":"Pace; Commercialization; Demise; Citation; Bibliometrics; Nanotechnology; Data science; Engineering ethics; Computer science; Business; Engineering; Political science; Library science; Materials science; Marketing; Physics","score_opus":0.12886906034911313,"score_gpt":0.39970770308237213,"score_spread":0.27083864273325897,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2027998681","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9967789,0.000018862527,0.0017932827,0.00013349361,0.00034199338,0.00022771225,0.000028610917,0.000024504381,0.00065264554],"genre_scores_gemma":[0.9954367,0.000006252606,0.0044192155,0.000024610476,0.000037939935,0.0000057154048,0.000005948312,0.0000075321964,0.00005603437],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.996573,0.0003584166,0.00030938577,0.00034155947,0.0021286015,0.00028906635],"domain_scores_gemma":[0.99781746,0.0005661933,0.00026614522,0.00060642813,0.0006740112,0.00006978986],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005074818,0.00008279118,0.00014009366,0.0016017546,0.00037851225,0.000048100235,0.0009330355,0.0000492422,0.00036429823],"category_scores_gemma":[0.0038668797,0.000056225384,0.000025511916,0.01001083,0.0005976851,0.00022295267,0.00012678934,0.0001466383,0.00003408414],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014374236,0.000044732256,0.0022522209,0.000017625509,6.6173857e-7,3.9654398e-7,0.0010376071,0.0051091295,0.9871975,0.0036082028,0.00012861774,0.00058891973],"study_design_scores_gemma":[0.00012186703,0.00022494423,0.053995945,0.000016563012,0.0000025025727,0.0000041900194,0.00021900516,0.016029587,0.92718744,0.001785341,0.00032161787,0.00009097543],"about_ca_topic_score_codex":0.000120928096,"about_ca_topic_score_gemma":0.0000065500835,"teacher_disagreement_score":0.06001005,"about_ca_system_score_codex":0.000053793905,"about_ca_system_score_gemma":0.000087771565,"threshold_uncertainty_score":0.4809872},"labels":[],"label_agreement":null},{"id":"W2035191842","doi":"10.1007/s11192-006-0133-x","title":"Comparing business competition positions based on Web co-link data: The global market vs. the Chinese market","year":2006,"lang":"en","type":"article","venue":"Scientometrics","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":47,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Multidimensional scaling; Competition (biology); Similarity (geometry); Business; Competitive advantage; Link (geometry); Web site; Market share; Industrial organization; Marketing; Computer science; The Internet; World Wide Web","score_opus":0.0310848158395885,"score_gpt":0.29805954544449237,"score_spread":0.26697472960490387,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2035191842","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.046374306,0.00027962885,0.60979944,0.017971046,0.00327514,0.0010129334,0.0007309603,0.000450414,0.32010615],"genre_scores_gemma":[0.9929336,0.000020364552,0.00432794,0.0019585697,0.00027573798,0.0000107321475,0.0001720753,0.000008337264,0.000292648],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9961457,0.00024143024,0.00047414866,0.0006806065,0.0018937739,0.00056429586],"domain_scores_gemma":[0.9950925,0.0013138939,0.000226017,0.002721716,0.00051906187,0.00012681822],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.005143519,0.0002542378,0.00023872305,0.0014459273,0.0011245776,0.0016868894,0.004861056,0.00008810701,0.00008963929],"category_scores_gemma":[0.0016775143,0.00014924518,0.000095692165,0.04925736,0.00035075063,0.0012321789,0.00092419464,0.00031064363,0.00010032365],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001482008,0.0014403517,0.54388815,0.00013333971,0.000042919037,0.000024749092,0.00009040601,0.019373458,0.000020940042,0.10510345,0.30744654,0.022287482],"study_design_scores_gemma":[0.00031461322,0.000034001223,0.48828745,0.000013513046,0.0000073773163,0.000007970184,0.000006744972,0.48476943,0.0000040765385,0.00068257435,0.025740372,0.00013189451],"about_ca_topic_score_codex":0.00020779893,"about_ca_topic_score_gemma":0.00012661012,"teacher_disagreement_score":0.9465593,"about_ca_system_score_codex":0.00031401287,"about_ca_system_score_gemma":0.0003097867,"threshold_uncertainty_score":0.9993495},"labels":[],"label_agreement":null},{"id":"W2035792097","doi":"10.1007/s11192-009-0008-z","title":"Discovery of factors influencing patent value based on machine learning in patents in the field of nanotechnology","year":2009,"lang":"en","type":"article","venue":"Scientometrics","topic":"Intellectual Property and Patents","field":"Business, Management and Accounting","cited_by":56,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Selection (genetic algorithm); Field (mathematics); Patent analysis; Value (mathematics); Data science; Computer science; Feature selection; Set (abstract data type); Artificial intelligence; Machine learning; Knowledge management; Mathematics","score_opus":0.10528866240336661,"score_gpt":0.2640623858484547,"score_spread":0.1587737234450881,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2035792097","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99745345,0.000046689674,0.00023593278,0.0003198988,0.00014991559,0.00016306975,0.000001357925,0.0000147628825,0.0016149508],"genre_scores_gemma":[0.99858254,0.000008567377,0.000020120571,0.0013296319,0.0000148757845,0.0000015856701,0.000007906641,0.0000049857063,0.0000297777],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9986776,0.00003309224,0.00033360417,0.00018806555,0.0005225496,0.00024508388],"domain_scores_gemma":[0.999263,0.0002824431,0.00019308133,0.0001777993,0.00007914849,0.0000045485767],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001071215,0.00011252085,0.00018833719,0.0038320841,0.000054851404,0.000059943148,0.00042185126,0.00008734403,0.00001827388],"category_scores_gemma":[0.0029255494,0.00007092739,0.00005821371,0.008873057,0.000039821483,0.00033731325,0.000066210385,0.00029470286,0.000007120503],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014442038,0.00053225545,0.9641641,0.000116134506,0.000004948662,0.000010932257,0.00060695346,0.018165976,0.003193869,0.0014256159,0.00006249646,0.011572333],"study_design_scores_gemma":[0.0019738842,0.0007359416,0.7907759,0.0004892279,0.000025846693,4.820376e-7,0.0010640679,0.1794335,0.022736877,0.0014885239,0.0008541521,0.0004215843],"about_ca_topic_score_codex":0.0014576378,"about_ca_topic_score_gemma":0.000044968463,"teacher_disagreement_score":0.17338815,"about_ca_system_score_codex":0.000033667005,"about_ca_system_score_gemma":0.000016948397,"threshold_uncertainty_score":0.42632094},"labels":[],"label_agreement":null},{"id":"W2038196424","doi":"10.1007/s11192-013-1221-3","title":"Coverage and adoption of altmetrics sources in the bibliometric community","year":2014,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":266,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Altmetrics; Bookmarking; Social media; Citation; Tracking (education); Scopus; World Wide Web; Download; Computer science; Internet privacy; Psychology; Political science; MEDLINE","score_opus":0.5482835610101827,"score_gpt":0.547941311386074,"score_spread":0.0003422496241086703,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2038196424","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9725812,0.0019702993,0.013103785,0.00034024505,0.0004496927,0.00035365165,0.000032408647,0.000019516745,0.011149187],"genre_scores_gemma":[0.9976739,0.0009003761,0.00083874917,0.00027388628,0.00004887113,0.000009419979,0.000006606601,0.000011991764,0.00023618668],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9744112,0.0026961304,0.0014232779,0.00076830934,0.019740634,0.00096047786],"domain_scores_gemma":[0.9366072,0.054657523,0.00093499303,0.002370915,0.0048499224,0.0005794386],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.17132866,0.00023811031,0.00055871904,0.78389746,0.0006015102,0.0029663448,0.0052209995,0.0001921364,0.000063904175],"category_scores_gemma":[0.24673264,0.00014596799,0.0001639573,0.95854455,0.00067227427,0.0009260565,0.001274417,0.0009095567,0.00008467943],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000122395495,0.00041877685,0.6253552,0.000022350547,0.0000053483654,0.0000020033563,0.00071411935,0.00020221413,0.00011460862,0.003432768,0.0015933355,0.36812705],"study_design_scores_gemma":[0.0007053017,0.000458395,0.9631696,0.0000090040185,0.0000074582513,0.000011186824,0.001316197,0.0063808975,0.00017975848,0.008942391,0.018627973,0.00019187413],"about_ca_topic_score_codex":0.00040684664,"about_ca_topic_score_gemma":0.000046126213,"teacher_disagreement_score":0.36793518,"about_ca_system_score_codex":0.00011344444,"about_ca_system_score_gemma":0.0001189067,"threshold_uncertainty_score":0.9980687},"labels":[],"label_agreement":null},{"id":"W2044326206","doi":"10.1007/s11192-008-0220-2","title":"A new look at evidence of scholarly citation in citation indexes and from web sources","year":2007,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":85,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Citation; Web of science; Library science; Citation analysis; Webometrics; Citation impact; Bibliometrics; Scholarly communication; Computer science; Sample (material); Impact factor; World Wide Web; Political science; MEDLINE; Publishing","score_opus":0.5726813817650799,"score_gpt":0.5567807477511177,"score_spread":0.015900634013962134,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2044326206","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9728765,0.00912236,0.01575318,0.0004166511,0.00052818324,0.00027749315,0.000022548435,0.000018153753,0.0009849267],"genre_scores_gemma":[0.9885912,0.0006787465,0.009100274,0.000081218095,0.000061686624,0.0000022913212,0.000003677136,0.000009933621,0.001471007],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9797834,0.00024062127,0.0012943669,0.0010849801,0.016842943,0.0007536688],"domain_scores_gemma":[0.97244275,0.020947376,0.00067869056,0.00073543336,0.004372299,0.00082347664],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.061583053,0.0001795445,0.0003808583,0.24832967,0.00020909487,0.0036119826,0.0018837327,0.00020708622,0.0002956545],"category_scores_gemma":[0.19217694,0.00014401932,0.000095451316,0.5391781,0.00028726715,0.0039991075,0.0009399208,0.00033359253,0.00013273461],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005946088,0.00004929365,0.82608587,0.000006706121,0.000004549861,0.000005882124,0.0008868973,0.00003081033,0.012182006,0.0002786964,0.0011107539,0.15929906],"study_design_scores_gemma":[0.0007069465,0.00019801552,0.9824387,0.0000535971,0.0000056810386,0.0000019660008,0.00080319255,0.0018490063,0.005668396,0.0061341,0.0019612918,0.00017911794],"about_ca_topic_score_codex":0.0008616144,"about_ca_topic_score_gemma":0.00068535877,"teacher_disagreement_score":0.2908484,"about_ca_system_score_codex":0.0003777984,"about_ca_system_score_gemma":0.00043073774,"threshold_uncertainty_score":0.99742234},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2046437697","doi":"10.1007/s11192-007-0205-6","title":"“The Triple-Helix collaboration: Why do researchers collaborate with industry and the government? What are the factors that influence the perceived barriers?”","year":2007,"lang":"en","type":"article","venue":"Scientometrics","topic":"Innovation and Knowledge Management","field":"Business, Management and Accounting","cited_by":108,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Foundation for Healthcare Improvement; Université Laval","funders":"","keywords":"Government (linguistics); Affect (linguistics); Set (abstract data type); Public relations; Knowledge production; Business; Marketing; Knowledge management; Political science; Sociology; Computer science","score_opus":0.03674851862201558,"score_gpt":0.28574997305715,"score_spread":0.2490014544351344,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2046437697","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9240853,0.0012343915,0.00008831487,0.039015435,0.0014337773,0.0017831597,0.000009172094,0.000064310734,0.032286104],"genre_scores_gemma":[0.9907241,0.0002008628,0.000009036633,0.0057114046,0.00018929626,0.000039253988,0.0000033375593,0.000018883222,0.0031038271],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.996395,0.00009996162,0.00031288544,0.00035589916,0.0022854146,0.00055085286],"domain_scores_gemma":[0.9965748,0.00081515574,0.0004533456,0.0006358324,0.0014717715,0.000049103088],"candidate_categories":["bibliometrics","sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.009280633,0.00022366647,0.00015918456,0.00053260144,0.003132919,0.008273514,0.0010600047,0.000104531835,0.00005066221],"category_scores_gemma":[0.0034713903,0.00008841281,0.000037347738,0.026565678,0.0013765986,0.0015912958,0.0004749602,0.00056297187,0.000028615043],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00092359283,0.00008678954,0.71322995,0.00012077044,0.00025026395,0.000017637076,0.007175857,0.00012485748,0.000052912554,0.22141394,0.04518018,0.011423273],"study_design_scores_gemma":[0.0014482846,0.00001838621,0.20769559,0.00004862967,0.000052614083,7.289087e-7,0.327364,0.0005007259,0.00003773671,0.00042976448,0.46218938,0.00021417004],"about_ca_topic_score_codex":0.0000661365,"about_ca_topic_score_gemma":0.0009446733,"teacher_disagreement_score":0.50553435,"about_ca_system_score_codex":0.00020648207,"about_ca_system_score_gemma":0.00010302467,"threshold_uncertainty_score":0.9981649},"labels":[],"label_agreement":null},{"id":"W2049647990","doi":"10.1007/s11192-006-0099-8","title":"Visualization of citation patterns of some Canadian journals","year":2006,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Western University","funders":"","keywords":"Citation; Subject (documents); Visualization; Computer science; Information visualization; Citation analysis; Data science; Information retrieval; Informetrics; Information science; Order (exchange); Library science; Bibliometrics; Data mining","score_opus":0.5363669360498906,"score_gpt":0.5783523202939667,"score_spread":0.04198538424407605,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2049647990","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9727825,0.0013094939,0.019304305,0.0003075955,0.0009237425,0.00022610811,0.00019905563,0.00001095551,0.0049362704],"genre_scores_gemma":[0.99770874,0.00013441147,0.00056787365,0.000045736586,0.00008047979,0.0000023007317,0.000016148239,0.000010093128,0.001434199],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9818531,0.00020489024,0.0013963715,0.0005002524,0.015397932,0.000647441],"domain_scores_gemma":[0.98569995,0.0022620875,0.00088554656,0.0006113002,0.009994958,0.00054616225],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.031663522,0.00012957526,0.00036986903,0.34124595,0.00018063834,0.0009568856,0.0019329994,0.00013293732,0.00062331185],"category_scores_gemma":[0.047036353,0.000101962505,0.00017372576,0.5154192,0.00015546186,0.0010328296,0.00020736168,0.00012682796,0.00006591029],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004307996,0.00014024497,0.9479783,0.000013519589,0.0000065258755,0.0000032999087,0.00009223126,0.0004922885,0.0031071412,0.011792397,0.0054124272,0.030957295],"study_design_scores_gemma":[0.00034745983,0.00012371971,0.9666323,0.000016475811,0.000006426451,0.0000025557063,0.00018115583,0.0034991582,0.010575983,0.013076852,0.0053973706,0.00014056156],"about_ca_topic_score_codex":0.030472688,"about_ca_topic_score_gemma":0.00477067,"teacher_disagreement_score":0.1741732,"about_ca_system_score_codex":0.00024677344,"about_ca_system_score_gemma":0.0006273125,"threshold_uncertainty_score":0.9971062},"labels":[],"label_agreement":null},{"id":"W2055778626","doi":"10.1007/s11192-009-0127-6","title":"Citer analysis as a measure of research impact: library and information science as a case study","year":2009,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":40,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Citation; Measure (data warehouse); Exploratory research; Information science; Citation analysis; Psychology; Field (mathematics); Computer science; Information retrieval; Library science; Sociology; Social science; Mathematics; Data mining","score_opus":0.5600534252037995,"score_gpt":0.6469165712479984,"score_spread":0.08686314604419887,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2055778626","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9764564,0.0005082302,0.0003535739,0.00045571275,0.00018457317,0.0007256606,0.000036863228,0.00003782187,0.021241168],"genre_scores_gemma":[0.9985774,0.00012096288,0.0005429459,0.00014613982,0.000029634992,0.0000071998725,0.0000028864551,0.0000060860125,0.00056672737],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9561572,0.00056408497,0.0012993098,0.0011726125,0.03947521,0.0013316008],"domain_scores_gemma":[0.97893137,0.0051108007,0.0005062384,0.0018343744,0.011947321,0.0016699069],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.10394198,0.00023225548,0.0005910348,0.5961124,0.0010874551,0.010722064,0.0039037208,0.00013685107,0.0005129838],"category_scores_gemma":[0.12828812,0.00015568691,0.00025607346,0.90832627,0.0013261617,0.01781663,0.0017556424,0.0005703786,0.00019395647],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008595954,0.0009263084,0.7521572,0.0000069711987,0.00008751387,0.00044349825,0.0047021722,0.00012802746,0.0004229229,0.0031077694,0.0010119859,0.23691969],"study_design_scores_gemma":[0.0011095102,0.0036274106,0.96178657,0.000010134071,0.000078255536,0.00045882454,0.015711246,0.0092134625,0.0010591149,0.0045524635,0.0020264208,0.00036657284],"about_ca_topic_score_codex":0.00042754295,"about_ca_topic_score_gemma":0.0000052949426,"teacher_disagreement_score":0.3122139,"about_ca_system_score_codex":0.00018141737,"about_ca_system_score_gemma":0.0015132255,"threshold_uncertainty_score":0.99592066},"labels":[],"label_agreement":null},{"id":"W2057264385","doi":"10.1007/s11192-013-1171-9","title":"An overview of forestry journals in the period 2006–2010 as basis for ascertaining research trends","year":2013,"lang":"en","type":"article","venue":"Scientometrics","topic":"Forest ecology and management","field":"Environmental Science","cited_by":41,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institute for Biological Sciences","funders":"","keywords":"Ordination; Ecophysiology; Forest management; Ecology; Vegetation (pathology); Plant ecology; Plant science; Forest ecology; Biology; Forestry; Geography; Botany; Ecosystem","score_opus":0.1892962841462507,"score_gpt":0.43934751497824204,"score_spread":0.2500512308319913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2057264385","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98084027,0.00037947696,0.00013038839,0.0022480655,0.00022353571,0.0004699254,0.000007304086,0.000008467544,0.015692558],"genre_scores_gemma":[0.99607277,0.00010560306,0.0015807314,0.00029182082,0.000024115017,0.00011346375,0.000002603821,0.00000699401,0.0018018959],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9977782,0.00018124041,0.0002562912,0.00027359737,0.000992652,0.00051804073],"domain_scores_gemma":[0.9988206,0.0004116063,0.000090868736,0.0005119308,0.000057274283,0.00010771343],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0071288855,0.00008328147,0.00013267278,0.0018761532,0.0002667429,0.00013145003,0.0011978161,0.00007546146,0.0048529864],"category_scores_gemma":[0.0018740577,0.000057613957,0.00006265669,0.011256471,0.00034206692,0.00041282346,0.00028090712,0.00021826703,0.00020815684],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000033035987,0.00094228366,0.7695544,0.00006412301,0.000022975444,0.000012254167,0.004849581,0.003571514,0.00064368447,0.009671887,0.099251196,0.11138309],"study_design_scores_gemma":[0.00026083284,0.00039354578,0.97617733,0.000012770969,0.0000064625,0.0000036238423,0.0012509822,0.00192033,0.00007296771,0.011268121,0.008549364,0.0000836925],"about_ca_topic_score_codex":0.00088221167,"about_ca_topic_score_gemma":0.00040920175,"teacher_disagreement_score":0.20662293,"about_ca_system_score_codex":0.00013471834,"about_ca_system_score_gemma":0.00002148494,"threshold_uncertainty_score":0.99605674},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2058472726","doi":"10.1007/s11192-007-1707-y","title":"Why are Websites co-linked? The case of Canadian universities","year":2007,"lang":"en","type":"article","venue":"Scientometrics","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":35,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Western University","funders":"","keywords":"Similarity (geometry); Context (archaeology); World Wide Web; Sample (material); Set (abstract data type); Co-occurrence; Web page; Computer science; Information retrieval; Geography; Chemistry","score_opus":0.04234576219966546,"score_gpt":0.288850666366687,"score_spread":0.24650490416702153,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2058472726","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7869862,0.00068021246,0.10875255,0.0023023593,0.001424595,0.00033730906,0.000087303415,0.000117494434,0.09931198],"genre_scores_gemma":[0.9930051,0.000027814118,0.004704607,0.0015312947,0.00003544358,3.7379195e-7,0.0000020205684,0.000003946421,0.000689387],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984185,0.00003175212,0.0002517288,0.00024819595,0.0005538357,0.0004959546],"domain_scores_gemma":[0.99766546,0.0006517378,0.00015543758,0.00069879944,0.00049686205,0.00033171108],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0033388846,0.000102859995,0.00013101706,0.013330006,0.00042087075,0.00022139038,0.0013958534,0.0000794125,0.00004305231],"category_scores_gemma":[0.0006811267,0.000078637895,0.00007690924,0.05604397,0.00026016575,0.0007331033,0.0001917669,0.00014553229,0.000041662275],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021071968,0.00034142198,0.10320693,0.0002015042,0.00010266006,0.0038050143,0.015541338,0.00071445684,0.00018498156,0.5209118,0.25738752,0.09758128],"study_design_scores_gemma":[0.00065411086,0.00030942142,0.09771025,0.000029289371,0.00002407533,0.0006521512,0.015409845,0.014519621,0.0045740693,0.0012904075,0.86421937,0.0006074183],"about_ca_topic_score_codex":0.050752215,"about_ca_topic_score_gemma":0.08462959,"teacher_disagreement_score":0.6068318,"about_ca_system_score_codex":0.00030049478,"about_ca_system_score_gemma":0.00043640437,"threshold_uncertainty_score":0.99785304},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W2058498906","doi":"10.1007/s11192-011-0369-y","title":"Sex differences in research funding, productivity and impact: an analysis of Québec university professors","year":2011,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":204,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université du Québec; Université du Québec à Montréal","funders":"McGill University","keywords":"Productivity; Scientometrics; Political science; Regional science; Sociology; Economics; Economic growth; Social science","score_opus":0.872966831153213,"score_gpt":0.6254438142378042,"score_spread":0.2475230169154089,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2058498906","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9955975,0.0003410447,0.00036256813,0.000091500915,0.00015208228,0.00030880512,0.00005928618,0.000015664875,0.0030715687],"genre_scores_gemma":[0.9969999,0.00009431289,0.0003836482,0.0000030843282,0.000013581109,0.0000019112956,0.0000023901964,0.000006587633,0.0024945633],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98272353,0.0013595909,0.0006469616,0.0013847505,0.012784578,0.0011005736],"domain_scores_gemma":[0.9875354,0.00497048,0.00034513656,0.0012781391,0.0049561406,0.00091468595],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.09664137,0.00017537712,0.00069418584,0.4405308,0.0003897616,0.0007019419,0.0031819332,0.0001542178,0.00048667061],"category_scores_gemma":[0.059531037,0.00012562706,0.00017343384,0.766765,0.0012252252,0.0014253677,0.0014101858,0.0005116217,0.00001835959],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000039398612,0.0005245767,0.96795774,0.000005277978,0.00005193958,0.000009488515,0.003533811,0.000011560521,0.0001515115,0.0005357205,0.00010656422,0.027072396],"study_design_scores_gemma":[0.0002270595,0.0003935306,0.9861859,0.0000036472331,0.000029198762,8.675518e-7,0.004411916,0.0073114973,0.00036163637,0.00073248474,0.0001932674,0.00014899974],"about_ca_topic_score_codex":0.046965227,"about_ca_topic_score_gemma":0.018731663,"teacher_disagreement_score":0.32623416,"about_ca_system_score_codex":0.0005574236,"about_ca_system_score_gemma":0.0008836918,"threshold_uncertainty_score":0.99917394},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W2059929112","doi":"10.1007/s11192-008-2139-z","title":"Increasing dominance of English in publications archived by PubMed","year":2009,"lang":"en","type":"article","venue":"Scientometrics","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":17,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Dominance (genetics); English language; Period (music); Civilization; History; Library science; Linguistics; Computer science; Biology; Art; Philosophy; Archaeology","score_opus":0.01977561581092457,"score_gpt":0.28266987312396274,"score_spread":0.26289425731303817,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2059929112","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9915062,0.001213479,0.0012014126,0.00023580913,0.00015444284,0.00009481195,0.000019175437,0.000014596496,0.005560065],"genre_scores_gemma":[0.9956307,0.00012355894,0.003697482,0.00010658679,0.000050984112,0.00000885506,0.000039501698,0.000002965405,0.00033936085],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99909604,0.000044483655,0.00017870324,0.00022540241,0.00020613131,0.00024922454],"domain_scores_gemma":[0.9994083,0.00005322282,0.000075243326,0.00021970502,0.00015781453,0.000085686384],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0008400497,0.00006354355,0.00010436613,0.000626801,0.000029138206,0.000021443971,0.00027214704,0.00008545365,0.0000024196502],"category_scores_gemma":[0.009245105,0.00005861234,0.00003325969,0.002439401,0.00016922975,0.0000037614186,0.00005219062,0.00006845847,6.8259794e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000092382055,0.0006322774,0.14751019,0.000019096453,0.000014838951,0.0000010674217,0.0006092825,0.00001113986,0.19905058,0.000491998,0.022220504,0.62934667],"study_design_scores_gemma":[0.0008281023,0.00023399523,0.7313621,0.000013737851,0.000005319364,0.0000023821178,0.00011241578,0.00003714583,0.054560017,0.00019344444,0.2124645,0.00018682292],"about_ca_topic_score_codex":0.000013603024,"about_ca_topic_score_gemma":0.000007116401,"teacher_disagreement_score":0.6291598,"about_ca_system_score_codex":0.000014344052,"about_ca_system_score_gemma":0.000056575,"threshold_uncertainty_score":0.99910045},"labels":[],"label_agreement":null},{"id":"W2064363624","doi":"10.1023/b:scie.0000034389.14825.bc","title":"Links to commercial websites as a source of business information","year":2004,"lang":"en","type":"article","venue":"Scientometrics","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":78,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Webometrics; Hyperlink; Business; Revenue; Scopus; China; Link analysis; Profit (economics); Competitive intelligence; Business information; Marketing; World Wide Web; Computer science; Web page; Accounting; Economics","score_opus":0.020912191600320575,"score_gpt":0.2720124649186993,"score_spread":0.2511002733183787,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2064363624","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.272632,0.000028210121,0.72036934,0.00097584014,0.0005581227,0.00018938782,0.0000035640564,0.00008975972,0.0051537915],"genre_scores_gemma":[0.94707215,0.000011297375,0.050679486,0.0020799097,0.000040667222,0.000004267336,0.0000056417525,0.000003893517,0.00010267766],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9978217,0.00001535905,0.00043765327,0.00020365827,0.0011892265,0.00033241653],"domain_scores_gemma":[0.99781644,0.00013602064,0.00017395786,0.00059047213,0.0010667907,0.00021629267],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0013089936,0.00011629859,0.00017869993,0.006098521,0.00014234397,0.000366619,0.0013665794,0.0001348917,0.000014054839],"category_scores_gemma":[0.004229892,0.00010860395,0.00006137189,0.056624424,0.00007244525,0.0025189733,0.0005601599,0.00018385769,0.00077663385],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000031819407,0.0004927347,0.020253075,0.00019098009,0.000019524814,0.000002511484,0.008781149,0.034311876,0.00030872825,0.10926352,0.003084969,0.8232591],"study_design_scores_gemma":[0.002745883,0.0012311086,0.6695571,0.000120600926,0.000019005975,0.000023554465,0.00031143863,0.010972631,0.021675995,0.006867254,0.28545657,0.0010188633],"about_ca_topic_score_codex":0.00025295504,"about_ca_topic_score_gemma":0.000009367783,"teacher_disagreement_score":0.82224023,"about_ca_system_score_codex":0.00014979986,"about_ca_system_score_gemma":0.00034120757,"threshold_uncertainty_score":0.9982316},"labels":[],"label_agreement":null},{"id":"W2064596000","doi":"10.1007/s11192-013-1118-1","title":"The effect of characteristics of title on citation rates of articles","year":2013,"lang":"en","type":"article","venue":"Scientometrics","topic":"Academic Writing and Publishing","field":"Arts and Humanities","cited_by":79,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University Health Centre","funders":"","keywords":"Citation; Acronym; Information retrieval; Computer science; Sentence; Scopus; Citation analysis; Statistics; Library science; Mathematics; Natural language processing; MEDLINE; Linguistics; Biology","score_opus":0.037650109452268925,"score_gpt":0.277116055331689,"score_spread":0.23946594587942005,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2064596000","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9626379,0.00008224913,0.000009050928,0.000059795882,0.00022138582,0.000041396772,0.0000090083095,0.000004099501,0.036935132],"genre_scores_gemma":[0.99829584,0.0000067456854,0.000010928336,0.000007659738,0.00004269054,0.0000010109183,0.0000015650029,0.0000018968102,0.0016316798],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99961925,0.000018975279,0.00010612243,0.000030466717,0.0001754234,0.000049780872],"domain_scores_gemma":[0.9990969,0.00061512913,0.00009574262,0.00005250202,0.00012884622,0.000010882882],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00054204185,0.000021433625,0.00005285872,0.0002309674,0.000040216284,0.00004258592,0.00007991374,0.000012172011,0.00023682516],"category_scores_gemma":[0.00132267,0.000012514351,0.000017299175,0.00031924926,0.00010249814,0.000059647107,0.000010428637,0.00003654135,0.000034771143],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024206487,0.00007092529,0.036342856,0.00035427717,0.000036144193,1.5495185e-7,0.013010898,0.000007677155,0.0033749372,0.5112465,0.12341837,0.31211305],"study_design_scores_gemma":[0.0021661457,0.004159965,0.33296725,0.00083866605,0.0001761235,0.0000014656796,0.01059985,0.008710251,0.23299587,0.042812813,0.36374512,0.0008264786],"about_ca_topic_score_codex":0.000027886159,"about_ca_topic_score_gemma":5.9971666e-7,"teacher_disagreement_score":0.46843368,"about_ca_system_score_codex":0.0000036462852,"about_ca_system_score_gemma":0.0000052151563,"threshold_uncertainty_score":0.25930694},"labels":[],"label_agreement":null},{"id":"W2065923934","doi":"10.1007/s11192-014-1489-y","title":"Prospective evaluation of the accessibility of Internet references in leading general medical journals","year":2014,"lang":"en","type":"article","venue":"Scientometrics","topic":"Health Literacy and Information Accessibility","field":"Health Professions","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University; Institute for Clinical Evaluative Sciences; St Mary's Hospital Centre; Women's College Hospital; University of Toronto","funders":"","keywords":"The Internet; Sample (material); Library science; Medicine; Computer science; World Wide Web","score_opus":0.2332333098912759,"score_gpt":0.5662370323276832,"score_spread":0.3330037224364073,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2065923934","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9793142,0.00009300073,0.0011215981,0.00023949832,0.00095785764,0.0006567484,0.000007857774,0.000007770272,0.017601486],"genre_scores_gemma":[0.9990654,0.000011678083,0.0003058064,0.00035207404,0.0000633989,0.000029039054,0.000002351836,0.0000027284505,0.0001675102],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99298316,0.0018457604,0.001953492,0.00021028248,0.0027089724,0.0002983458],"domain_scores_gemma":[0.9951205,0.0009210624,0.0013293562,0.00038935276,0.0021017145,0.00013796774],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.061824113,0.000074473835,0.00027336736,0.0011199183,0.00016233721,0.000016785361,0.0006420128,0.00015925328,0.0012068248],"category_scores_gemma":[0.047559045,0.000044021795,0.00005150836,0.007250172,0.0001677337,0.0006650752,0.00023390519,0.00059266575,0.000013564804],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012804698,0.00005728179,0.96787196,0.00018705014,0.000002079021,8.607219e-9,0.0036269904,0.000019922714,0.000019656625,0.0010773606,0.00020488976,0.026919967],"study_design_scores_gemma":[0.00063758594,0.00002812608,0.9483506,0.00031832917,0.0000049997525,1.478333e-7,0.0003424511,0.046791274,0.00040206866,0.0027068683,0.00037624416,0.000041316936],"about_ca_topic_score_codex":0.0002396154,"about_ca_topic_score_gemma":0.00022714637,"teacher_disagreement_score":0.04677135,"about_ca_system_score_codex":0.0003812645,"about_ca_system_score_gemma":0.0012156672,"threshold_uncertainty_score":0.9997062},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2067745687","doi":"10.1007/s11192-009-0126-7","title":"Patent activity on water pollution and treatment in China—a scientometric perspective","year":2009,"lang":"en","type":"article","venue":"Scientometrics","topic":"Innovation Policy and R&D","field":"Economics, Econometrics and Finance","cited_by":17,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thomson Reuters (Canada)","funders":"","keywords":"Perspective (graphical); China; Patent analysis; Pollution; Scientometrics; Regional science; Geography; Sociology; Computer science; Data science; Social science; Biology","score_opus":0.08727446852174127,"score_gpt":0.28987522875459804,"score_spread":0.20260076023285678,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2067745687","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96948457,0.00026369552,0.0002805282,0.0037542656,0.0003858335,0.00027022656,0.000050846385,0.00003120349,0.025478842],"genre_scores_gemma":[0.99786675,0.00013195837,0.00011173556,0.00046399332,0.00006130483,0.000008500844,0.0000051453344,0.0000075396974,0.0013430568],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99843884,0.00001928283,0.00032123574,0.00056830037,0.00015195996,0.0005003647],"domain_scores_gemma":[0.9994181,0.000025043437,0.00012790263,0.00024754537,0.00006787365,0.00011358375],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0012309797,0.00016907374,0.0002718906,0.01489421,0.00019397626,0.00015623684,0.00014523983,0.000093283175,0.00006912188],"category_scores_gemma":[0.0003525537,0.00013728168,0.00006271603,0.017206416,0.00008002314,0.00031802626,0.000034962562,0.00013311108,0.00022317881],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000100354686,0.0019117342,0.059663825,0.000008773936,0.000027130773,0.000012247239,0.005350922,0.00047045818,0.0009090779,0.8876349,0.0002709629,0.04363965],"study_design_scores_gemma":[0.0012473252,0.00072505826,0.96633446,0.000006143099,0.0000024407973,0.0000031897428,0.00007103573,0.0015258038,0.0028072745,0.023140002,0.0038719038,0.00026537527],"about_ca_topic_score_codex":0.0005623742,"about_ca_topic_score_gemma":0.000013683549,"teacher_disagreement_score":0.90667063,"about_ca_system_score_codex":0.0011669996,"about_ca_system_score_gemma":0.000021775211,"threshold_uncertainty_score":0.99627113},"labels":[],"label_agreement":null},{"id":"W2068893145","doi":"10.1007/s11192-012-0839-x","title":"Bibliometric analysis of biotechnology research in Pakistan","year":2012,"lang":"en","type":"article","venue":"Scientometrics","topic":"Genetics, Bioinformatics, and Biomedical Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":37,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"International Development Research Centre; University of Karachi","keywords":"Bibliometrics; Index (typography); Citation index; Biotechnology; Citation; Science Citation Index; Scientometrics; Annual growth %; Library science; Political science; Biology; Agricultural economics; Economics; Computer science","score_opus":0.08999657371647511,"score_gpt":0.44234703883899634,"score_spread":0.3523504651225212,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2068893145","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.990977,0.004399931,0.0013510254,0.00013122988,0.0002071647,0.00015698904,0.00003739236,0.0000071968097,0.0027320588],"genre_scores_gemma":[0.9961801,0.0021530052,0.0012387241,0.00003330144,0.000059604346,0.000007385898,0.00006673774,0.00000808622,0.00025304413],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9966908,0.000120383025,0.00046678897,0.00028153072,0.0015014526,0.0009390361],"domain_scores_gemma":[0.9984099,0.00015250155,0.0000936533,0.0005907402,0.00044860764,0.00030459434],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0073917517,0.00010718943,0.00025243548,0.34532988,0.00006441468,0.000038257418,0.00068525394,0.00032558822,0.00004459477],"category_scores_gemma":[0.0023906152,0.00009288188,0.00012254172,0.6301925,0.0006414753,0.0000095980695,0.0005309853,0.00024598994,0.000030073472],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004810447,0.0007750416,0.5903131,0.000102436104,0.00028357262,0.0000022050285,0.00023351153,0.000025773094,0.12515883,0.00062538486,0.0020047533,0.28042734],"study_design_scores_gemma":[0.0005767972,0.00059336826,0.6956849,0.000010693627,0.00006497434,0.0000024792098,0.0006285975,0.00086049706,0.239249,0.00006907715,0.06200115,0.0002584522],"about_ca_topic_score_codex":0.000080664795,"about_ca_topic_score_gemma":0.000041827396,"teacher_disagreement_score":0.28486264,"about_ca_system_score_codex":0.000064832944,"about_ca_system_score_gemma":0.00012311638,"threshold_uncertainty_score":0.66209024},"labels":[],"label_agreement":null},{"id":"W2070973569","doi":"10.1007/s11192-013-1203-5","title":"Trends in and contributions to entrepreneurship research: a broad review of literature from 1996 to June 2012","year":2013,"lang":"en","type":"review","venue":"Scientometrics","topic":"Entrepreneurship Studies and Influences","field":"Business, Management and Accounting","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Social Sciences Citation Index; Citation index; Entrepreneurship; Citation; Categorization; Field (mathematics); Sign (mathematics); Web of science; Library science; Social science; Science Citation Index; Sociology; Period (music); Bibliometrics; Political science; Computer science; MEDLINE","score_opus":0.14742338298890986,"score_gpt":0.424124825057443,"score_spread":0.2767014420685331,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2070973569","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00015195212,0.99367374,0.000002545188,0.0020385233,0.0007220138,0.0011880933,0.0003977286,0.000029346103,0.0017960753],"genre_scores_gemma":[0.000045841352,0.9964845,0.000082006205,0.0010098881,0.00052871386,0.00020533269,0.00026314252,0.000026247553,0.0013543188],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9959119,0.00013898365,0.0009528367,0.00096086453,0.0012572926,0.0007781399],"domain_scores_gemma":[0.99681306,0.0006401069,0.00041474763,0.00072307215,0.001284198,0.00012483339],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.003120363,0.00040925937,0.0016058042,0.013243594,0.00016563054,0.0006176383,0.00092247495,0.00019958885,0.00036874218],"category_scores_gemma":[0.007600287,0.00030981083,0.00027247015,0.060239337,0.00012962655,0.000731333,0.0012747435,0.0005426457,0.0004247503],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000029462808,0.000078575096,0.0013252399,0.018856673,0.000032941567,0.000006785349,0.00008329648,4.2238167e-7,6.339839e-7,0.0006294932,0.11000219,0.8689808],"study_design_scores_gemma":[0.00011184827,0.000017706128,0.0012074435,0.07057704,0.00012197188,8.0973956e-7,0.000022185804,8.88083e-7,3.8073952e-7,0.00012203921,0.9275386,0.00027910352],"about_ca_topic_score_codex":0.0008638824,"about_ca_topic_score_gemma":0.00018943525,"teacher_disagreement_score":0.8687017,"about_ca_system_score_codex":0.00013517654,"about_ca_system_score_gemma":0.000056330275,"threshold_uncertainty_score":0.9999354},"labels":[],"label_agreement":null},{"id":"W2075464897","doi":"10.1007/s11192-012-0859-6","title":"Characterizing a scientific elite (B): publication and citation patterns of the most highly cited scientists in environmental science and ecology","year":2012,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":62,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Citation; Publishing; Publication; Elite; Productivity; Citation analysis; Library science; Ecology; Political science; Sociology; Biology; Computer science; Law","score_opus":0.26447491485703406,"score_gpt":0.4612345109392629,"score_spread":0.19675959608222882,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2075464897","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9956776,0.0006994798,0.00013463551,0.0007187071,0.0016700912,0.00045266986,0.00010116226,0.000010109165,0.0005355787],"genre_scores_gemma":[0.99883133,0.00013830276,0.00034268596,0.00016298507,0.00003054688,0.000013478271,0.000008483442,0.000008122991,0.0004640881],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98372483,0.00026619187,0.0008648708,0.0010733446,0.012976815,0.0010939189],"domain_scores_gemma":[0.9935562,0.0018035665,0.00059426547,0.0009216093,0.0024443201,0.0006800312],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.068968765,0.00016955637,0.0002728926,0.14825515,0.0009451085,0.0040509864,0.0022124788,0.000104637824,0.000064110296],"category_scores_gemma":[0.06760544,0.00011683832,0.000048373186,0.45360926,0.0026510481,0.0032238746,0.0021370817,0.0002516458,0.000028697012],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000458263,0.00014713492,0.8878684,0.000007167814,0.00000163685,3.3263163e-7,0.00082012836,0.0000011938992,0.041489635,0.0006697598,0.000107689186,0.06888234],"study_design_scores_gemma":[0.0003440464,0.000037609912,0.9871478,0.000010713667,0.000003806649,0.000010180237,0.00036640657,0.0025412806,0.004473042,0.00019655743,0.004734856,0.00013367478],"about_ca_topic_score_codex":0.000055574124,"about_ca_topic_score_gemma":0.000022160319,"teacher_disagreement_score":0.3053541,"about_ca_system_score_codex":0.000433768,"about_ca_system_score_gemma":0.0003471444,"threshold_uncertainty_score":0.99698293},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2076071133","doi":"10.1007/s11192-012-0822-6","title":"Do citations and impact factors relate to the real numbers in publications? A case study of citation rates, impact, and effect sizes in ecology and evolutionary biology","year":2012,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University; York University","funders":"","keywords":"Citation; Harm; Impact factor; Proxy (statistics); Ecology; Value (mathematics); Metric (unit); Set (abstract data type); Biology; Psychology; Statistics; Computer science; Mathematics; Social psychology; Library science; Economics","score_opus":0.37632678244284634,"score_gpt":0.5872092963679597,"score_spread":0.21088251392511337,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2076071133","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99684346,0.0014611391,0.00005452258,0.00025869027,0.00021084548,0.0009433344,0.00006680362,0.000008086902,0.00015312916],"genre_scores_gemma":[0.99948907,0.00019607016,0.0002028513,0.000011366642,0.000013264397,0.00003906236,0.0000057372454,0.00000680475,0.00003577737],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9949861,0.001026733,0.0007189172,0.0006286569,0.0019599258,0.00067965046],"domain_scores_gemma":[0.97907573,0.018579938,0.00026527906,0.00044273803,0.0010852754,0.00055104506],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.02995713,0.00017192888,0.00037741574,0.12591848,0.0002878303,0.00069136836,0.0005148041,0.00012907131,0.000029511695],"category_scores_gemma":[0.070761286,0.00009904343,0.000048032885,0.32032874,0.00031438,0.00095150684,0.00055808085,0.00026406883,0.0000060229413],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019866524,0.00025101952,0.98634416,0.0000031367097,0.000012175252,0.0000052423293,0.004259727,0.000100751604,0.000070876595,0.00013920631,0.00023547235,0.008558375],"study_design_scores_gemma":[0.00077484705,0.00096299604,0.98997694,0.000002773579,0.000007714173,0.000059043035,0.0065358197,0.0009765993,0.000005217995,0.00053012045,0.00006179313,0.00010615292],"about_ca_topic_score_codex":0.004534582,"about_ca_topic_score_gemma":0.0012639653,"teacher_disagreement_score":0.19441028,"about_ca_system_score_codex":0.00028330187,"about_ca_system_score_gemma":0.00017113915,"threshold_uncertainty_score":0.9988633},"labels":[],"label_agreement":null},{"id":"W2080497488","doi":"10.1007/s11192-014-1457-6","title":"Exploring the interdisciplinary evolution of a discipline: the case of Biochemistry and Molecular Biology","year":2014,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":56,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"","keywords":"Discipline; Variety (cybernetics); Interdisciplinarity; Identification (biology); Epistemology; Sociology; Data science; Engineering ethics; Biology; Social science; Computer science; Ecology; Philosophy","score_opus":0.4283907410291913,"score_gpt":0.5475682690263628,"score_spread":0.11917752799717152,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2080497488","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9571957,0.0013209905,0.038435597,0.0010992491,0.00049222994,0.00017256188,0.000031432206,0.000007765928,0.0012444644],"genre_scores_gemma":[0.9991814,0.000054743516,0.00058013096,0.00001811711,0.00004251545,0.000012679625,0.0000010828247,0.00000667514,0.00010268293],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99478793,0.0003404771,0.0007558698,0.0005908344,0.003080929,0.0004439789],"domain_scores_gemma":[0.98852783,0.0073299715,0.00045512215,0.0012264843,0.0022433593,0.00021720388],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.032988917,0.00013683637,0.00029258375,0.037022043,0.00040971756,0.00039033729,0.0022997432,0.000067540364,0.00002067387],"category_scores_gemma":[0.050511062,0.00006417531,0.00014897766,0.2764794,0.0013678196,0.00029461615,0.0034563353,0.00024384457,0.000007794844],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000060062295,0.00039199943,0.098682374,0.000088588284,0.000069323054,0.000045996603,0.0023178442,0.00031777166,0.096142575,0.056403916,0.0017079121,0.7437716],"study_design_scores_gemma":[0.0026945805,0.0022412941,0.29979628,0.00011907288,0.00015079488,0.0011317048,0.031596515,0.15953434,0.2225431,0.27340248,0.0055663493,0.0012234894],"about_ca_topic_score_codex":0.000091476606,"about_ca_topic_score_gemma":0.000009920851,"teacher_disagreement_score":0.74254817,"about_ca_system_score_codex":0.00008286251,"about_ca_system_score_gemma":0.00010668882,"threshold_uncertainty_score":0.9957414},"labels":[],"label_agreement":null},{"id":"W2093594766","doi":"10.1007/s11192-012-0640-x","title":"Exploring Web keyword analysis as an alternative to link analysis: a multi-industry case","year":2012,"lang":"en","type":"article","venue":"Scientometrics","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Revenue; Link analysis; Computer science; Context (archaeology); Web page; Word lists by frequency; Information retrieval; Visibility; Index (typography); World Wide Web; Business; Accounting; Natural language processing; Geography","score_opus":0.25821070208708496,"score_gpt":0.3850153108853049,"score_spread":0.12680460879821992,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2093594766","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6789488,0.00009328902,0.31889832,0.00014378253,0.0007647004,0.00017406367,0.000018505616,0.00015443879,0.0008041021],"genre_scores_gemma":[0.9344845,0.000022363189,0.06436267,0.00050227245,0.00022177363,0.000031922293,0.000008325677,0.0000103684415,0.0003558508],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99545074,0.00016116945,0.00061854423,0.0009219017,0.0017277184,0.0011199553],"domain_scores_gemma":[0.9953256,0.0003407351,0.00024515876,0.001805656,0.00068612804,0.0015966938],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.004953866,0.00029728038,0.00052115304,0.03341801,0.00040397875,0.0009209956,0.0019935772,0.00019379462,0.00007141401],"category_scores_gemma":[0.0028197912,0.00027847447,0.0004297879,0.2517508,0.0000723115,0.0058917394,0.000920646,0.00057273393,0.00031555345],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000023963206,0.0020249197,0.64703894,0.00003771384,0.0041424106,0.0005968201,0.041065454,0.045101743,0.00011651555,0.012107141,0.00018450414,0.24755989],"study_design_scores_gemma":[0.00070884597,0.00042390896,0.168301,0.000009289274,0.0017399265,0.000104551014,0.0021698512,0.81627226,0.0021978933,0.00007723553,0.006860561,0.0011347121],"about_ca_topic_score_codex":0.00055399886,"about_ca_topic_score_gemma":0.00017219942,"teacher_disagreement_score":0.7711705,"about_ca_system_score_codex":0.00037671215,"about_ca_system_score_gemma":0.00015428186,"threshold_uncertainty_score":0.99996674},"labels":[],"label_agreement":null},{"id":"W2097269731","doi":"10.1023/b:scie.0000027677.79173.b8","title":"The classification of biomedical journals by research level","year":2004,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":89,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Science North","funders":"","keywords":"Computer science; Citation; Science Citation Index; Information retrieval; Basic research; Data science; Library science","score_opus":0.9272708597111224,"score_gpt":0.7093757539334359,"score_spread":0.21789510577768656,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2097269731","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8835268,0.01285592,0.06419225,0.020682685,0.0030844128,0.0008100205,0.00031111806,0.00004413114,0.0144926775],"genre_scores_gemma":[0.99374163,0.0012607563,0.0012231361,0.000047540336,0.000102174934,0.000012813608,0.0000053089684,0.000012666737,0.0035939931],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9392418,0.0007403781,0.0016692668,0.0009770918,0.05584848,0.0015229603],"domain_scores_gemma":[0.95609325,0.0199717,0.00067840895,0.0018405326,0.020194031,0.0012220936],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.20571366,0.00015457203,0.00032768692,0.1777989,0.0015367926,0.004603872,0.007405369,0.00021015077,0.00020755702],"category_scores_gemma":[0.30833465,0.000084796695,0.00020159756,0.7157363,0.002311401,0.0006741668,0.0011950918,0.000835082,0.0006112434],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003475098,0.00084697886,0.02419321,0.0000072707385,0.000025773312,0.0000072678627,0.00036233422,0.000033834884,0.053912126,0.019815782,0.26467213,0.63608855],"study_design_scores_gemma":[0.0013741108,0.0006347309,0.40077558,0.000026228985,0.00000609612,0.00001834893,0.0022291394,0.0013206109,0.01171471,0.08020875,0.50139576,0.00029593564],"about_ca_topic_score_codex":0.0001161045,"about_ca_topic_score_gemma":0.0000130394565,"teacher_disagreement_score":0.6357926,"about_ca_system_score_codex":0.0005187356,"about_ca_system_score_gemma":0.0010557031,"threshold_uncertainty_score":0.9997631},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2098507569","doi":"10.1007/s11192-015-1788-y","title":"Bibliometric analysis of output and impact based on CRIS data: a case study on the registered output of a Dutch university","year":2015,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":47,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Regional science; Sociology; Operations research; Computer science; Econometrics; Data science; Library science; Economics; Engineering","score_opus":0.8230981755742243,"score_gpt":0.5877561517828117,"score_spread":0.23534202379141256,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2098507569","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9924242,0.00031583683,0.0030082287,0.00033719672,0.00030467034,0.00065892114,0.0009107823,0.000018068544,0.0020220887],"genre_scores_gemma":[0.99881154,0.000041456187,0.00028269182,0.000060803442,0.000018434226,0.0000015857662,0.000012185662,0.0000123644495,0.0007589151],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9740435,0.001167949,0.0011139103,0.0015930135,0.021348821,0.0007327967],"domain_scores_gemma":[0.96223694,0.021159993,0.0012336256,0.0051588314,0.009058973,0.0011516481],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.08615682,0.0003086135,0.0010039281,0.9048277,0.0003301841,0.001443172,0.0058023953,0.00014108872,0.000073583986],"category_scores_gemma":[0.18275133,0.00017951052,0.00036465723,0.98116934,0.0006142933,0.0007273988,0.0023093624,0.0003870095,0.000026731663],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00020685753,0.0020560229,0.92641646,0.000008197001,0.00037756676,0.0004309792,0.0008712417,0.0020481262,0.0000112717635,0.00006946288,0.01462197,0.052881848],"study_design_scores_gemma":[0.0017785599,0.0028630614,0.82521796,0.000008988508,0.0003901591,0.000019443925,0.012684992,0.154682,0.000041024337,0.000060354625,0.0019845546,0.00026889323],"about_ca_topic_score_codex":0.0027090725,"about_ca_topic_score_gemma":0.00011737597,"teacher_disagreement_score":0.15263386,"about_ca_system_score_codex":0.0003396087,"about_ca_system_score_gemma":0.0007678303,"threshold_uncertainty_score":0.99959344},"labels":[],"label_agreement":null},{"id":"W2102705305","doi":"10.1007/s11192-008-0106-3","title":"The declining scientific impact of theses: Implications for electronic thesis and dissertation repositories and graduate studies","year":2007,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":50,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; McGill University","funders":"","keywords":"Library science; Altmetrics; Graduate students; Peer review; Sociology; Political science; Public relations; Computer science; Pedagogy; Law","score_opus":0.7238816379522245,"score_gpt":0.6807475618396931,"score_spread":0.043134076112531394,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2102705305","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9818325,0.009835767,0.0066321557,0.00051023206,0.0005240319,0.00045071793,0.0000269202,0.00001534268,0.00017232267],"genre_scores_gemma":[0.99712825,0.0013424626,0.0009149256,0.000008130403,0.00004002873,0.000020155185,0.0000023211767,0.00000938659,0.0005343276],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99322665,0.00011100606,0.0008981966,0.0007831079,0.0040883967,0.0008926188],"domain_scores_gemma":[0.94417477,0.044209804,0.0006803746,0.00088267814,0.00966383,0.0003885278],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.073450044,0.00015682058,0.0003083725,0.04414637,0.0020475897,0.003741698,0.0011878463,0.00006487826,0.0000015446439],"category_scores_gemma":[0.16636352,0.00008314037,0.00016062449,0.23673761,0.001217727,0.000598426,0.00040670228,0.00014999232,0.0000019878],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000069667665,0.00007401209,0.5932082,0.0000140862785,0.00008713714,3.5983314e-7,0.0023175923,0.000017344633,0.0036106675,0.016161615,0.0009971547,0.3834422],"study_design_scores_gemma":[0.00024675686,0.00037430256,0.941357,0.000007885856,0.000020874442,0.0000060073335,0.004757581,0.0009631797,0.0020442544,0.04886956,0.0012230116,0.00012957356],"about_ca_topic_score_codex":0.000043252636,"about_ca_topic_score_gemma":0.00008884861,"teacher_disagreement_score":0.3833126,"about_ca_system_score_codex":0.00018816313,"about_ca_system_score_gemma":0.00031400818,"threshold_uncertainty_score":0.9992516},"labels":[],"label_agreement":null},{"id":"W2103013209","doi":"10.1007/s11192-009-0020-3","title":"Matrices of science and technology interactions and patterns of structured growth: implications for development","year":2009,"lang":"en","type":"article","venue":"Scientometrics","topic":"Economic and Technological Innovation","field":"Economics, Econometrics and Finance","cited_by":41,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Conselho Nacional de Desenvolvimento Científico e Tecnológico; International Development Research Centre","keywords":"Dimension (graph theory); Articulation (sociology); Computer science; Component (thermodynamics); Key (lock); Scientific literature; Regional science; Data science; Sociology; Political science; Mathematics","score_opus":0.059368761363235184,"score_gpt":0.2929687513041478,"score_spread":0.2335999899409126,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2103013209","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9810764,0.00040747953,0.016514406,0.001149251,0.00008134004,0.0001674228,0.00007876622,0.00001988226,0.0005050513],"genre_scores_gemma":[0.9869646,0.00007008525,0.012886064,0.00003253387,0.0000044557446,0.000008915634,0.0000031113946,0.0000022790662,0.000027960978],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9991687,9.019705e-7,0.0004069901,0.00025299116,0.00003301316,0.00013741784],"domain_scores_gemma":[0.9990955,0.00005042479,0.0003480884,0.00013639881,0.00033798648,0.000031574968],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006003584,0.000057030506,0.00017059423,0.0045008603,0.00012032038,0.000028164734,0.00020979176,0.000054154545,0.0000054227935],"category_scores_gemma":[0.0009311485,0.000058965335,0.0000118492335,0.006246259,0.0003083748,0.0001916394,0.00006845313,0.000049302747,7.559952e-7],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000011125779,0.000026925389,0.14452596,0.000014609753,0.0000037826615,1.0353388e-8,0.0000502418,4.1116283e-7,0.0014468271,0.8344076,0.00001548974,0.019507024],"study_design_scores_gemma":[0.00023735041,0.00010027926,0.6831429,0.000008368225,0.000002848495,0.0000026379776,0.00008451492,0.00019471004,0.012804368,0.3013046,0.0020168035,0.000100617195],"about_ca_topic_score_codex":0.0000062163804,"about_ca_topic_score_gemma":0.0000023183431,"teacher_disagreement_score":0.53861696,"about_ca_system_score_codex":0.00005920239,"about_ca_system_score_gemma":0.000029424438,"threshold_uncertainty_score":0.4016071},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2109757953","doi":"10.1007/s11192-011-0515-6","title":"Exploring the web visibility of world-class universities","year":2011,"lang":"en","type":"article","venue":"Scientometrics","topic":"Web visibility and informetrics","field":"Computer Science","cited_by":44,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"University of British Columbia","keywords":"Visibility; Proxy (statistics); The Internet; Computer science; World Wide Web; Test (biology); Corporate governance; Information retrieval; Geography; Business","score_opus":0.22676155982887738,"score_gpt":0.28013886868423954,"score_spread":0.053377308855362154,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2109757953","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.86196464,0.00012351816,0.04310448,0.00018955719,0.002349965,0.00020496685,0.000009140083,0.00015244978,0.091901295],"genre_scores_gemma":[0.98916465,0.0000423004,0.009839698,0.00009632475,0.000024994924,0.0000027278224,3.985235e-7,0.0000036649403,0.00082526426],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99794966,0.00006681293,0.00032867794,0.00033900104,0.0009580336,0.00035778582],"domain_scores_gemma":[0.99757284,0.0004061844,0.00016220628,0.0011359314,0.00059703,0.00012579268],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0028797481,0.00011450955,0.00016174895,0.0043100584,0.0002076769,0.00011462221,0.00232329,0.000027123171,0.000041612726],"category_scores_gemma":[0.00083309424,0.00008257484,0.000113291855,0.042302538,0.00030691255,0.002138501,0.0007919016,0.00017252592,0.000046537873],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000033422264,0.00060147897,0.054225236,0.00013973001,0.000057889207,0.00000983531,0.021356717,0.000113446906,0.0002436468,0.7705525,0.0027188426,0.14994726],"study_design_scores_gemma":[0.0011136519,0.0007071398,0.824467,0.00005757432,0.000041561987,0.000007844661,0.0051981555,0.048310816,0.035971504,0.011144331,0.072054066,0.00092633045],"about_ca_topic_score_codex":0.00008843798,"about_ca_topic_score_gemma":0.00002615241,"teacher_disagreement_score":0.7702418,"about_ca_system_score_codex":0.00016704066,"about_ca_system_score_gemma":0.00029357226,"threshold_uncertainty_score":0.9780538},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2113501793","doi":"10.1007/s11192-015-1726-z","title":"Tree responses, tolerance and acclimation to stress: Does current research depend on the cultivation status of studied species?","year":2015,"lang":"en","type":"article","venue":"Scientometrics","topic":"Plant responses to elevated CO2","field":"Agricultural and Biological Sciences","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institute for Biological Sciences","funders":"Ministarstvo Prosvete, Nauke i Tehnološkog Razvoja","keywords":"Biology; Abiotic component; Acclimatization; Abiotic stress; Ecology; Tree (set theory); Botany; Gene","score_opus":0.2472618973623742,"score_gpt":0.3793187442655228,"score_spread":0.13205684690314862,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2113501793","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9963057,0.00042342508,0.0000067687524,0.0015580738,0.00019078088,0.00043195154,0.00024805282,0.000020774325,0.00081449514],"genre_scores_gemma":[0.9990481,0.00036647113,0.000073755036,0.000044935034,0.000073239426,0.000019159841,0.000026841659,7.9958374e-7,0.00034666862],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9970127,0.00042686748,0.00022185809,0.00032211703,0.0015572277,0.00045922794],"domain_scores_gemma":[0.99635804,0.0025238676,0.000101157886,0.00011352789,0.0006653085,0.00023810026],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0033341986,0.00010263491,0.00013296638,0.00045808806,0.00024107409,0.0001526206,0.00035642533,0.00004414289,0.000030354398],"category_scores_gemma":[0.010173918,0.000031272833,0.000024201247,0.009156636,0.0001145451,0.00013627602,0.0001933684,0.00018390403,0.000024667972],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0020485178,0.00067205983,0.1344614,0.000029098848,0.000021999715,0.0000059350587,0.0024633491,0.000115919494,0.34251973,0.004684142,0.011276489,0.50170135],"study_design_scores_gemma":[0.00025286243,0.0006853136,0.90807253,0.00006536279,0.000004614438,9.3189936e-7,0.0023212098,0.00024333256,0.023874594,0.0003153513,0.06402054,0.00014337101],"about_ca_topic_score_codex":0.000106088766,"about_ca_topic_score_gemma":0.0005563505,"teacher_disagreement_score":0.7736111,"about_ca_system_score_codex":0.00012741228,"about_ca_system_score_gemma":0.00004426876,"threshold_uncertainty_score":0.9981638},"labels":[],"label_agreement":null},{"id":"W2114111800","doi":"10.1007/s11192-009-0035-9","title":"Why it has become more difficult to predict Nobel Prize winners: a bibliometric analysis of nominees and winners of the chemistry and physics prizes (1901–2007)","year":2009,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":86,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Citation; Fragmentation (computing); Attribution; Hierarchy; Field (mathematics); Citation analysis; Data science; Sociology; Library science; Political science; Psychology; Computer science; Mathematics; Law","score_opus":0.2500444637851816,"score_gpt":0.4672281133466309,"score_spread":0.2171836495614493,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2114111800","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9890583,0.0017600422,0.0033364631,0.0035030313,0.0003197615,0.00054609275,0.00036780266,0.000025570262,0.0010829418],"genre_scores_gemma":[0.99534196,0.00045946438,0.0028283214,0.0006746614,0.000071258364,0.000008222213,0.000011109582,0.000015802616,0.00058918673],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9818282,0.00015578093,0.0015028804,0.0014305976,0.014086009,0.0009965448],"domain_scores_gemma":[0.9856835,0.004478426,0.0009552816,0.0016578439,0.00623274,0.0009921911],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.014614199,0.00040432945,0.0011627743,0.46733135,0.00040089778,0.0021269815,0.0036570553,0.00023661014,0.00011328191],"category_scores_gemma":[0.07676464,0.00025740347,0.00042822698,0.96132904,0.0011950658,0.00076726597,0.0016037195,0.00036974446,0.000004913722],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000841349,0.00093458913,0.75776744,0.00009212612,0.0004545572,0.000006762201,0.0016683943,0.001730066,0.024938686,0.000108580185,0.028421363,0.18379332],"study_design_scores_gemma":[0.0007807306,0.00031132874,0.969685,0.000030566425,0.00024808294,0.000004843462,0.00068022084,0.015901372,0.005036637,0.00022733095,0.006741942,0.00035195076],"about_ca_topic_score_codex":0.00013113688,"about_ca_topic_score_gemma":0.0000108477425,"teacher_disagreement_score":0.4939977,"about_ca_system_score_codex":0.00015832581,"about_ca_system_score_gemma":0.00029120225,"threshold_uncertainty_score":0.99998784},"labels":[],"label_agreement":null},{"id":"W2115701185","doi":"10.1007/s11192-007-2036-x","title":"History of the journal impact factor: Contingencies and consequences","year":2009,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":332,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Université du Québec à Montréal","funders":"","keywords":"Impact factor; Citation; Bibliometrics; Computer science; Factor (programming language); Data science; Library science; Operations research; Political science; Law; Mathematics","score_opus":0.6874658205142047,"score_gpt":0.5792682361347077,"score_spread":0.108197584379497,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2115701185","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9750594,0.012934373,0.0008045767,0.0007212727,0.0016288704,0.00016331048,0.000028931012,0.000012158176,0.008647101],"genre_scores_gemma":[0.9953232,0.0006607349,0.00035919508,0.00017156308,0.000045123197,3.211648e-7,1.7124403e-7,0.000004073224,0.003435623],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98498774,0.00036803505,0.00080786046,0.00040226686,0.0127982125,0.0006359106],"domain_scores_gemma":[0.9897329,0.0036259028,0.00074771885,0.0006658464,0.004657912,0.0005696746],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.02947145,0.0001613262,0.00035172785,0.08029769,0.00031147437,0.0011072627,0.0031542212,0.00009922497,0.00041323982],"category_scores_gemma":[0.07822573,0.00007991691,0.00023243073,0.2551544,0.0012612247,0.0007456327,0.00043384198,0.00038774623,0.000017634518],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024156012,0.00014280627,0.57381636,0.0000038788457,0.000017969796,0.000006182272,0.001584694,0.000026673124,0.017097304,0.00088516745,0.06077237,0.34562242],"study_design_scores_gemma":[0.00029848487,0.00038146615,0.95722824,0.000008484637,0.000007951964,0.00007679209,0.0003609042,0.00060483033,0.0012664278,0.0056630895,0.033955745,0.00014755432],"about_ca_topic_score_codex":0.000053374173,"about_ca_topic_score_gemma":0.0000041045673,"teacher_disagreement_score":0.3834119,"about_ca_system_score_codex":0.000628156,"about_ca_system_score_gemma":0.0011596772,"threshold_uncertainty_score":0.99992967},"labels":[],"label_agreement":null},{"id":"W2121258465","doi":"10.1007/s11192-011-0423-9","title":"Bibliographical research in the study of Hebrew printing: a bibliometric analysis","year":2011,"lang":"en","type":"article","venue":"Scientometrics","topic":"Library Collection Development and Digital Resources","field":"Computer Science","cited_by":8,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Hebrew; Bibliometrics; Quarter (Canadian coin); Field (mathematics); History; Period (music); State (computer science); Library science; Classics; Computer science; Philosophy; Mathematics; Archaeology; Algorithm","score_opus":0.2997094169988252,"score_gpt":0.3817411787088426,"score_spread":0.08203176171001741,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2121258465","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9645272,0.00026570488,0.002941426,0.00006610739,0.00009641455,0.00032243098,4.0848064e-7,0.000039659444,0.031740632],"genre_scores_gemma":[0.9978554,0.000035615703,0.0018841367,0.00006711621,0.000008253364,0.000020760539,3.322498e-7,0.000003033837,0.00012533728],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99548596,0.00034185866,0.00039779887,0.00046041986,0.0028847242,0.00042924163],"domain_scores_gemma":[0.9978655,0.00084090035,0.00010549096,0.00069486414,0.00038099452,0.00011222709],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0075138444,0.00009291418,0.00019345,0.8011888,0.00018221936,0.00076872826,0.00287231,0.000043517244,0.00002946012],"category_scores_gemma":[0.0007963581,0.000060905644,0.00010486916,0.9814398,0.00012046226,0.0010736318,0.00078133,0.0002309141,0.000016010767],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004852372,0.0011847668,0.9753365,0.0000048126535,0.000048685142,0.00001209928,0.004718684,0.000007186696,0.0000015637327,0.0020883724,0.000767143,0.015825363],"study_design_scores_gemma":[0.00024225308,0.0003181314,0.9952111,0.0000023498355,0.000009650801,0.000001200237,0.00085042685,0.00076629966,0.000056695557,0.0010084237,0.0014547156,0.000078755955],"about_ca_topic_score_codex":0.00017713553,"about_ca_topic_score_gemma":0.000025402851,"teacher_disagreement_score":0.18025099,"about_ca_system_score_codex":0.00001923694,"about_ca_system_score_gemma":0.00008604914,"threshold_uncertainty_score":0.74128646},"labels":[],"label_agreement":null},{"id":"W2127199286","doi":"10.1007/s11192-010-0331-4","title":"Exploring the management information systems discipline: a scientometric study of ICIS, PACIS and ASAC","year":2010,"lang":"en","type":"article","venue":"Scientometrics","topic":"Information Systems Theories and Implementation","field":"Social Sciences","cited_by":36,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lakehead University; Athabasca University","funders":"","keywords":"Delegate; Loyalty; Political science; Maturity (psychological); Information system; Library science; Identity (music); Field (mathematics); Management; Law; Computer science; Economics; Philosophy","score_opus":0.09714758848338591,"score_gpt":0.3650010739259219,"score_spread":0.267853485442536,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127199286","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97706324,0.000017297729,0.00091157435,0.00007050059,0.003073384,0.0014762805,0.000010697692,0.000035523837,0.017341517],"genre_scores_gemma":[0.999363,0.000060316375,0.00013180856,0.000016351985,0.00007536547,0.00012075686,0.000005087517,0.0000033960405,0.00022388351],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9970572,0.00009392906,0.00054669107,0.0001280847,0.0018756427,0.00029845914],"domain_scores_gemma":[0.9985872,0.00017375464,0.00035654023,0.00027163464,0.0004974186,0.000113465576],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0061947065,0.000089108944,0.00012913508,0.004435527,0.0011216344,0.0006909924,0.00039373958,0.000026878732,0.000016535056],"category_scores_gemma":[0.00051396,0.00006296631,0.000028070208,0.022914903,0.00021621173,0.002616808,0.00017649353,0.00011190614,0.000016629847],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000038496295,0.00026146296,0.046577863,0.00021400124,0.00007927262,6.21932e-7,0.39433083,0.00015552956,0.000034163822,0.37046486,0.00061917776,0.18722373],"study_design_scores_gemma":[0.0018178724,0.00018637998,0.11439874,0.000017870138,0.000045005938,0.000001701892,0.78497136,0.00050280575,0.000032411102,0.0001135869,0.09770161,0.00021066154],"about_ca_topic_score_codex":0.00078253157,"about_ca_topic_score_gemma":0.00016946302,"teacher_disagreement_score":0.39064053,"about_ca_system_score_codex":0.00007068533,"about_ca_system_score_gemma":0.00004220035,"threshold_uncertainty_score":0.9978535},"labels":[],"label_agreement":null},{"id":"W2127540269","doi":"10.1007/s11192-012-0729-2","title":"Counting citations in the field of business and management: why use Google Scholar rather than the Web of Science","year":2012,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":104,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"","keywords":"Web of science; Field (mathematics); Bibliometrics; Personalization; Scientometrics; Computer science; Library science; Data science; Sociology; Psychology; Political science; World Wide Web; Mathematics; MEDLINE; Law","score_opus":0.4692757887462023,"score_gpt":0.5425769812812546,"score_spread":0.07330119253505235,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127540269","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98823243,0.0013844221,0.0017214635,0.0012751675,0.000581145,0.0003995134,0.00001929888,0.0000057636976,0.0063808123],"genre_scores_gemma":[0.99739397,0.0005067901,0.0014248671,0.0003692921,0.000026873797,0.000009082928,5.6057564e-7,0.0000054933867,0.0002630868],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.985712,0.0002854629,0.00069662236,0.00046134382,0.012101539,0.0007430388],"domain_scores_gemma":[0.9795646,0.014024588,0.00046689576,0.0011850325,0.0045405985,0.00021827946],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.09023855,0.00012096726,0.00022197695,0.10341745,0.00063544017,0.0022955996,0.0038302557,0.000058905727,0.000051728264],"category_scores_gemma":[0.12700452,0.00006180266,0.00006694698,0.64596534,0.0014442573,0.0026046094,0.0010652951,0.00028426966,0.00001637487],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000007152057,0.00023358721,0.95974296,0.000020600288,0.0000066779694,0.0000011344981,0.0013449047,0.000037570615,0.0010512193,0.012762531,0.0038914725,0.02090021],"study_design_scores_gemma":[0.00025386616,0.000040626255,0.9702234,0.000016239952,0.000008614816,0.0000030241615,0.002744947,0.0012616065,0.0006892487,0.0007508964,0.02391418,0.00009333761],"about_ca_topic_score_codex":0.00018836274,"about_ca_topic_score_gemma":0.00003064539,"teacher_disagreement_score":0.5425479,"about_ca_system_score_codex":0.000053558713,"about_ca_system_score_gemma":0.00018888933,"threshold_uncertainty_score":0.99874014},"labels":[],"label_agreement":null},{"id":"W2139085708","doi":"10.1007/s11192-011-0603-7","title":"ASEAN benchmarking in terms of science, technology, and innovation from 1999 to 2009","year":2012,"lang":"en","type":"article","venue":"Scientometrics","topic":"Global trade and economics","field":"Economics, Econometrics and Finance","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institute on Governance","funders":"","keywords":"Benchmarking; Regional science; Scientometrics; Political science; Business; Data science; Knowledge management; Computer science; Library science; Sociology; Marketing","score_opus":0.06945695606174171,"score_gpt":0.26587844729325943,"score_spread":0.19642149123151773,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139085708","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9776813,0.000744177,0.00044971707,0.00018094327,0.0006883514,0.00012619559,0.00011656059,0.000013603655,0.019999169],"genre_scores_gemma":[0.9947583,0.000051621588,0.004918362,0.00015451208,0.000051646366,0.0000043123573,0.000006592676,0.000005846162,0.000048790083],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99861366,0.0000027797116,0.0005232958,0.00033881122,0.00007745357,0.00044401243],"domain_scores_gemma":[0.999342,0.00003298938,0.00021970211,0.00025049227,0.000040108356,0.0001146788],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0020627365,0.00009036923,0.00023442086,0.026313908,0.00008443224,0.000069351496,0.0003379466,0.00008250154,0.000040468276],"category_scores_gemma":[0.0007342769,0.00010720072,0.000018043233,0.053836618,0.00031490458,0.0005893244,0.00019078514,0.000098389275,0.00007656736],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000001567384,0.00004511898,0.7840638,0.0000034075458,0.0000020262257,1.5599664e-7,0.0001747396,0.00001901465,0.0002508665,0.2113278,0.00010185099,0.0040096263],"study_design_scores_gemma":[0.00030821003,0.000056628363,0.9450662,0.000014926394,0.0000016567412,0.000001200846,0.00011381964,0.00062862324,0.0015397337,0.028347256,0.023724973,0.00019674428],"about_ca_topic_score_codex":0.000121897065,"about_ca_topic_score_gemma":0.000008406612,"teacher_disagreement_score":0.18298054,"about_ca_system_score_codex":0.00016160082,"about_ca_system_score_gemma":0.000016222342,"threshold_uncertainty_score":0.984722},"labels":[],"label_agreement":null},{"id":"W2139770987","doi":"10.1007/s11192-006-0127-8","title":"Canadian collaboration networks: A comparative analysis of the natural sciences, social sciences and the humanities","year":2006,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":215,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Digital humanities; Humanities; Natural science; The arts; Citation index; Index (typography); Sociology; Social science; Citation; Library science; Political science; Epistemology; Computer science; Philosophy; World Wide Web","score_opus":0.43608455016924813,"score_gpt":0.5388048553623775,"score_spread":0.10272030519312941,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2139770987","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9516325,0.0055379667,0.00064213277,0.004157518,0.0013402897,0.0006155912,0.00012917444,0.000015599904,0.035929225],"genre_scores_gemma":[0.99766755,0.00003482786,0.00020691761,0.00036625692,0.000082766346,0.000008522271,0.0000037511709,0.000002866187,0.0016265344],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9856075,0.00071395846,0.0007841504,0.00073701766,0.0113566555,0.00080070825],"domain_scores_gemma":[0.98838294,0.006461841,0.0006148186,0.00044104853,0.003980749,0.000118588294],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.045255516,0.00017393919,0.0005177922,0.12537162,0.00542787,0.008359406,0.0038698844,0.00008531764,0.000072147865],"category_scores_gemma":[0.012719167,0.0000822153,0.00025502633,0.75563353,0.009273396,0.000741326,0.00067605835,0.00026276932,0.0000070178535],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000030437637,0.000103575105,0.5593642,0.0000042972524,0.00016232397,0.0000016001686,0.002907607,0.02553765,0.000054577366,0.3731199,0.033742134,0.0049716895],"study_design_scores_gemma":[0.0005735985,0.000063907515,0.6948568,0.0000036809804,0.00012390276,0.0000017886155,0.0037867024,0.28230578,0.00007599826,0.0045476207,0.013459983,0.00020023761],"about_ca_topic_score_codex":0.04476495,"about_ca_topic_score_gemma":0.13636485,"teacher_disagreement_score":0.63026196,"about_ca_system_score_codex":0.0003015797,"about_ca_system_score_gemma":0.0014160976,"threshold_uncertainty_score":0.99586695},"labels":[],"label_agreement":null},{"id":"W2140162241","doi":"10.1007/s11192-011-0589-1","title":"Author disambiguation using multi-aspect similarity indicators","year":2011,"lang":"en","type":"article","venue":"Scientometrics","topic":"Advanced Text Analysis Techniques","field":"Computer Science","cited_by":48,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Social Sciences and Humanities Research Council of Canada; Genome Canada","keywords":"Similarity (geometry); Recall; Precision and recall; Computer science; Task (project management); Measure (data warehouse); Information retrieval; Key (lock); Data mining; Artificial intelligence; Natural language processing; Psychology; Cognitive psychology","score_opus":0.20036793714287734,"score_gpt":0.3850923397812263,"score_spread":0.18472440263834897,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2140162241","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06664888,0.00007596157,0.9314774,0.000038421644,0.00029422637,0.00014131174,0.0000020713671,0.00044541506,0.00087626715],"genre_scores_gemma":[0.55480254,0.0000030720614,0.4450245,0.000050520845,0.000012374585,0.000003334687,0.0000010514233,0.0000062903405,0.000096309035],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99773276,0.00007072586,0.00030025182,0.00058342825,0.00090902316,0.00040379516],"domain_scores_gemma":[0.9984842,0.00006259727,0.00023685933,0.0008169024,0.00019717048,0.00020230534],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0014503961,0.000154435,0.00018253864,0.0056326855,0.0002468751,0.00017071022,0.0014374069,0.00009159531,0.000027306176],"category_scores_gemma":[0.0009301903,0.0001457812,0.00010276734,0.029235197,0.00012419117,0.0011284503,0.0005025721,0.00017776723,0.000032558073],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009446847,0.0018646246,0.3969854,0.000044108485,0.000110020366,0.00007000408,0.010153784,0.0002472493,0.008101798,0.125403,0.0006532187,0.45635733],"study_design_scores_gemma":[0.00059774757,0.00017903237,0.3545432,0.000027531969,0.00007081143,0.000015265747,0.000116224415,0.52875143,0.090665884,0.021524392,0.0024248082,0.001083638],"about_ca_topic_score_codex":0.000073182295,"about_ca_topic_score_gemma":0.000011718475,"teacher_disagreement_score":0.5285042,"about_ca_system_score_codex":0.00024355153,"about_ca_system_score_gemma":0.00008151416,"threshold_uncertainty_score":0.9913989},"labels":[],"label_agreement":null},{"id":"W2141246632","doi":"10.1023/a:1016007205344","title":"Contribution of Hacettepe University Faculty of Medicine to the world's biomedical literature (1988-1997)","year":2002,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Turkish; Science Citation Index; Library science; Quarter (Canadian coin); Citation; MEDLINE; Bibliometrics; Political science; Medical education; Medicine; Computer science; History","score_opus":0.38260139108866204,"score_gpt":0.5123331948262367,"score_spread":0.1297318037375747,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2141246632","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8270816,0.008086269,0.043394364,0.089746326,0.005856201,0.0020423552,0.0032019054,0.000105607774,0.020485317],"genre_scores_gemma":[0.9876325,0.0001803516,0.00077860587,0.00026475437,0.000120343364,0.0000012675667,0.0000353337,0.000006896274,0.01097996],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.97939265,0.00038544412,0.000959283,0.0007836906,0.017739499,0.00073942926],"domain_scores_gemma":[0.98215014,0.004967541,0.0005808116,0.001168145,0.010304725,0.00082862],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.029692553,0.0001902538,0.0005595677,0.1754737,0.00032167442,0.00039661044,0.0043919366,0.00015536784,0.0012024626],"category_scores_gemma":[0.119134694,0.00010633474,0.00021125418,0.7391614,0.0011882654,0.00044604414,0.0010381409,0.0003952731,0.00021741471],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000057451467,0.0005029702,0.037470795,0.000017050079,0.000038524817,0.000026996411,0.0013428053,0.00006844608,0.0019403773,0.007372724,0.7886925,0.16246936],"study_design_scores_gemma":[0.0012674049,0.0004744406,0.118648805,0.00005394365,0.000019253566,0.0000067676674,0.0004687295,0.0048400178,0.0014398205,0.0006224755,0.8719763,0.00018201883],"about_ca_topic_score_codex":0.000055290613,"about_ca_topic_score_gemma":0.000027522818,"teacher_disagreement_score":0.5636877,"about_ca_system_score_codex":0.00021560832,"about_ca_system_score_gemma":0.00009047071,"threshold_uncertainty_score":0.99971056},"labels":[],"label_agreement":null},{"id":"W2142066148","doi":"10.1007/s11192-015-1676-5","title":"Benchmarking bibliometrics in biomedical research: research performance of the University of Toronto’s Faculty of Medicine, 2008–2012","year":2015,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Hospital for Sick Children; University of Toronto","funders":"","keywords":"Bibliometrics; Benchmarking; Citation; Citation analysis; Web of science; Library science; Impact factor; Medical education; MEDLINE; Computer science; Medicine; Political science; Management","score_opus":0.8554237139510291,"score_gpt":0.6280879688612137,"score_spread":0.22733574508981547,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2142066148","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97074926,0.005911822,0.00039847963,0.00083512056,0.0013091587,0.0005700495,0.000113564405,0.000008939991,0.02010359],"genre_scores_gemma":[0.99511844,0.0018351391,0.0012762776,0.000008806418,0.00008168112,0.0000015438246,0.0000059540794,0.000012153271,0.0016600172],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.93211627,0.0022440546,0.0017912084,0.0010779446,0.06114908,0.0016214364],"domain_scores_gemma":[0.9467135,0.015157316,0.00090553216,0.0021757702,0.033871733,0.0011761212],"candidate_categories":["metaresearch","bibliometrics","sts","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.25774965,0.00020151296,0.0008022854,0.3996344,0.00040005674,0.0001999276,0.009313801,0.00033268382,0.00080709043],"category_scores_gemma":[0.2301776,0.00012935871,0.00019471558,0.8761707,0.0067171846,0.0013183093,0.004277602,0.0010736196,0.0000434158],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014440298,0.0008517007,0.69681776,0.00009461898,0.000019148043,0.000009516255,0.0029290183,0.0001483899,0.0016378337,0.0011921974,0.15040843,0.14574696],"study_design_scores_gemma":[0.0021434515,0.0017956053,0.88497525,0.00016894873,0.000008075362,0.0000058654005,0.008069461,0.011161698,0.0016871506,0.0013820861,0.08838829,0.00021411874],"about_ca_topic_score_codex":0.0085304035,"about_ca_topic_score_gemma":0.0006619983,"teacher_disagreement_score":0.4765363,"about_ca_system_score_codex":0.001025088,"about_ca_system_score_gemma":0.0020934455,"threshold_uncertainty_score":0.99807185},"labels":[],"label_agreement":null},{"id":"W2143612877","doi":"10.1007/s11192-011-0580-x","title":"Validating online reference managers for scholarly impact measurement","year":2011,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":260,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; York University","funders":"","keywords":"Citation; Computer science; Citation impact; Sample (material); Citation analysis; Web of science; Information retrieval; Scholarly communication; Impact factor; World Wide Web; Data science; Political science; MEDLINE; Publishing","score_opus":0.9469149544521874,"score_gpt":0.6472795575330219,"score_spread":0.2996353969191655,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2143612877","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8102085,0.0018029857,0.13279562,0.00034027637,0.002484117,0.0015697593,0.0005850005,0.00018067272,0.050033085],"genre_scores_gemma":[0.96615493,0.00008863856,0.03145721,0.00016474507,0.00009907644,0.00002988908,0.00001709604,0.00002809965,0.0019602978],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.968701,0.00027037435,0.0012499349,0.0015562727,0.026639005,0.001583426],"domain_scores_gemma":[0.9707146,0.003109583,0.0006914015,0.001836327,0.02224485,0.0014031943],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.09005316,0.0003228453,0.00050356897,0.18648191,0.00068039267,0.0069597946,0.005891509,0.00019925224,0.00094686856],"category_scores_gemma":[0.23355797,0.00022214335,0.0004149593,0.492865,0.00024840876,0.0030273125,0.0011143233,0.0004992308,0.0004698049],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012026401,0.0011978964,0.23904707,0.000028177426,0.000095182506,0.00001899053,0.00066901074,0.00006849066,0.0037167408,0.004157508,0.024212776,0.7266679],"study_design_scores_gemma":[0.00209964,0.0016555167,0.8873448,0.000036939426,0.000048129867,0.000013965155,0.0015528861,0.010463558,0.005317414,0.024259526,0.066176124,0.0010315485],"about_ca_topic_score_codex":0.00024279398,"about_ca_topic_score_gemma":0.000029896744,"teacher_disagreement_score":0.72563636,"about_ca_system_score_codex":0.000882061,"about_ca_system_score_gemma":0.00058745296,"threshold_uncertainty_score":0.9999664},"labels":[],"label_agreement":null},{"id":"W2145694442","doi":"10.1007/s11192-010-0278-5","title":"Is the academic Ivory Tower becoming a managed structure? A nested analysis of the variance in activities of researchers from natural sciences and engineering in Canada","year":2010,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université Laval","funders":"Social Sciences and Humanities Research Council of Canada; Natural Sciences and Engineering Research Council of Canada","keywords":"Ivory tower; Autonomy; Context (archaeology); Variance (accounting); Sociology; Natural (archaeology); Adaptation (eye); Institutional theory; Social science; Political science; Psychology; Geography; Law; Business","score_opus":0.3143140146529494,"score_gpt":0.4913641469124948,"score_spread":0.1770501322595454,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2145694442","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9973474,0.00092942605,0.00003841584,0.00071460236,0.0005701259,0.00021846182,0.00008268875,0.0000025126592,0.0000963677],"genre_scores_gemma":[0.9993473,0.00006970411,0.00037627618,0.0000824009,0.000012414965,0.0000034594664,6.6656884e-7,0.0000046730192,0.000103047336],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98824805,0.0002584451,0.00076713687,0.0006657005,0.009414913,0.00064577],"domain_scores_gemma":[0.9841796,0.013815248,0.00039454838,0.0007736201,0.0006656064,0.0001713842],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.023630831,0.0001443833,0.00043674925,0.09748745,0.00019347749,0.0005183533,0.004082181,0.000107056345,0.00010928355],"category_scores_gemma":[0.057198226,0.00007916786,0.000106695035,0.5808077,0.00076256087,0.00054791494,0.0011815027,0.0012145229,2.509117e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000008496918,0.000013997169,0.98062366,0.0000049806195,0.000030397034,0.0000019323397,0.0009995882,0.0015256724,0.008892696,0.00016821174,0.000109735796,0.0076206485],"study_design_scores_gemma":[0.00021641837,0.0000110938045,0.8948194,0.000012089274,0.000014020122,5.562162e-7,0.0015195423,0.10055468,0.0022078957,0.00045284504,0.0001051838,0.00008627036],"about_ca_topic_score_codex":0.35006437,"about_ca_topic_score_gemma":0.56524616,"teacher_disagreement_score":0.48332027,"about_ca_system_score_codex":0.00027265825,"about_ca_system_score_gemma":0.0012316637,"threshold_uncertainty_score":0.9507434},"labels":[],"label_agreement":null},{"id":"W2153046986","doi":"10.1007/s11192-012-0891-6","title":"Discovering and assessing fields of expertise in nanomedicine: a patent co-citation network perspective","year":2012,"lang":"en","type":"article","venue":"Scientometrics","topic":"Intellectual Property and Patents","field":"Business, Management and Accounting","cited_by":33,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Quality (philosophy); Patent analysis; Modularity (biology); Data science; Computer science; Patent office; Perspective (graphical); Patent visualisation; Intellectual property; Simple (philosophy); Business; Citation; Patent portfolio; Knowledge management; Industrial organization; Engineering; Artificial intelligence; World Wide Web","score_opus":0.239983206839301,"score_gpt":0.3122728714126446,"score_spread":0.0722896645733436,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2153046986","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9834958,0.0012501895,0.002587848,0.00009591523,0.00060946686,0.0001259596,4.471991e-7,0.00001362792,0.011820765],"genre_scores_gemma":[0.9991127,0.000026361133,0.00013957769,0.00027474295,0.00035343601,0.00000418554,0.0000038384765,0.0000069661073,0.000078179495],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9990631,0.0000113728875,0.00018114902,0.00013910854,0.00032241963,0.00028289633],"domain_scores_gemma":[0.99959785,0.00006835815,0.00010456601,0.00007591235,0.00013579031,0.000017520535],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00089223677,0.00008142502,0.00012794982,0.0013258187,0.00009224899,0.00012582485,0.000085702704,0.00004400291,0.000053479864],"category_scores_gemma":[0.000804904,0.00006433035,0.00002348443,0.0045457445,0.000060949587,0.0012773375,0.00007621756,0.00008983237,0.000014106032],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000056873418,0.00023755348,0.9558433,0.00021608753,0.00001809571,0.0000039338606,0.00883526,0.0006738151,0.0009906695,0.0051672016,0.001340813,0.02661639],"study_design_scores_gemma":[0.0017392021,0.00008383198,0.92582256,0.0005105312,0.000053165335,0.00000431235,0.013901752,0.05158628,0.000816323,0.0015995807,0.003302328,0.00058015535],"about_ca_topic_score_codex":0.0006089257,"about_ca_topic_score_gemma":0.000018425737,"teacher_disagreement_score":0.050912466,"about_ca_system_score_codex":0.00006443504,"about_ca_system_score_gemma":0.000010175337,"threshold_uncertainty_score":0.26233155},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2165122403","doi":"10.1007/s11192-006-0115-z","title":"Benchmarking scientific output in the social sciences and humanities: The limits of existing databases","year":2006,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":506,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Benchmarking; Bibliometrics; Benchmark (surveying); Database; Bibliographic database; Computer science; Data science; Humanities; Library science; Geography; Management","score_opus":0.8593883432201264,"score_gpt":0.609782680977184,"score_spread":0.24960566224294234,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2165122403","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9622965,0.002488917,0.0007473676,0.0009150481,0.0009381889,0.00036706106,0.00007395211,0.000014547473,0.032158423],"genre_scores_gemma":[0.9974456,0.00004071513,0.000840781,0.00013877799,0.00020497724,0.000010502574,0.0000063023595,0.000006962935,0.0013053843],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9785785,0.0006287548,0.0010892374,0.0010849964,0.017603062,0.0010154401],"domain_scores_gemma":[0.97927314,0.016922055,0.00061378593,0.0008311887,0.0022730045,0.00008680503],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.1096677,0.00019499946,0.0003290084,0.09872509,0.003340084,0.008372738,0.0050019273,0.00006397161,0.000053477812],"category_scores_gemma":[0.042466298,0.00009916171,0.0001264725,0.48992166,0.0042155017,0.0010241781,0.0010909914,0.00035715642,0.00002100305],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013964016,0.0005389262,0.64587545,0.000038921375,0.0000094992565,0.00002085178,0.005744464,0.00029302444,0.00083869847,0.15052934,0.044210084,0.15188676],"study_design_scores_gemma":[0.00055380835,0.0002009237,0.84608567,0.00003439848,0.000014937524,0.000020836429,0.012896725,0.008677776,0.0006360377,0.012397576,0.11810393,0.0003774033],"about_ca_topic_score_codex":0.00061841053,"about_ca_topic_score_gemma":0.00032557797,"teacher_disagreement_score":0.39119658,"about_ca_system_score_codex":0.000088103574,"about_ca_system_score_gemma":0.00032179165,"threshold_uncertainty_score":0.99849445},"labels":[],"label_agreement":null},{"id":"W2226524777","doi":"10.1007/s11192-015-1825-x","title":"How to boost scientific production? A statistical analysis of research funding and other influencing factors","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":137,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Quality (philosophy); Work (physics); Production (economics); Relation (database); Impact factor; Statistical analysis; Business; Marketing; Political science; Economics; Computer science; Statistics; Mathematics; Engineering; Microeconomics","score_opus":0.7558819682205911,"score_gpt":0.6277056305859157,"score_spread":0.1281763376346754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2226524777","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97538346,0.0002480292,0.020243814,0.001788594,0.0009827956,0.00041431043,0.00033213166,0.000027903794,0.0005789458],"genre_scores_gemma":[0.9891614,0.000017919803,0.0023091617,0.000027378368,0.00005451293,0.0000104403225,0.0000021506107,0.000014889618,0.008402177],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.96931726,0.0005634705,0.00087940705,0.0020031785,0.02569866,0.001538015],"domain_scores_gemma":[0.96522546,0.01815615,0.00035888527,0.0017940674,0.012726195,0.0017392618],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.09974855,0.00020882227,0.0006025062,0.4767476,0.0009271588,0.0074615646,0.0025741048,0.00012406956,0.00034432148],"category_scores_gemma":[0.3991574,0.000117539435,0.00016430025,0.7982325,0.0018128299,0.0011355258,0.001701463,0.00026152498,0.000098767356],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016677262,0.00010814637,0.9263775,0.000008778137,0.000083605184,0.0000029118844,0.0007880807,0.000027860131,0.019895807,0.0045163003,0.0046243216,0.043550026],"study_design_scores_gemma":[0.00024930958,0.0002517567,0.9469564,0.000026717704,0.000057071327,0.0000018969605,0.0022264076,0.0006100925,0.009076896,0.0023162907,0.0379461,0.00028109134],"about_ca_topic_score_codex":0.00012039453,"about_ca_topic_score_gemma":0.00009319081,"teacher_disagreement_score":0.3214849,"about_ca_system_score_codex":0.00043435112,"about_ca_system_score_gemma":0.00036772818,"threshold_uncertainty_score":0.9935688},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2264463174","doi":"10.1007/s11192-016-1846-0","title":"Factors influencing research collaboration in LIS schools in South Africa","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":31,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"University of South Africa","keywords":"Bureaucracy; Library science; Productivity; Medical education; Population; Public relations; Psychology; Sociology; Political science; Medicine; Economic growth; Computer science","score_opus":0.7350202013577418,"score_gpt":0.6208490315127927,"score_spread":0.11417116984494913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2264463174","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9906318,0.00073485594,0.0006557208,0.0009051833,0.0007326345,0.00056053215,0.0000675769,0.000033469005,0.0056782207],"genre_scores_gemma":[0.99639803,0.000086808315,0.00041810327,0.000027011158,0.000057018027,0.00003033521,0.0000016917224,0.000018713892,0.0029622575],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9657006,0.0011704097,0.00150628,0.0017150487,0.027746128,0.002161516],"domain_scores_gemma":[0.971301,0.016063169,0.00036653606,0.001599673,0.009592143,0.001077432],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.11080867,0.0002549729,0.00051031256,0.49142045,0.0003508513,0.004645818,0.0043446654,0.0002920899,0.00040166415],"category_scores_gemma":[0.3862859,0.00014706452,0.00011687501,0.8414483,0.0003983643,0.002547789,0.0014909299,0.0007956754,0.0009874034],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021088566,0.00014015404,0.96061635,0.0000033644578,0.0000022060267,0.00001669694,0.0037783762,0.000046527886,0.0017449015,0.00061443076,0.0026039619,0.030411916],"study_design_scores_gemma":[0.0012986293,0.00023008013,0.9503058,0.0000624748,0.0000012514514,5.751936e-7,0.0120748775,0.0007106786,0.003132249,0.009093488,0.022717817,0.00037203188],"about_ca_topic_score_codex":0.00013789732,"about_ca_topic_score_gemma":0.00024373917,"teacher_disagreement_score":0.35002786,"about_ca_system_score_codex":0.0019613714,"about_ca_system_score_gemma":0.0012376616,"threshold_uncertainty_score":0.99979043},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2270179582","doi":"10.1007/s11192-015-1828-7","title":"Automated Research Impact Assessment: a new bibliometrics approach","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"Evaluation and Performance Assessment","field":"Decision Sciences","cited_by":33,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"U.S. Food and Drug Administration; National Institutes of Health; National Institute of Environmental Health Sciences; Réseau Provincial de Recherche en Adaptation-Réadaptation","keywords":"Bibliometrics; Resource (disambiguation); Benchmark (surveying); Computer science; Management science; Data science; Library science; Engineering; Geography","score_opus":0.6967903147110235,"score_gpt":0.6874424372991984,"score_spread":0.009347877411825078,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2270179582","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5418639,0.0006420049,0.2162102,0.0050355545,0.0025086624,0.0012672313,0.00009059326,0.0009172055,0.23146464],"genre_scores_gemma":[0.9632463,0.00016126665,0.015639607,0.0001309856,0.00019343436,0.000019914578,0.0000077985615,0.00002208249,0.020578627],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.97665507,0.0008609963,0.0010198152,0.0011070597,0.01915693,0.0012001456],"domain_scores_gemma":[0.9862934,0.006228049,0.00037263863,0.0015860513,0.004489397,0.0010305174],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.06576112,0.00023960258,0.00039497874,0.116039366,0.0005627021,0.0020718279,0.0025833403,0.00017146739,0.0044099195],"category_scores_gemma":[0.032625973,0.00012132884,0.00022520231,0.41322955,0.00031930063,0.0018108455,0.00061828695,0.00035501167,0.0033322307],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000018263507,0.0003152726,0.23926719,0.000004828919,0.000026282292,0.000003503389,0.00031699587,0.0002884184,0.0011618069,0.0038604664,0.38468292,0.37005407],"study_design_scores_gemma":[0.0018114697,0.00050375226,0.7955891,0.00002105336,0.000009974387,0.000015217668,0.0003743796,0.08485515,0.00031187446,0.008278965,0.10783882,0.00039023362],"about_ca_topic_score_codex":0.000080420345,"about_ca_topic_score_gemma":0.000005186433,"teacher_disagreement_score":0.5563219,"about_ca_system_score_codex":0.00090755435,"about_ca_system_score_gemma":0.0028314725,"threshold_uncertainty_score":0.99896413},"labels":[],"label_agreement":null},{"id":"W2271548616","doi":"10.1007/s11192-016-1852-2","title":"iSEER: an intelligent automatic computer system for scientific evaluation of researchers","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"Online Learning and Analytics","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Computer science; Data science; Information retrieval","score_opus":0.21723149811846532,"score_gpt":0.42346161614393535,"score_spread":0.20623011802547003,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2271548616","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20287551,0.00010348566,0.7944215,0.0006451614,0.0013030932,0.00035635123,0.000010927445,0.00014623528,0.0001377571],"genre_scores_gemma":[0.925708,0.0000021400228,0.07371392,0.000010563788,0.000074854586,0.000012247165,0.00000406625,0.000008109264,0.00046612896],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.996005,0.0002531317,0.00036463205,0.0005343652,0.0024459471,0.00039691335],"domain_scores_gemma":[0.996207,0.00043999802,0.0001998386,0.0008242523,0.0021104715,0.00021843499],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.014702676,0.00010338673,0.00017411953,0.0040315003,0.0002224264,0.00042046164,0.0013154268,0.000053373322,0.000009406509],"category_scores_gemma":[0.0015832133,0.00007022752,0.00009055128,0.011355542,0.0001975521,0.00053207023,0.00022531666,0.00006248649,0.00004740453],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000025227862,0.00017288706,0.0012083079,0.00013216249,0.000019489922,6.661563e-7,0.0006835384,0.0018960881,0.0013155725,0.026451537,0.000718293,0.96739894],"study_design_scores_gemma":[0.0004161682,0.00021909499,0.0011021051,0.0000999182,0.000018529645,0.0000020801222,0.000068165,0.9926604,0.0025374934,0.0012779674,0.0014870432,0.00011103184],"about_ca_topic_score_codex":0.0000036244305,"about_ca_topic_score_gemma":0.0000017905247,"teacher_disagreement_score":0.9907643,"about_ca_system_score_codex":0.00032568543,"about_ca_system_score_gemma":0.00040976473,"threshold_uncertainty_score":0.5455961},"labels":[],"label_agreement":null},{"id":"W2280374399","doi":"10.1007/s11192-015-1807-z","title":"Quantifying the scientific output of new researchers using the zp-index","year":2015,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":13,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Disadvantage; Index (typography); Computer science; Metric (unit); Citation; Position (finance); Quality (philosophy); Bibliometrics; Scientific literature; Data science; Citation index; Operations research; Library science; Mathematics; Marketing; Economics; World Wide Web; Epistemology; Business; Artificial intelligence","score_opus":0.9550993674466709,"score_gpt":0.6719727838580462,"score_spread":0.28312658358862475,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2280374399","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93958366,0.0059018144,0.03765344,0.0044144955,0.00556417,0.0008520088,0.000036454327,0.00004433195,0.0059496197],"genre_scores_gemma":[0.9840502,0.00004293285,0.0017659124,0.000121371544,0.0001921975,0.0000040957343,0.0000018045247,0.000020955296,0.013800516],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9508038,0.00093480933,0.0013106144,0.0012979019,0.04405071,0.0016021507],"domain_scores_gemma":[0.9568042,0.014736644,0.0011390125,0.004768381,0.020087076,0.0024646942],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.18200314,0.0002549535,0.00044676568,0.1317207,0.0014463781,0.010658663,0.011210214,0.00016100133,0.00016577695],"category_scores_gemma":[0.26241818,0.0001202309,0.0003309421,0.6987687,0.0024168538,0.0011238639,0.003133236,0.00075011933,0.00033170832],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008076246,0.00029619117,0.48426223,0.0000134709035,0.00005384137,0.000016138049,0.005734059,0.006513942,0.0014244035,0.0037505077,0.23635527,0.2614992],"study_design_scores_gemma":[0.0020091427,0.00040706675,0.14527929,0.000038149985,0.000041234358,0.000045456247,0.020975498,0.2806717,0.003899914,0.017665206,0.5282596,0.0007078125],"about_ca_topic_score_codex":0.0009794218,"about_ca_topic_score_gemma":0.00009258748,"teacher_disagreement_score":0.567048,"about_ca_system_score_codex":0.00043948097,"about_ca_system_score_gemma":0.0033158995,"threshold_uncertainty_score":0.9998536},"labels":[],"label_agreement":null},{"id":"W2287531290","doi":"10.1007/s11192-016-1848-y","title":"The effect of holding a research chair on scientists’ productivity","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Productivity; Novelty; Matching (statistics); Set (abstract data type); Field (mathematics); Computer science; Psychology; Statistics; Mathematics; Economics; Social psychology; Economic growth","score_opus":0.7473508328001948,"score_gpt":0.6643655597955263,"score_spread":0.08298527300466851,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2287531290","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9797401,0.00072425534,0.00131532,0.0029660556,0.0027899889,0.00082038774,0.00004222076,0.000038465343,0.011563243],"genre_scores_gemma":[0.9868195,0.00016913896,0.00010405413,0.000011840638,0.00016088429,0.000036835754,3.864746e-7,0.000017586159,0.012679789],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9428851,0.0024390898,0.0011031021,0.0019229711,0.04950253,0.0021472152],"domain_scores_gemma":[0.8820871,0.10123921,0.0005950801,0.0036537272,0.011456199,0.0009686986],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.31661305,0.00024386321,0.00047753885,0.19908656,0.0019412878,0.003301554,0.0073662833,0.0001413917,0.00011493149],"category_scores_gemma":[0.545146,0.00009615324,0.00028839827,0.6657676,0.0027678607,0.0008130836,0.0022205987,0.0005210298,0.00085968134],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000108340595,0.00012657001,0.2115906,0.000010882653,0.000012860889,0.000005552972,0.00006717925,0.000009806328,0.011977594,0.0047006123,0.022298865,0.74909115],"study_design_scores_gemma":[0.0022315257,0.0054053804,0.4698769,0.00011264347,0.000009143299,0.000015197744,0.00017128202,0.00074682856,0.2789234,0.009931369,0.23206158,0.0005147954],"about_ca_topic_score_codex":0.00002823019,"about_ca_topic_score_gemma":0.00000889988,"teacher_disagreement_score":0.74857634,"about_ca_system_score_codex":0.0005108209,"about_ca_system_score_gemma":0.0003504165,"threshold_uncertainty_score":0.99994606},"labels":[],"label_agreement":null},{"id":"W2308596676","doi":"10.1007/s11192-016-1923-4","title":"“Listing author contribution” does not alter the author inflation in the publications in basic research in four major gastroenterology journals in 10 years","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"Health and Medical Research Impacts","field":"Medicine","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Hepatology; Listing (finance); Internal medicine; Medicine; Gastroenterology; Impact factor; Political science; Law","score_opus":0.3680554104418968,"score_gpt":0.5282761647174701,"score_spread":0.16022075427557325,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2308596676","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.73884785,0.00008517282,0.00017464501,0.25977364,0.00012155523,0.00076954387,0.000016020462,0.0000067372594,0.00020481872],"genre_scores_gemma":[0.9965853,0.000100386635,0.00008864624,0.0021955424,0.00013771922,0.000119182354,0.000006429034,0.000007705515,0.0007590739],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99348897,0.0015557765,0.0007207033,0.00032461304,0.0022633476,0.0016465709],"domain_scores_gemma":[0.99012214,0.0071625714,0.00015188009,0.00053317,0.0007893742,0.001240891],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.058909856,0.000091572336,0.00023207015,0.008784912,0.000117059106,0.00011655446,0.00046499263,0.0001356485,0.000187415],"category_scores_gemma":[0.29962647,0.000040949726,0.00003654836,0.017412363,0.00030853794,0.00026434567,0.0001258924,0.0011551249,0.00006289796],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032648828,0.00027364062,0.9350943,0.000041285533,0.000003980642,0.000093525916,0.0013382183,0.0000041000917,0.0016573261,0.00076776417,0.0060692765,0.05433007],"study_design_scores_gemma":[0.0025607173,0.00021233199,0.98218846,0.00024673357,0.0000021106764,0.000016697777,0.00033551583,0.0005126963,0.00007675523,0.0005984833,0.013209796,0.00003969221],"about_ca_topic_score_codex":0.0002967541,"about_ca_topic_score_gemma":0.0019612005,"teacher_disagreement_score":0.25773746,"about_ca_system_score_codex":0.0009612596,"about_ca_system_score_gemma":0.0016735505,"threshold_uncertainty_score":0.96905035},"labels":[],"label_agreement":null},{"id":"W2340061204","doi":"10.1007/s11192-016-1953-y","title":"Characterization, description, and considerations for the use of funding acknowledgement data in Web of Science","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":121,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Social Sciences and Humanities Research Council of Canada; Fonds de Recherche du Québec-Société et Culture; Austrian Science Fund; Multiple Sclerosis Society; Impact Fund","keywords":"Acknowledgement; Citation; Computer science; Citation index; Library science; Science Citation Index; Bibliometrics; Index (typography); Scientometrics; Data science; Political science; World Wide Web","score_opus":0.8693361966360683,"score_gpt":0.5759535144275153,"score_spread":0.293382682208553,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2340061204","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8921446,0.00032043748,0.10391869,0.0013191287,0.00095562905,0.0006445516,0.0006125288,0.0000069090897,0.00007751803],"genre_scores_gemma":[0.9939979,0.0004799074,0.0049152444,0.000037135505,0.000024350862,0.000011309658,0.000003794743,0.000005120336,0.0005252762],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99139744,0.00010431869,0.00093649945,0.0007579539,0.0063215876,0.00048221182],"domain_scores_gemma":[0.95254314,0.030124981,0.0007329033,0.0022990662,0.01395954,0.00034036004],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.04605835,0.00009658866,0.00024244092,0.10291528,0.00039691658,0.0014502677,0.002508396,0.000045341774,0.000114417286],"category_scores_gemma":[0.36143756,0.0000492015,0.000039007307,0.32556343,0.002002122,0.002766682,0.0017667165,0.00006190809,0.000008367295],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000016940223,0.00022621934,0.64958555,0.000016519767,0.0000103505645,4.5587484e-7,0.00033486992,0.000021669017,0.21301134,0.011333207,0.004210554,0.121232316],"study_design_scores_gemma":[0.0012187925,0.00019422578,0.8849644,0.000052436553,0.000015771977,0.0000034501948,0.0003141648,0.0443354,0.015794115,0.0044037965,0.04849716,0.00020626638],"about_ca_topic_score_codex":0.000017685321,"about_ca_topic_score_gemma":0.00006611066,"teacher_disagreement_score":0.31537923,"about_ca_system_score_codex":0.00011335717,"about_ca_system_score_gemma":0.00085828133,"threshold_uncertainty_score":0.99958634},"labels":[],"label_agreement":null},{"id":"W2489636575","doi":"10.1007/s11192-016-2094-z","title":"Global neuroscience and mental health research: a bibliometrics case study","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":15,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canada Research Chairs; University of Toronto","funders":"Faculty of Arts and Sciences; University of Toronto","keywords":"Bibliometrics; Mental health; Citation; Public health; Citation analysis; Social network analysis; Psychology; Gerontology; Library science; Medicine; Sociology; Social science; Psychiatry; Computer science","score_opus":0.8921782491564988,"score_gpt":0.7000762143745934,"score_spread":0.19210203478190535,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2489636575","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9792413,0.0024845374,0.0062059658,0.0061749467,0.002243348,0.0014290295,0.0003165502,0.00009791809,0.0018063708],"genre_scores_gemma":[0.9954416,0.0010078377,0.0007887029,0.00036498602,0.00009355159,0.000026213747,8.0478793e-7,0.000020809995,0.0022554782],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9439397,0.0021868355,0.0016979671,0.003217576,0.045863487,0.003094433],"domain_scores_gemma":[0.9732373,0.011228657,0.00056411617,0.0023201092,0.009001081,0.0036487875],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.18153585,0.0003737144,0.0006664534,0.5509861,0.0026286077,0.00893684,0.0049269726,0.00014897849,0.00017546891],"category_scores_gemma":[0.22749335,0.00021543005,0.00016181271,0.9261959,0.0018187593,0.0022878665,0.0052224942,0.00043205897,0.0003165629],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013015356,0.00064890453,0.53365463,0.0000037106588,0.000004854585,0.0005890521,0.0003562862,0.0000014567478,0.00007900839,0.00048095232,0.03398333,0.43018478],"study_design_scores_gemma":[0.0038804335,0.0051875003,0.802931,0.000026841042,0.0000063334764,0.0043282197,0.013583754,0.0017125923,0.000047348218,0.005301657,0.16216376,0.0008305142],"about_ca_topic_score_codex":0.0008608061,"about_ca_topic_score_gemma":0.0002653798,"teacher_disagreement_score":0.42935425,"about_ca_system_score_codex":0.0011585227,"about_ca_system_score_gemma":0.0010561439,"threshold_uncertainty_score":0.99866986},"labels":[],"label_agreement":null},{"id":"W2517558721","doi":"10.1007/s11192-016-2098-8","title":"University–industry R&D linkage metrics: validity and applicability in world university rankings","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"Innovation and Knowledge Management","field":"Business, Management and Accounting","cited_by":41,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Directorate-General for Education and Culture; Newfoundland and Labrador; Ministerie van Onderwijs, Cultuur en Wetenschap; European Commission","keywords":"Ranking (information retrieval); Linkage (software); Metric (unit); Computer science; Data science; Index (typography); Strengths and weaknesses; Information retrieval; Operations research; Data mining; Mathematics; Marketing; Psychology; Business; World Wide Web","score_opus":0.05807529621679081,"score_gpt":0.25030281545862426,"score_spread":0.19222751924183346,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2517558721","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.67716384,0.000022731472,0.0051755765,0.0016652239,0.000548836,0.00038930855,0.000009254714,0.00016809157,0.31485713],"genre_scores_gemma":[0.98152345,0.000014977438,0.0003783974,0.0003203903,0.0001393381,3.909426e-7,0.0000048294824,0.000008418319,0.017609816],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99853355,0.000031175998,0.00017538424,0.00046603335,0.00045458053,0.00033925506],"domain_scores_gemma":[0.99897736,0.0002080653,0.00015168166,0.000293761,0.00033633513,0.000032809185],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0026268528,0.00014029682,0.00017140049,0.010013203,0.00020616195,0.0001543717,0.00040120652,0.00011470019,0.00020621838],"category_scores_gemma":[0.00086624443,0.00012477642,0.000045757573,0.038387477,0.00016490939,0.0008167187,0.0006124422,0.00021560836,0.00023844161],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003854123,0.0001971645,0.71600384,0.000114888935,0.000010280785,0.00001951598,0.000041610005,0.0000020951777,0.00009393642,0.24945953,0.003582604,0.030435989],"study_design_scores_gemma":[0.001345325,0.000004858448,0.1292542,0.000030852632,0.000024838075,2.5596788e-7,0.00034561125,0.00021689312,0.00001890902,0.0015482953,0.86693823,0.0002717024],"about_ca_topic_score_codex":0.00008886478,"about_ca_topic_score_gemma":0.00026555316,"teacher_disagreement_score":0.86335564,"about_ca_system_score_codex":0.00031697552,"about_ca_system_score_gemma":0.0000251967,"threshold_uncertainty_score":0.9820521},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2524791359","doi":"10.1007/s11192-016-2134-8","title":"The linguistic patterns and rhetorical structure of citation context: an approach using n-grams","year":2016,"lang":"en","type":"article","venue":"Scientometrics","topic":"Topic Modeling","field":"Computer Science","cited_by":50,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Université du Québec à Montréal","funders":"","keywords":"Rhetorical question; Citation; Linguistics; Context (archaeology); Computer science; Linguistic context; Function (biology); Natural (archaeology); Natural language processing; Information retrieval; Linguistic analysis; History; Library science; Philosophy; Biology","score_opus":0.08286568018633636,"score_gpt":0.31700194935228493,"score_spread":0.23413626916594857,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2524791359","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4092207,0.000083046754,0.5901917,0.000033963413,0.00037702103,0.000050035076,0.0000030368537,0.000015353175,0.000025112313],"genre_scores_gemma":[0.9626554,0.0000088113575,0.037225477,0.00002188315,0.00006384227,5.666193e-7,4.369285e-7,0.0000037298405,0.000019841456],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9986974,0.0000668227,0.00018635808,0.00029595263,0.0005499657,0.0002034971],"domain_scores_gemma":[0.9988305,0.0003087219,0.00010824207,0.0004034295,0.00025095226,0.00009814324],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007221449,0.000072240444,0.000092365706,0.000492243,0.00015618376,0.00018082403,0.000571169,0.00004254263,8.407481e-7],"category_scores_gemma":[0.0012404454,0.000039582857,0.000020120959,0.0024055745,0.000080834405,0.00022174948,0.00016544797,0.00005785301,3.6149223e-7],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004558396,0.000055961696,0.06857125,0.000035298064,0.000012423536,0.0000018266538,0.0032938137,0.0002860831,0.006552438,0.14211494,0.000008265692,0.77906317],"study_design_scores_gemma":[0.00053674943,0.000106804495,0.037400834,0.000037149377,0.000011767871,0.000018847286,0.00019492552,0.946605,0.0017674393,0.0127042355,0.0003728088,0.00024342308],"about_ca_topic_score_codex":0.000041549876,"about_ca_topic_score_gemma":0.0000073739034,"teacher_disagreement_score":0.9463189,"about_ca_system_score_codex":0.00005509664,"about_ca_system_score_gemma":0.000043279968,"threshold_uncertainty_score":0.17436902},"labels":[],"label_agreement":null},{"id":"W2579971484","doi":"10.1007/s11192-017-2283-4","title":"How to reduce the number of rating scale items without predictability loss?","year":2017,"lang":"en","type":"article","venue":"Scientometrics","topic":"Psychometric Methodologies and Testing","field":"Decision Sciences","cited_by":39,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Laurentian University","funders":"European Social Fund","keywords":"Predictability; Rating scale; Receiver operating characteristic; Statistics; Scale (ratio); Reliability (semiconductor); Confirmatory factor analysis; Computer science; Econometrics; Mathematics; Structural equation modeling","score_opus":0.5004341825675646,"score_gpt":0.5355242632854347,"score_spread":0.03509008071787012,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2579971484","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95333606,0.000104457424,0.022722062,0.0029715262,0.0025115188,0.00033775292,0.00002391916,0.000036421432,0.017956272],"genre_scores_gemma":[0.95590013,0.000008592112,0.039979696,0.00008058698,0.00019194199,0.000010241706,3.4033266e-7,0.00000873036,0.0038197304],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9934588,0.00045503792,0.0007144988,0.00082636724,0.0040115616,0.0005337535],"domain_scores_gemma":[0.9710071,0.021828309,0.0013367828,0.0036873696,0.0017885411,0.00035190553],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.038198587,0.00016531229,0.00043389184,0.001670183,0.0011507277,0.0023658401,0.004555833,0.00008918661,0.00009284819],"category_scores_gemma":[0.62063795,0.00009242002,0.00016081956,0.018928913,0.0007552415,0.0006426634,0.0011694068,0.00023960149,0.000060132388],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010060326,0.000033671095,0.9044705,0.000005542201,0.000004271328,7.003969e-7,0.0003743768,0.000020765396,0.0004377992,0.00014495988,0.0021094012,0.09238795],"study_design_scores_gemma":[0.0002543657,0.0000527118,0.9816327,0.000019374298,0.000008819454,0.0000114808445,0.0010234884,0.0011439697,0.002310631,0.004919627,0.008476413,0.00014643758],"about_ca_topic_score_codex":0.00009263085,"about_ca_topic_score_gemma":0.000010622051,"teacher_disagreement_score":0.58243936,"about_ca_system_score_codex":0.00005571187,"about_ca_system_score_gemma":0.00009943529,"threshold_uncertainty_score":0.9986698},"labels":[],"label_agreement":null},{"id":"W2585136728","doi":"10.1007/s11192-017-2252-y","title":"Development trend forecasting for coherent light generator technology based on patent citation network analysis","year":2017,"lang":"en","type":"article","venue":"Scientometrics","topic":"Innovation Diffusion and Forecasting","field":"Decision Sciences","cited_by":63,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National University of Defense Technology; National Natural Science Foundation of China","keywords":"Trademark; Computer science; Technology forecasting; Autoregressive integrated moving average; Citation; Network analysis; Development (topology); Patent office; Data science; Operations research; Artificial intelligence; Time series; Engineering; Machine learning; Mathematics","score_opus":0.36211515557363344,"score_gpt":0.4023682443934626,"score_spread":0.04025308881982914,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2585136728","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6810729,0.00004642702,0.30879956,0.0017067404,0.0023084355,0.00055856083,0.000030143352,0.00009776165,0.0053795036],"genre_scores_gemma":[0.92223203,6.749288e-7,0.07598914,0.0003219713,0.00013940448,0.0000668422,0.00003387668,0.000012503843,0.001203539],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99542814,0.000043503573,0.0009398152,0.00075107435,0.0022842446,0.00055323145],"domain_scores_gemma":[0.9957308,0.00062495144,0.0011690482,0.00093806273,0.0013844981,0.00015262743],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.007963471,0.00019341505,0.00036428845,0.0118797105,0.0021490552,0.0012757999,0.0012040309,0.00014102373,0.00017457624],"category_scores_gemma":[0.017181713,0.00014453805,0.00018914616,0.02979638,0.000097963246,0.0002114149,0.0002182095,0.00012767619,0.000049655366],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004947263,0.00016912282,0.13503753,0.000006946086,0.000079680874,0.0000044847106,0.00028649587,0.017384484,0.00042117655,0.004247686,0.007818945,0.834494],"study_design_scores_gemma":[0.0010038897,0.00013285094,0.051407702,0.000025848845,0.00006351523,7.7225826e-7,0.00018126723,0.8349457,0.0025339492,0.001543164,0.1078258,0.00033557424],"about_ca_topic_score_codex":0.0000016827968,"about_ca_topic_score_gemma":0.000066845496,"teacher_disagreement_score":0.8341584,"about_ca_system_score_codex":0.0001689057,"about_ca_system_score_gemma":0.00014014349,"threshold_uncertainty_score":0.999761},"labels":[],"label_agreement":null},{"id":"W2592688325","doi":"10.1007/s11192-017-2318-x","title":"Evolution monitoring for innovation sources using patent cluster analysis","year":2017,"lang":"en","type":"article","venue":"Scientometrics","topic":"Innovation Diffusion and Forecasting","field":"Decision Sciences","cited_by":10,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"University of Waterloo; National University of Defense Technology; National Natural Science Foundation of China","keywords":"Trademark; Computer science; Data science; Key (lock); Cluster (spacecraft); Cluster analysis; TRACE (psycholinguistics); Patent analysis; Citation; Scale (ratio); Spectral clustering; Data mining; Knowledge management; Artificial intelligence; World Wide Web; Geography","score_opus":0.4820857377646642,"score_gpt":0.4696832937935274,"score_spread":0.012402443971136767,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2592688325","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.670333,0.000018627812,0.32726282,0.00015568412,0.001428473,0.00014289809,0.000010082153,0.00002088647,0.0006274991],"genre_scores_gemma":[0.987481,8.2543994e-7,0.010611438,0.000066202934,0.00026296428,0.0000070177985,0.000005734163,0.000007859519,0.0015569483],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9960391,0.000045109315,0.0008067291,0.0005125931,0.0022476402,0.00034881363],"domain_scores_gemma":[0.9938993,0.00045132436,0.0011961575,0.00081985386,0.0035608087,0.00007258184],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.009797764,0.00012060167,0.0002450931,0.013595746,0.0019647041,0.002155757,0.0010019324,0.00008909284,0.000054578566],"category_scores_gemma":[0.0299106,0.00009509381,0.00015634151,0.03727334,0.00012495228,0.0008910192,0.00031033106,0.00009331971,0.00002974955],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000015779724,0.000039887273,0.9411489,0.000003964958,0.000035010977,3.7417763e-7,0.00027973705,0.0057704425,0.0027064811,0.004368864,0.0003349467,0.045295652],"study_design_scores_gemma":[0.00047207336,0.00002676273,0.54858744,0.000009456842,0.00006509512,8.0465446e-7,0.0003767355,0.44240323,0.0014418357,0.004308491,0.0021336447,0.00017443848],"about_ca_topic_score_codex":0.00004361418,"about_ca_topic_score_gemma":0.0000055631226,"teacher_disagreement_score":0.43663278,"about_ca_system_score_codex":0.00018972319,"about_ca_system_score_gemma":0.000071287344,"threshold_uncertainty_score":0.9993346},"labels":[],"label_agreement":null},{"id":"W2596958794","doi":"10.1007/s11192-017-2350-x","title":"The impact of individual collaborative activities on knowledge creation and transmission","year":2017,"lang":"en","type":"article","venue":"Scientometrics","topic":"Knowledge Management and Sharing","field":"Social Sciences","cited_by":23,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Context (archaeology); Knowledge management; Knowledge creation; Scopus; General partnership; NetLogo; Data science; Collaborative network; Computer science; Loyalty; Business; Political science; Marketing; Geography","score_opus":0.0632680637246128,"score_gpt":0.42701903140369407,"score_spread":0.36375096767908127,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2596958794","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.67845476,0.00055757863,0.00006838896,0.0001573297,0.00021131871,0.00018730784,0.000007363989,0.00001312544,0.32034284],"genre_scores_gemma":[0.9941004,0.00046421232,0.000037121026,0.0000016961812,0.00005919722,0.0000035920766,9.349521e-7,0.0000031036727,0.005329728],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9990224,0.00008946627,0.000091389964,0.00013026873,0.00047966358,0.00018675854],"domain_scores_gemma":[0.9990572,0.00039405172,0.00013685378,0.00017085532,0.00016352689,0.0000775289],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.0023180319,0.00006202804,0.000086752625,0.00057571207,0.0026353446,0.00080779206,0.00045136025,0.000040071027,0.000017487713],"category_scores_gemma":[0.0010777721,0.000039638962,0.00004023918,0.0018950194,0.0005896534,0.00033631548,0.000094688345,0.000061277715,0.0000042739553],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004490144,0.00013370127,0.07825953,0.000013983833,0.000055330063,5.314801e-7,0.0884008,0.000008887935,0.00011172248,0.03639914,0.0026146166,0.7939569],"study_design_scores_gemma":[0.0006537594,0.00033964904,0.92779034,0.000063377396,0.000032989017,6.785854e-8,0.013148105,0.0005395573,0.0008768594,0.0020672819,0.054291267,0.00019672637],"about_ca_topic_score_codex":0.00014292209,"about_ca_topic_score_gemma":0.00015960216,"teacher_disagreement_score":0.8495308,"about_ca_system_score_codex":0.00007515148,"about_ca_system_score_gemma":0.00013706626,"threshold_uncertainty_score":0.99866307},"labels":[],"label_agreement":null},{"id":"W2605555090","doi":"10.1007/s11192-017-2387-x","title":"Western classical music development: a statistical analysis of composers similarity, differentiation and evolution","year":2017,"lang":"en","type":"article","venue":"Scientometrics","topic":"Music and Audio Processing","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"China Scholarship Council","keywords":"Similarity (geometry); Scientometrics; Statistical analysis; Sound analysis; Classical music; Musical; Association (psychology); Order (exchange); Evolutionary biology; Computer science; Biology; Artificial intelligence; Epistemology; Literature; Art; Mathematics; Statistics; Philosophy; World Wide Web; Physics","score_opus":0.05810941451857577,"score_gpt":0.3146336132466485,"score_spread":0.25652419872807275,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2605555090","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.41433114,0.00004278441,0.5851435,0.00013407257,0.00014192254,0.000032525844,0.000003706643,0.000015088368,0.00015525403],"genre_scores_gemma":[0.9693025,0.0000025237985,0.030571546,0.0000507,0.000015871952,0.0000013919401,0.0000073508027,0.000002492752,0.00004561913],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99852884,0.00003146242,0.00022736266,0.00029580368,0.00068079156,0.00023575014],"domain_scores_gemma":[0.9989657,0.0001112585,0.00026115726,0.0003721542,0.00014716301,0.00014260632],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000664423,0.000087264256,0.00021050111,0.0017426956,0.00045433023,0.0005060011,0.0005987297,0.000048831993,0.000008086883],"category_scores_gemma":[0.00034611803,0.00007854686,0.00003478562,0.00342354,0.00018388168,0.00060923817,0.00050303695,0.00007758335,0.0000032144026],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000025589122,0.0001142078,0.91599965,0.000039193677,0.0000935795,0.0000013380112,0.0008172656,0.000022095042,0.00038692812,0.015223227,0.00006949942,0.067230426],"study_design_scores_gemma":[0.00014726412,0.0000113624255,0.9239712,0.000008095383,0.00008010294,5.998598e-7,0.0000066554053,0.07499486,0.0002226323,0.00026090784,0.00020770456,0.00008864942],"about_ca_topic_score_codex":0.000009470835,"about_ca_topic_score_gemma":0.000032634427,"teacher_disagreement_score":0.55497134,"about_ca_system_score_codex":0.00006656245,"about_ca_system_score_gemma":0.00007026261,"threshold_uncertainty_score":0.48793802},"labels":[],"label_agreement":null},{"id":"W2719511526","doi":"10.1007/s11192-017-2438-3","title":"Online distribution channel increases article usage on Mendeley: a randomized controlled trial","year":2017,"lang":"en","type":"article","venue":"Scientometrics","topic":"Academic Writing and Publishing","field":"Arts and Humanities","cited_by":21,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre for Addiction and Mental Health; Brain and Cognition Discovery Foundation; Institute for Work & Health; University of Toronto; University Health Network; Institute of Health Services and Policy Research; MaRS","funders":"Ontario Centres of Excellence","keywords":"Randomized controlled trial; Computer science; Control (management); Distribution (mathematics); Medicine; Artificial intelligence; Mathematics; Surgery","score_opus":0.06917204241122363,"score_gpt":0.3152505920396897,"score_spread":0.24607854962846606,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2719511526","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95279723,0.00019613864,0.0002094904,0.0015210208,0.0023785515,0.0014176148,0.00033500505,0.00013752612,0.041007448],"genre_scores_gemma":[0.9905735,0.000036905396,0.000019995616,0.0002919216,0.0023032357,0.000079095014,0.00010179017,0.000011567573,0.0065819677],"study_design_codex":"randomized_trial","study_design_gemma":"randomized_trial","domain_scores_codex":[0.9976943,0.00022688662,0.0004928161,0.00030235123,0.00087717327,0.00040643412],"domain_scores_gemma":[0.9971951,0.0014037963,0.0005302449,0.0003916137,0.00030256168,0.0001766318],"candidate_categories":["metaresearch","sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.006307178,0.00016664037,0.0007135814,0.0006683351,0.0016441849,0.002310783,0.0005705681,0.00008199752,0.00034939212],"category_scores_gemma":[0.032267686,0.0001152486,0.0002882454,0.00032139328,0.000515237,0.00071565376,0.00011988466,0.00029967388,0.00005376789],"study_design_candidate":"randomized_trial","study_design_consensus":"randomized_trial","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.7079282,0.0016917638,0.00009425905,0.000024710987,0.00016846464,0.0000119431415,0.004460305,0.00002976519,0.000016007945,0.24967745,0.03089528,0.005001889],"study_design_scores_gemma":[0.96879685,0.00016365318,0.00010862372,0.00004538493,0.00010385782,9.396261e-7,0.0006744076,0.006797451,0.000024494684,0.0026451827,0.020422889,0.00021624578],"about_ca_topic_score_codex":0.0004677745,"about_ca_topic_score_gemma":0.000065885506,"teacher_disagreement_score":0.2608687,"about_ca_system_score_codex":0.000073983974,"about_ca_system_score_gemma":0.00006381011,"threshold_uncertainty_score":0.99965554},"labels":[],"label_agreement":null},{"id":"W2724711068","doi":"10.1007/s11192-017-2459-y","title":"The scaling relationship between degree centrality of countries and their citation-based performance on Management Information Systems","year":2017,"lang":"en","type":"article","venue":"Scientometrics","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Centrality; Degree (music); Citation; Exponent; Scaling; Degree distribution; Power law; Citation impact; China; Computer science; Complex network; Econometrics; Mathematics; Statistics; Political science; Physics; Law; Library science","score_opus":0.0761607441796123,"score_gpt":0.3116552854228882,"score_spread":0.23549454124327587,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2724711068","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94996464,0.00004766177,0.042510808,0.00009898615,0.00010452078,0.00026206195,0.00003578077,0.000018087188,0.0069574397],"genre_scores_gemma":[0.99959767,0.000006665988,0.00027045686,0.000006405529,0.0000356973,0.000011408261,0.000022051165,0.0000026241578,0.000047028796],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9991317,0.000031913674,0.00025047475,0.00009182462,0.0003486495,0.00014548232],"domain_scores_gemma":[0.99848586,0.00052065816,0.00037339187,0.00039993192,0.00018196822,0.000038157217],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013826151,0.00007501111,0.000106142485,0.00048731186,0.000980043,0.00049104507,0.00029712514,0.000016932867,0.0000025048257],"category_scores_gemma":[0.0000707947,0.000051292333,0.000036956506,0.00080912124,0.00014008915,0.00031991277,0.00006192111,0.00006896451,0.0000054683846],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000031826698,0.0000059372637,0.9312338,0.00002822988,0.000016445365,7.502476e-9,0.000082350394,0.00057375844,1.8334245e-7,0.040913343,0.0000810004,0.02706176],"study_design_scores_gemma":[0.0001370176,0.000016029053,0.9768681,0.000040316336,0.000018870664,1.1783343e-8,0.00016550823,0.01908134,0.00012579559,0.00087349373,0.0026107272,0.000062818996],"about_ca_topic_score_codex":0.00004920575,"about_ca_topic_score_gemma":0.0000011082379,"teacher_disagreement_score":0.049633007,"about_ca_system_score_codex":0.000036568308,"about_ca_system_score_gemma":0.000018055582,"threshold_uncertainty_score":0.7537797},"labels":[],"label_agreement":null},{"id":"W2734032589","doi":"10.1007/s11192-017-2452-5","title":"Tracking the emergence of synthetic biology","year":2017,"lang":"en","type":"article","venue":"Scientometrics","topic":"Gene Regulatory Network Analysis","field":"Biochemistry, Genetics and Molecular Biology","cited_by":109,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Office of Naval Research; Office of Defense Programs; National Key Research and Development Program of China; Biotechnology and Biological Sciences Research Council; Engineering and Physical Sciences Research Council; Natural Sciences and Engineering Research Council of Canada; Defense Advanced Research Projects Agency; Directorate for Biological Sciences; National Institutes of Health; Bundesministerium für Bildung und Forschung; Russian Foundation for Basic Research; National Natural Science Foundation of China; Deutsche Forschungsgemeinschaft; Wellcome Trust; Council of Scientific and Industrial Research, India; U.S. Department of Energy; U.S. Department of Defense; European Commission; Canadian Institutes of Health Research; National Science Foundation; American Heart Association; Ministry of Education, Culture, Sports, Science and Technology; Howard Hughes Medical Institute","keywords":"Synthetic biology; Systems biology; Domain (mathematical analysis); Computer science; Set (abstract data type); Benchmark (surveying); Multidisciplinary approach; Data science; Biology; Computational biology; Mathematics; Political science","score_opus":0.03642355739386737,"score_gpt":0.32851887235283084,"score_spread":0.29209531495896346,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2734032589","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9926937,0.001580049,0.003008665,0.00020196103,0.0004005225,0.00006538563,0.000005616296,0.0000035259595,0.0020405382],"genre_scores_gemma":[0.99905026,0.00021031068,0.00024257302,0.000025500996,0.000096018004,0.0000027480062,0.0000045894076,0.000006416781,0.0003615691],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99917406,0.000041450556,0.00015293462,0.00024515597,0.00018598865,0.00020039604],"domain_scores_gemma":[0.9985289,0.000017921808,0.00020958147,0.0010231213,0.00017115926,0.00004929112],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008364201,0.00007407833,0.00010350511,0.00023904628,0.00031562004,0.000040506096,0.00088702573,0.00006813406,0.000034224846],"category_scores_gemma":[0.0010275554,0.00005166066,0.000108754604,0.00084111426,0.0003761695,0.0000029825565,0.00025226857,0.000042879536,0.0000072005187],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012692345,0.000075900345,0.33604503,0.000017022643,0.00013801683,9.535326e-7,0.00008753241,0.0008272759,0.5895551,0.001276123,0.0024422598,0.06952211],"study_design_scores_gemma":[0.00038392143,0.00024559948,0.3406213,0.000015592856,0.00013380843,0.000010517568,0.0001679736,0.0018110318,0.6046208,0.0009085478,0.050694484,0.0003864427],"about_ca_topic_score_codex":0.000012634854,"about_ca_topic_score_gemma":0.000014020991,"teacher_disagreement_score":0.06913567,"about_ca_system_score_codex":0.000003892438,"about_ca_system_score_gemma":0.000035724475,"threshold_uncertainty_score":0.2427526},"labels":[],"label_agreement":null},{"id":"W2775516587","doi":"10.1007/s11192-017-2593-6","title":"Assessing the effect of the United States’ “citation advantage” on other countries’ scientific impact as measured in the Web of Science (WoS) database","year":2017,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":49,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Citation; Web of science; Relation (database); Citation database; Visibility; Citation analysis; Computer science; Database; Citation impact; Data science; Political science; Regional science; Library science; Scopus; Sociology; Geography; Law; MEDLINE","score_opus":0.4570312116954996,"score_gpt":0.6150197274678524,"score_spread":0.1579885157723528,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2775516587","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9921231,0.00024248712,0.0003370418,0.00097448274,0.0019005104,0.0009372737,0.00020027017,0.000009458673,0.0032753695],"genre_scores_gemma":[0.99940544,0.000045318433,0.00006423931,0.00014719885,0.000031486896,0.000013333337,0.0000070330225,0.000013713205,0.00027224587],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9587404,0.0017289486,0.0011571341,0.0011931257,0.036023866,0.0011565202],"domain_scores_gemma":[0.95961416,0.023175828,0.0021851638,0.0063910736,0.008257183,0.00037661582],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.2244995,0.00028395472,0.00046045138,0.12177536,0.0037416134,0.015463735,0.016700862,0.00009725178,0.00011205795],"category_scores_gemma":[0.2981451,0.00011081327,0.00027942008,0.5253583,0.008851208,0.0026728292,0.0016753853,0.00058482157,0.000078387035],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000113147355,0.00030028395,0.9568405,0.000035317855,0.000018989518,0.0000079387855,0.0017323726,0.0018985745,0.015025116,0.0015512467,0.0034718497,0.019004663],"study_design_scores_gemma":[0.0014675295,0.00069202296,0.9217571,0.00011587703,0.000027569311,0.0000085825,0.0027048727,0.031225316,0.031818394,0.0012038321,0.0087111,0.00026777934],"about_ca_topic_score_codex":0.00081091886,"about_ca_topic_score_gemma":0.00006985458,"teacher_disagreement_score":0.403583,"about_ca_system_score_codex":0.0003891235,"about_ca_system_score_gemma":0.0017638543,"threshold_uncertainty_score":0.9975554},"labels":[],"label_agreement":null},{"id":"W2784275186","doi":"10.1007/s11192-018-2991-4","title":"Edge Factors: Scientific Frontier Positions of Nations","year":2018,"lang":"en","type":"preprint","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"National Institute on Aging; National Institutes of Health","keywords":"Frontier; Work (physics); Position (finance); China; Political science; Enhanced Data Rates for GSM Evolution; Data science; Sociology; Epistemology; Public relations; Computer science; Engineering; Business; Law; Telecommunications","score_opus":0.6933349039972806,"score_gpt":0.6011293296615132,"score_spread":0.09220557433576737,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2784275186","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8979332,0.0031797665,0.033309385,0.00050351897,0.022732448,0.0012125255,0.003409441,0.00012341453,0.037596337],"genre_scores_gemma":[0.9709293,0.00014907883,0.008144205,0.000034747714,0.0003484687,0.000033022603,0.00025339396,0.000044873294,0.020062868],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9555929,0.0005347067,0.0024249677,0.0030055528,0.036915086,0.0015267918],"domain_scores_gemma":[0.9462385,0.00840401,0.0020214678,0.0048170937,0.037091125,0.0014277706],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","sts","open_science","insufficient_payload"],"category_scores_codex":[0.053577498,0.0005760624,0.0011582659,0.52635723,0.0015819465,0.012970526,0.011868752,0.0007754094,0.0027281912],"category_scores_gemma":[0.19140348,0.00044042733,0.0009515544,0.701194,0.0028271282,0.0010429792,0.009692999,0.0011652807,0.0010479243],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002664888,0.0022581255,0.39396006,0.00017011739,0.00018895455,0.000013743357,0.0023504826,0.0005941267,0.0017739418,0.0058294428,0.5652619,0.027572457],"study_design_scores_gemma":[0.0013401198,0.000727473,0.58576685,0.0002724732,0.00019471029,0.000010501634,0.0017085158,0.017906899,0.018023362,0.09850065,0.27316105,0.002387385],"about_ca_topic_score_codex":0.000112900176,"about_ca_topic_score_gemma":0.000042762713,"teacher_disagreement_score":0.29210085,"about_ca_system_score_codex":0.000673702,"about_ca_system_score_gemma":0.0024070058,"threshold_uncertainty_score":0.99988663},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2795529249","doi":"10.1007/s11192-018-2732-8","title":"Can Twitter increase the visibility of Chinese publications?","year":2018,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":22,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; McGill University","funders":"Fonds de Recherche du Québec-Société et Culture; Andrew W. Mellon Foundation","keywords":"Visibility; Social media; Citation; China; Political science; Library science; Computer science; Geography; World Wide Web","score_opus":0.5383689152925458,"score_gpt":0.5912490065647195,"score_spread":0.052880091272173635,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2795529249","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96969336,0.00027217172,0.0027456244,0.005366747,0.0013832361,0.0003821808,0.00008202449,0.000034952554,0.020039676],"genre_scores_gemma":[0.99265933,0.000024721443,0.0009858559,0.00059492793,0.00025631327,0.000013127723,0.000004766575,0.000010365961,0.0054506017],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.97975105,0.00044550953,0.0011445164,0.0010198916,0.01682199,0.0008170138],"domain_scores_gemma":[0.9506176,0.0114873415,0.0008553653,0.00419161,0.031827778,0.0010202732],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.06967175,0.00019309807,0.0003565547,0.13477136,0.00065837736,0.0026180057,0.0065782056,0.000112762595,0.0016527589],"category_scores_gemma":[0.3528187,0.000097065844,0.0002302911,0.7164576,0.0017481673,0.0006596506,0.0017635167,0.00029813283,0.0005692058],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014496874,0.00024895545,0.8961473,0.0000036025485,0.000011027422,6.646571e-7,0.00051791564,0.0000052859154,0.00038164636,0.0011915013,0.061440058,0.04003753],"study_design_scores_gemma":[0.0003015943,0.00018313453,0.9253265,0.0000020206844,0.0000045855654,0.0000040948344,0.00020834425,0.0039487095,0.000733805,0.019010196,0.050125718,0.00015124872],"about_ca_topic_score_codex":0.00043327236,"about_ca_topic_score_gemma":0.00010175389,"teacher_disagreement_score":0.58168626,"about_ca_system_score_codex":0.00016675744,"about_ca_system_score_gemma":0.0004933741,"threshold_uncertainty_score":0.9992599},"labels":[],"label_agreement":null},{"id":"W2888141929","doi":"10.1007/s11192-018-2889-1","title":"Bibliometric study of family business succession between 1939 and 2017: mapping and analyzing authors’ networks","year":2018,"lang":"en","type":"article","venue":"Scientometrics","topic":"Family Business Performance and Succession","field":"Business, Management and Accounting","cited_by":216,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; HEC Montréal","funders":"","keywords":"Scopus; Scope (computer science); Ecological succession; Family business; Social network analysis; Scientometrics; Bibliometrics; Data science; Citation impact; Sociology; Citation; Computer science; Knowledge management; Library science; Social science; Political science; Management; Social capital; Economics; MEDLINE","score_opus":0.07934125063747188,"score_gpt":0.31367773812755384,"score_spread":0.23433648749008196,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2888141929","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9936663,0.0011330853,0.0026356732,0.000060940652,0.0009458981,0.0003213004,0.0000018382273,0.00007512839,0.0011598681],"genre_scores_gemma":[0.9982703,0.00029480265,0.00019481604,0.00013064725,0.0010064067,0.000005682007,0.000013850737,0.000024433319,0.000059099813],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9973231,0.000020933769,0.00056671037,0.00059745193,0.0009835189,0.00050826167],"domain_scores_gemma":[0.9977188,0.00013737039,0.000480114,0.00035250728,0.0012524445,0.000058736496],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.002291296,0.00026758356,0.00045643246,0.10285612,0.0006938124,0.0008171863,0.00043939904,0.00014080109,0.000016339263],"category_scores_gemma":[0.00063243386,0.00021818867,0.000032988937,0.33465225,0.00025916114,0.002674218,0.0008743663,0.00019036529,0.000014535536],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000156641,0.000104356775,0.90639347,0.0001546945,0.000016379467,0.0000029545315,0.00010996017,0.0000387522,0.00020540095,0.00002239882,0.000683058,0.09225292],"study_design_scores_gemma":[0.00085966353,0.000045962704,0.98223376,0.00014194837,0.00006718189,8.289336e-7,0.0007048099,0.013223711,0.0000070610085,0.000043220083,0.0023979351,0.00027390145],"about_ca_topic_score_codex":0.00091691595,"about_ca_topic_score_gemma":0.000029527033,"teacher_disagreement_score":0.23179613,"about_ca_system_score_codex":0.000025860367,"about_ca_system_score_gemma":0.000025451873,"threshold_uncertainty_score":0.9073122},"labels":[],"label_agreement":null},{"id":"W2900764028","doi":"10.1007/s11192-018-2959-4","title":"Capturing the economic value of triadic patents","year":2018,"lang":"en","type":"article","venue":"Scientometrics","topic":"Intellectual Property and Patents","field":"Business, Management and Accounting","cited_by":42,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Value (mathematics); Patent analysis; Cover (algebra); Test (biology); European patent office; Patent application; Business; Computer science; Regional science; International trade; Data science; Political science; Sociology; Engineering; Law; Machine learning","score_opus":0.17710943657854503,"score_gpt":0.26109806541647806,"score_spread":0.08398862883793304,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2900764028","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93558496,0.000078730154,0.00025073896,0.0001655923,0.0022979483,0.00017126386,0.000002533049,0.000032268927,0.061415978],"genre_scores_gemma":[0.9974622,0.000006689481,0.000036490233,0.0006642556,0.0008614958,0.0000025997854,0.0000030837634,0.000010464607,0.00095273607],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.99897665,0.000010530993,0.00022959175,0.00019051123,0.0003327577,0.00025996668],"domain_scores_gemma":[0.99933463,0.00006756844,0.00016328573,0.00023561591,0.0001873976,0.000011477412],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0011982005,0.000092625836,0.00011788378,0.0012540227,0.00025108294,0.00018965767,0.00051743526,0.000039253056,0.00055520394],"category_scores_gemma":[0.00059483905,0.000057143807,0.00006796859,0.0033192958,0.00019599467,0.00040628074,0.00021256518,0.000078671066,0.0021861119],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010251898,0.00091360667,0.3594777,0.0010419443,0.000576424,0.000021815535,0.006285347,0.0032578588,0.0065138005,0.17717917,0.20788623,0.2358209],"study_design_scores_gemma":[0.0036382563,0.00025239793,0.11352896,0.00017874865,0.00024193394,0.000008204767,0.00093918096,0.23175842,0.01578495,0.011811246,0.6206933,0.0011643473],"about_ca_topic_score_codex":0.00087401364,"about_ca_topic_score_gemma":0.000021385313,"teacher_disagreement_score":0.4128071,"about_ca_system_score_codex":0.00005203119,"about_ca_system_score_gemma":0.000028368902,"threshold_uncertainty_score":0.9985908},"labels":[],"label_agreement":null},{"id":"W2905418571","doi":"10.1007/s11192-018-2981-6","title":"An approach to conference selection and evaluation: advice to avoid “predatory” conferences","year":2018,"lang":"en","type":"article","venue":"Scientometrics","topic":"Conferences and Exhibitions Management","field":"Social Sciences","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Alberta Hip and Knee Clinic; University of Calgary","funders":"","keywords":"Medical education; Selection (genetic algorithm); Psychology; Public relations; Political science; Computer science; Medicine","score_opus":0.11067442992493984,"score_gpt":0.40623693506175124,"score_spread":0.29556250513681137,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2905418571","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5665724,0.00004219768,0.032724883,0.0018909962,0.0008648196,0.0012024296,0.000017126576,0.00020020833,0.3964849],"genre_scores_gemma":[0.9904509,0.000019142424,0.0074627544,0.0004919512,0.00031998177,0.00007100864,0.000008368763,0.0000053772674,0.0011704703],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99712855,0.00024306955,0.00018911234,0.0005731772,0.0014458998,0.00042021257],"domain_scores_gemma":[0.99733144,0.00005976605,0.000062909516,0.00023320329,0.0017653047,0.00054739096],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0043143854,0.00011775405,0.00013199983,0.0017127714,0.0010695963,0.0009126558,0.00046137633,0.00006788972,0.0006604325],"category_scores_gemma":[0.0008405395,0.00011656473,0.00002070028,0.010024333,0.00024681643,0.00046128972,0.00012054216,0.00007795334,0.00017444801],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004911548,0.00059721776,0.023415262,0.000049476013,0.00003753912,5.4627077e-7,0.07775321,0.00034313378,0.00049342756,0.43891317,0.01714823,0.44119966],"study_design_scores_gemma":[0.0007780374,0.0014813247,0.117036134,0.00006966342,0.000103131795,0.000002330595,0.056243658,0.07278205,0.00036561594,0.0056708558,0.7443962,0.0010710192],"about_ca_topic_score_codex":0.0018655792,"about_ca_topic_score_gemma":0.0047556204,"teacher_disagreement_score":0.72724795,"about_ca_system_score_codex":0.00016590724,"about_ca_system_score_gemma":0.00051217614,"threshold_uncertainty_score":0.8800761},"labels":[],"label_agreement":null},{"id":"W2906809950","doi":"10.1007/s11192-018-2987-0","title":"Publishing in English or another language: An inclusive study of scholar’s language publication preferences in the natural, social and interdisciplinary sciences","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"Discourse Analysis in Language Studies","field":"Arts and Humanities","cited_by":70,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Publishing; Publication; Reputation; English language; Contrast (vision); Sociology; Linguistics; First language; Inclusion (mineral); Social science; Psychology; Computer science; Political science; Mathematics education","score_opus":0.061364195641559476,"score_gpt":0.3791982291499634,"score_spread":0.3178340335084039,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2906809950","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9628813,0.000783492,3.0655164e-7,0.0004208078,0.00029740794,0.00032136845,0.000014079217,0.000019458923,0.035261758],"genre_scores_gemma":[0.9990767,0.000005662615,0.000029891278,0.00013784162,0.00016649636,0.000018522049,0.000008809657,0.0000060826137,0.00054995116],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9980698,0.00025539793,0.00026142132,0.00034700142,0.0008352449,0.00023109748],"domain_scores_gemma":[0.9989723,0.0003349571,0.0001827003,0.00021354482,0.00027789324,0.00001861217],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0037753184,0.00011428675,0.00021153936,0.0027476633,0.00043517936,0.0019618124,0.00081190286,0.000031798132,0.00029280406],"category_scores_gemma":[0.002230651,0.00006612758,0.000026473894,0.004022789,0.00044762692,0.003387478,0.0005769239,0.00029486028,0.0000025549507],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001264097,0.0004000026,0.034777816,0.000020868512,0.000014771914,0.0000038546873,0.95240784,0.0000036453787,0.000018404284,0.004336594,0.00037789386,0.00762569],"study_design_scores_gemma":[0.00035786786,0.00024050039,0.046898972,0.000021157648,0.000014305555,3.55576e-7,0.95180196,0.00011144244,0.000007673232,0.00012287492,0.00031717782,0.0001056987],"about_ca_topic_score_codex":0.00093599834,"about_ca_topic_score_gemma":0.025781736,"teacher_disagreement_score":0.036195423,"about_ca_system_score_codex":0.000046623303,"about_ca_system_score_gemma":0.000053625183,"threshold_uncertainty_score":0.9990742},"labels":[],"label_agreement":null},{"id":"W2909243384","doi":"10.1007/s11192-019-03011-4","title":"Bibliometric and text mining approaches to evaluate landfill design standards","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"Municipal Solid Waste Management","field":"Environmental Science","cited_by":33,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Regina","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Index (typography); Per capita; Government (linguistics); Rank (graph theory); Computer science; Table (database); Statistics; Environmental science; Mathematics; Linguistics; Database; Sociology","score_opus":0.1828976481759,"score_gpt":0.32018661246674013,"score_spread":0.13728896429084012,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2909243384","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9219944,0.00024394477,0.012708171,0.00022545927,0.0003287779,0.0009039491,0.000015088181,0.00004785815,0.06353234],"genre_scores_gemma":[0.98170745,0.000035151428,0.014251538,0.0002473417,0.000017957242,0.000020739848,0.000001826524,0.00002012382,0.0036978954],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9959681,0.00010713317,0.00023198235,0.00064322393,0.0024698493,0.0005796955],"domain_scores_gemma":[0.9987841,0.00023698057,0.000083139304,0.00050061767,0.000031343,0.00036380554],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.006028485,0.00017889573,0.00020679813,0.0359484,0.00013322377,0.00028940465,0.0005628916,0.000052819756,0.0016650935],"category_scores_gemma":[0.0009360415,0.00015332756,0.00003621962,0.23204675,0.00009983101,0.00028569798,0.0011786013,0.00009453299,0.0011814425],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000064053114,0.00018384508,0.35240436,0.00006436628,0.00003798241,0.000008642863,0.0021852802,0.25655672,0.00027981595,0.0002633206,0.02343832,0.3645133],"study_design_scores_gemma":[0.0018459884,0.0010767397,0.50534236,0.000042016698,0.00006363242,0.000009444196,0.0009672646,0.29934078,0.0007175303,0.0002548761,0.18929774,0.0010416576],"about_ca_topic_score_codex":0.00006964748,"about_ca_topic_score_gemma":0.0000068275176,"teacher_disagreement_score":0.36347166,"about_ca_system_score_codex":0.0004156311,"about_ca_system_score_gemma":0.00002040998,"threshold_uncertainty_score":0.99959624},"labels":[],"label_agreement":null},{"id":"W2921565448","doi":"10.1007/s11192-019-03066-3","title":"What characterises funded biomedical research? Evidence from a basic and a clinical domain","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":22,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Health Canada; National Cancer Institute; Consejo Superior de Investigaciones Científicas; National Institutes of Health; Impact Fund; Ministerio de Economía y Competitividad; European Commission; National Research Centre; Nature","keywords":"Basic research; Public funding; Political science; Institution; Public domain; Public institution; Public relations; Public administration; Library science","score_opus":0.8248587450899906,"score_gpt":0.6784602035793619,"score_spread":0.1463985415106287,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2921565448","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9788005,0.0059650526,0.00094551546,0.0060846196,0.006460113,0.0007689867,0.00006892059,0.000054926993,0.0008513621],"genre_scores_gemma":[0.9831362,0.00815594,0.002782578,0.00059531245,0.000411343,0.000017415903,0.000012538805,0.000027241162,0.0048614587],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9524676,0.0023975002,0.0019470884,0.003001601,0.03823586,0.0019503055],"domain_scores_gemma":[0.896863,0.087455474,0.0006376483,0.0031718812,0.008704609,0.003167387],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.18920518,0.0003118426,0.0008577723,0.18693626,0.000572421,0.01713773,0.005766135,0.00061724073,0.0027570107],"category_scores_gemma":[0.26600295,0.00021740037,0.00031138648,0.5423129,0.0019528959,0.0037935795,0.0043972386,0.0018676395,0.0060082274],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010832862,0.00039112847,0.6647955,0.0000126154955,0.000033531473,0.000051240226,0.00051523506,4.8900995e-7,0.0022932847,0.0004091629,0.024813857,0.3065756],"study_design_scores_gemma":[0.0013222301,0.0008867355,0.83853304,0.00016659494,0.000009933419,0.000012114549,0.0026974282,0.0028529086,0.0002023066,0.014578817,0.13831387,0.00042402637],"about_ca_topic_score_codex":0.00026492868,"about_ca_topic_score_gemma":0.000044327102,"teacher_disagreement_score":0.35537666,"about_ca_system_score_codex":0.00031392026,"about_ca_system_score_gemma":0.0007006329,"threshold_uncertainty_score":0.99961317},"labels":[],"label_agreement":null},{"id":"W2921935945","doi":"10.1007/s11192-019-03059-2","title":"The progressive substitution of hazard ratios for relative risks in biomedical research","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Agence Nationale de la Recherche","keywords":"Hazard ratio; Relative risk; Medicine; Hazard; Statistics; Confidence interval; Actuarial science; Econometrics; Demography; Mathematics; Business; Chemistry; Sociology","score_opus":0.9042976211939502,"score_gpt":0.6664328341718819,"score_spread":0.23786478702206826,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2921935945","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96572423,0.002344461,0.022358796,0.0010910966,0.0011327672,0.00321903,0.000022936418,0.000002648494,0.0041040587],"genre_scores_gemma":[0.99270374,0.000022622897,0.0027096528,0.000013105709,0.000039683244,0.000085270054,0.0000044804074,0.0000038853623,0.0044175684],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9721274,0.0064168,0.005097999,0.00085161,0.015029858,0.00047636268],"domain_scores_gemma":[0.95381314,0.034294564,0.0028041552,0.002569978,0.006345374,0.00017281539],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.3580447,0.00011079245,0.0010775084,0.0051720515,0.00024690275,0.00062414986,0.0019474463,0.00009378924,0.0004989521],"category_scores_gemma":[0.30676004,0.00004415738,0.000526331,0.04337439,0.000489419,0.0002909859,0.000182702,0.00023010517,0.001034751],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005876504,0.00029085606,0.7353748,0.00010975271,0.00012529116,0.0000033532544,0.0022987875,0.000318887,0.00033827918,0.15875071,0.03769043,0.06464005],"study_design_scores_gemma":[0.0014286236,0.000645537,0.39487678,0.00014468697,0.000076736425,0.0000047330277,0.0037852537,0.0981456,0.00039391255,0.07297694,0.42722568,0.00029553534],"about_ca_topic_score_codex":0.000009617283,"about_ca_topic_score_gemma":0.000021653292,"teacher_disagreement_score":0.38953525,"about_ca_system_score_codex":0.00010357321,"about_ca_system_score_gemma":0.0002628099,"threshold_uncertainty_score":0.99974304},"labels":[],"label_agreement":null},{"id":"W2945259118","doi":"10.1007/s11192-019-03120-0","title":"Identifying emerging scholars: seeing through the crystal ball of scholarship selection committees","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec en Outaouais","funders":"","keywords":"Scholarship; Percentage point; Competition (biology); Psychology; Point (geometry); Selection (genetic algorithm); Medical education; Political science; Medicine; Computer science; Statistics; Mathematics; Law","score_opus":0.5381788305996781,"score_gpt":0.5673536692852789,"score_spread":0.02917483868560078,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2945259118","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96795917,0.002327723,0.014000928,0.00056792045,0.002319686,0.0005516954,0.000023706298,0.00006038281,0.012188766],"genre_scores_gemma":[0.991471,0.00017215784,0.004858774,0.00019968733,0.00009989599,0.000007697975,0.0000028956586,0.000025269646,0.0031626003],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.97251636,0.00072345487,0.0013944014,0.0012336134,0.022901423,0.0012307302],"domain_scores_gemma":[0.9826747,0.0078114374,0.00089006504,0.0015239262,0.006724983,0.00037489427],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.06449754,0.00028014425,0.0005306949,0.09870425,0.000914438,0.0063294284,0.005844918,0.0002015694,0.0011437703],"category_scores_gemma":[0.08997458,0.0001815673,0.00037719271,0.5707337,0.00035795476,0.0039975834,0.0018612507,0.0010440482,0.00089122215],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000020713947,0.00017432368,0.93383193,0.00003397526,0.000040779265,0.0000034069642,0.0010879834,0.00078883226,0.0168165,0.0037862062,0.002913797,0.040501565],"study_design_scores_gemma":[0.0015235633,0.0004461774,0.80859876,0.00010404251,0.0000385526,0.000041077365,0.005807728,0.014464337,0.022459041,0.027279511,0.11845284,0.00078437064],"about_ca_topic_score_codex":0.00013371467,"about_ca_topic_score_gemma":0.00001857007,"teacher_disagreement_score":0.4720295,"about_ca_system_score_codex":0.0003159849,"about_ca_system_score_gemma":0.00032035162,"threshold_uncertainty_score":0.9998867},"labels":[],"label_agreement":null},{"id":"W2954205856","doi":"10.1007/s11192-019-03157-1","title":"How collaboration type, publication place, funding and author’s role affect citations received by publications from Africa: A bibliometric study of LIS research from 1996 to 2015","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":77,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Citation; Incentive; Library science; Political science; Citation impact; Bibliometrics; Public relations; Economics; Computer science","score_opus":0.5915490014438257,"score_gpt":0.5835091969205798,"score_spread":0.00803980452324593,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2954205856","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9819898,0.0029477142,0.0010413174,0.00606135,0.0010003165,0.002474082,0.0011330147,0.00007582845,0.0032765476],"genre_scores_gemma":[0.9829507,0.00013989785,0.001991683,0.000045371668,0.00009510496,0.00013453883,0.00022076153,0.000032923814,0.014389017],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.97177213,0.0014446697,0.0012374086,0.0022560048,0.02205972,0.0012300529],"domain_scores_gemma":[0.93256354,0.025812984,0.0009091271,0.0026024175,0.036434043,0.0016778708],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.04756975,0.00031015082,0.00066982117,0.7025377,0.00078468776,0.015614636,0.0037983586,0.00029417855,0.0004759513],"category_scores_gemma":[0.26319402,0.00026311405,0.00008684335,0.9555584,0.00023542183,0.003501321,0.0016792578,0.0006130276,0.0006081595],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007055468,0.0011750667,0.61699533,0.0000067399587,0.000061611594,7.375871e-7,0.003891633,0.00005384746,0.014558907,0.0006094109,0.33280566,0.029770462],"study_design_scores_gemma":[0.0013994131,0.0015527327,0.8760884,0.000014699933,0.000027934573,5.1814453e-7,0.022885477,0.0046703247,0.0007849813,0.0026503012,0.08950081,0.00042440085],"about_ca_topic_score_codex":0.0019627712,"about_ca_topic_score_gemma":0.0003318974,"teacher_disagreement_score":0.25909305,"about_ca_system_score_codex":0.000587261,"about_ca_system_score_gemma":0.00067902525,"threshold_uncertainty_score":0.9999821},"labels":[],"label_agreement":null},{"id":"W2955327859","doi":"10.1007/s11192-019-03111-1","title":"On the development of China’s leadership in international collaborations","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"International Student and Expatriate Challenges","field":"Social Sciences","cited_by":24,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"China Scholarship Council","keywords":"China; Political science; Regional science; Sociology; Law","score_opus":0.21961104703660078,"score_gpt":0.3793214619532546,"score_spread":0.15971041491665383,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2955327859","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8899283,0.000035802088,0.000028247632,0.0048489193,0.00082491094,0.0001566295,0.0000027767865,0.000007524084,0.10416691],"genre_scores_gemma":[0.9963259,0.000020115487,0.00015043205,0.00009085015,0.000039787257,0.0000038550734,0.0000026299474,0.0000022226598,0.003364173],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.9984939,0.000047843216,0.00015625864,0.00010960071,0.0010426475,0.00014972257],"domain_scores_gemma":[0.99938697,0.000278403,0.00006661994,0.0000573901,0.00018508208,0.000025548776],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0019467432,0.000040842362,0.000055164284,0.00087377464,0.0001303862,0.000057825535,0.0004539696,0.00003229279,0.0004919447],"category_scores_gemma":[0.001117174,0.000030340001,0.000021801232,0.0033192488,0.00007015052,0.00009306869,0.000039688133,0.00006930416,0.00017770963],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000018524157,0.00020667737,0.032837454,0.000006569572,0.000020371843,2.5706478e-7,0.33978948,0.0001635887,0.0001410098,0.62259406,0.00069085526,0.00353117],"study_design_scores_gemma":[0.0010747318,0.000090156864,0.3737316,0.00010932279,0.0000043794726,1.10654064e-7,0.55091834,0.00036545968,0.0016871761,0.0068223826,0.06487018,0.0003261451],"about_ca_topic_score_codex":0.00012073132,"about_ca_topic_score_gemma":0.0011690138,"teacher_disagreement_score":0.61577165,"about_ca_system_score_codex":0.00021036991,"about_ca_system_score_gemma":0.00015179341,"threshold_uncertainty_score":0.53864497},"labels":[],"label_agreement":null},{"id":"W2955873301","doi":"10.1007/s11192-019-03146-4","title":"The citation advantage of foreign language references for Chinese social science papers","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":24,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Office for Philosophy and Social Sciences; China Scholarship Council","keywords":"Citation; Internationalization; Foreign language; Citation index; Impact factor; Scientific communication; Sociology of scientific knowledge; Social science; Consumption (sociology); Artifact (error); Scientometrics; Diversity (politics); Sociology; Computer science; Library science; Political science; Law; Business","score_opus":0.34542183294758166,"score_gpt":0.5814282516665357,"score_spread":0.236006418718954,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2955873301","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9018667,0.00061490666,0.0013327581,0.00024741524,0.0014756516,0.0006967943,0.00009095738,0.000022638364,0.09365218],"genre_scores_gemma":[0.9944675,0.00007381278,0.0010178257,0.000058068163,0.00007487888,0.000019324729,0.000005186096,0.000010364544,0.0042730127],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.97952145,0.00012466194,0.00082433457,0.00090268505,0.017645406,0.000981442],"domain_scores_gemma":[0.9799633,0.011449912,0.0006208788,0.0008780792,0.0067739016,0.00031390542],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.062178716,0.00017452335,0.000344946,0.08917785,0.0011526555,0.0027499814,0.0052497867,0.00009178874,0.00016474965],"category_scores_gemma":[0.11688691,0.00009311091,0.00023739002,0.5141261,0.0011355059,0.0010036702,0.00073695445,0.00019063121,0.00012322004],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000104797145,0.00017411394,0.38074687,0.000032231805,0.000018574585,0.0000012871133,0.0019325325,0.000033459808,0.0406542,0.064988695,0.0021479782,0.5091652],"study_design_scores_gemma":[0.0017745101,0.0008786184,0.8985406,0.000012157224,0.000013497525,0.0000036860718,0.019622473,0.010949568,0.007718425,0.025862094,0.03410206,0.00052229466],"about_ca_topic_score_codex":0.00001615571,"about_ca_topic_score_gemma":0.000007617001,"teacher_disagreement_score":0.5177937,"about_ca_system_score_codex":0.00017799286,"about_ca_system_score_gemma":0.00050655234,"threshold_uncertainty_score":0.99828523},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2962062541","doi":"10.1007/s11192-019-03166-0","title":"Visualizing music similarity: clustering and mapping 500 classical music composers","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"Music and Audio Processing","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Multidimensional scaling; Similarity (geometry); Cluster analysis; Dendrogram; Distance matrices in phylogeny; Canonical correlation; Computer science; Artificial intelligence; Mathematics; Pattern recognition (psychology); Combinatorics; Machine learning","score_opus":0.06368608730831339,"score_gpt":0.28835309034117484,"score_spread":0.22466700303286147,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2962062541","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4944817,0.0003590097,0.4944955,0.00083550427,0.0017007972,0.00021979056,0.0000014177732,0.00023796094,0.0076682707],"genre_scores_gemma":[0.95733243,0.000012395721,0.039442874,0.0025924346,0.00010766219,0.0000030353626,0.0000016908956,0.000014945586,0.000492527],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99732494,0.000054911787,0.00030474784,0.0007729595,0.0009253955,0.0006170158],"domain_scores_gemma":[0.99874717,0.00021871008,0.0001485863,0.00049339776,0.00013212112,0.0002600442],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0012832872,0.00020248652,0.00027986098,0.0020356986,0.00033723275,0.00105122,0.00088612625,0.00009595259,0.00004394568],"category_scores_gemma":[0.00018933999,0.00019360222,0.00006501888,0.009540397,0.000119694836,0.0010294217,0.001161227,0.00027078186,0.000094905925],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014260591,0.0003145605,0.0543004,0.00095736206,0.00007239966,0.00009321661,0.014695221,0.0011155158,0.033270717,0.022599442,0.008990113,0.86357677],"study_design_scores_gemma":[0.0011713716,0.00013957583,0.03223128,0.00023634902,0.000013288134,0.00006521301,0.0004559021,0.899071,0.0009887968,0.0010391817,0.06374192,0.00084611407],"about_ca_topic_score_codex":0.000009564011,"about_ca_topic_score_gemma":0.00000467852,"teacher_disagreement_score":0.8979555,"about_ca_system_score_codex":0.000102694095,"about_ca_system_score_gemma":0.0000946876,"threshold_uncertainty_score":0.9999858},"labels":[],"label_agreement":null},{"id":"W2962098473","doi":"10.1007/s11192-019-03181-1","title":"Territorial innovation models: to be or not to be, that’s the question","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"Regional Development and Policy","field":"Social Sciences","cited_by":26,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"HEC Montréal","funders":"H2020 Societal Challenges; Eusko Jaurlaritza","keywords":"Variety (cybernetics); Field (mathematics); Regional science; Divergence (linguistics); Urban agglomeration; Computer science; Economic geography; Data science; Sociology; Knowledge management; Geography; Linguistics; Artificial intelligence; Mathematics","score_opus":0.2503429839231957,"score_gpt":0.42876411249566576,"score_spread":0.17842112857247006,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2962098473","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.86359674,0.0000091502225,0.0012526457,0.076262034,0.010936247,0.00096623425,0.000018288174,0.00008805752,0.046870634],"genre_scores_gemma":[0.94484407,0.000021079479,0.0011847763,0.010317882,0.0012709767,0.000021098498,0.00001115093,0.000008882462,0.042320065],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9974417,0.00008819966,0.00017356605,0.0002455118,0.0017032116,0.00034778233],"domain_scores_gemma":[0.9989527,0.00024614722,0.00006870022,0.00018848409,0.00040226654,0.00014171012],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.003049096,0.00008587418,0.000095212716,0.002136512,0.0003914378,0.00031657002,0.00053621735,0.000079981444,0.0000911721],"category_scores_gemma":[0.0015322131,0.00005826,0.000026896736,0.022662409,0.00006327288,0.0004014378,0.00010950188,0.000086067,0.00023132339],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012439993,0.000056387897,0.00511974,0.000008123296,0.000011060515,9.1947595e-7,0.060377024,0.0004283042,0.0002960271,0.69628775,0.2182462,0.019044051],"study_design_scores_gemma":[0.00014687648,0.000063106236,0.010569967,0.000009571178,0.0000029699904,2.25696e-7,0.0014058865,0.00012304676,0.00018664177,0.0017570978,0.98558646,0.00014813885],"about_ca_topic_score_codex":0.0018668376,"about_ca_topic_score_gemma":0.00088844757,"teacher_disagreement_score":0.76734024,"about_ca_system_score_codex":0.00027588816,"about_ca_system_score_gemma":0.00043443398,"threshold_uncertainty_score":0.99811137},"labels":[],"label_agreement":null},{"id":"W2971101863","doi":"10.1007/s11192-019-03215-8","title":"Cooperation, scale-invariance and complex innovation systems: a generalization","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"Innovation and Knowledge Management","field":"Business, Management and Accounting","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Generalization; Scale (ratio); Scale invariance; Mathematical economics; Mathematics; Computer science; Statistics; Geography; Cartography; Mathematical analysis","score_opus":0.03896026622934054,"score_gpt":0.2602012405683491,"score_spread":0.22124097433900855,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2971101863","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5273571,0.00014914191,0.034213327,0.0010716247,0.004629312,0.0012640674,0.0000037036425,0.00028983862,0.43102187],"genre_scores_gemma":[0.9912259,0.0000050477956,0.00042631713,0.0025272998,0.00038334928,0.000013003191,0.00013118559,0.000012870076,0.005275024],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987905,0.000008512131,0.00031413633,0.00028266755,0.00041861806,0.00018553615],"domain_scores_gemma":[0.99825656,0.000012941299,0.00019766319,0.00018784482,0.001337357,0.000007650194],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0010462706,0.00010712009,0.00012091215,0.003540655,0.00018476688,0.0009440291,0.00014312341,0.000041711806,0.00012221519],"category_scores_gemma":[0.00021870028,0.00010264906,0.000011264917,0.023049098,0.000033477267,0.0008616677,0.0001338184,0.000050465416,0.0006625837],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000003518695,0.000031994357,0.04329791,0.00016318588,0.000005201914,2.7374236e-7,0.000028546392,0.00022306676,0.0008739779,0.94648594,0.0073126974,0.0015736651],"study_design_scores_gemma":[0.0012077633,0.000017793092,0.046739355,0.000048019152,0.000020998768,0.0000015622048,0.00034734898,0.3271549,0.000041859275,0.0010794132,0.6229269,0.0004140869],"about_ca_topic_score_codex":0.00002815178,"about_ca_topic_score_gemma":0.0000075827857,"teacher_disagreement_score":0.94540656,"about_ca_system_score_codex":0.000050654624,"about_ca_system_score_gemma":0.00001586177,"threshold_uncertainty_score":0.9977165},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2983983302","doi":"10.1007/s11192-019-03280-z","title":"Deep and narrow impact: introducing location filtered citation counting","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":21,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Bibliometrics; Citation; Ranking (information retrieval); Computer science; Information retrieval; Biomedicine; Data science; Citation analysis; Rank (graph theory); Scientometrics; Data mining; Mathematics; Library science; Bioinformatics; Biology","score_opus":0.35873592871164445,"score_gpt":0.5378950683652745,"score_spread":0.17915913965363,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2983983302","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97313267,0.0017678342,0.018835401,0.00041541195,0.0014547448,0.00041082414,0.000010813049,0.000044356133,0.003927924],"genre_scores_gemma":[0.99495274,0.000052429314,0.0026825867,0.00011874248,0.0001152168,0.0000058015557,0.00001169599,0.00001579531,0.0020450107],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9842342,0.0001961241,0.00078924804,0.0012485418,0.012654933,0.0008769259],"domain_scores_gemma":[0.98552126,0.0048628603,0.0004822576,0.0010471602,0.0074839443,0.00060251873],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.040652663,0.000207943,0.00035333412,0.1596791,0.00036914318,0.0056213946,0.0016153719,0.0001358539,0.0008115205],"category_scores_gemma":[0.10507772,0.00015373042,0.00011269413,0.5284157,0.00016888215,0.001811574,0.00064118876,0.00030125337,0.0011157828],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002043699,0.000085517815,0.7539529,0.000023659619,0.000012040391,0.0000027864485,0.0007535608,0.004075868,0.0037066701,0.0007453745,0.004320845,0.23230031],"study_design_scores_gemma":[0.00069819,0.00024723847,0.7247403,0.000011522019,0.000006542648,0.000013669928,0.00077583024,0.26633137,0.0005827675,0.0024133513,0.0038763005,0.00030287562],"about_ca_topic_score_codex":0.000103794446,"about_ca_topic_score_gemma":0.00001014519,"teacher_disagreement_score":0.3687366,"about_ca_system_score_codex":0.00035847008,"about_ca_system_score_gemma":0.00024091348,"threshold_uncertainty_score":0.999662},"labels":[],"label_agreement":null},{"id":"W2986816625","doi":"10.1007/s11192-019-03285-8","title":"The national system of researchers in Mexico: implications of publication incentives for researchers in social sciences","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"Science, Technology, and Education in Latin America","field":"Social Sciences","cited_by":20,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Université du Québec à Montréal","funders":"Consejo Nacional de Ciencia y Tecnología","keywords":"Incentive; Promotion (chess); SNi; Revocation; Political science; Sociology; Engineering; Economics; Law; Chemistry","score_opus":0.3660182259990773,"score_gpt":0.5331802589625843,"score_spread":0.16716203296350696,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2986816625","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7515376,0.00035475555,0.00035794833,0.035104934,0.00068740017,0.001973698,0.0000483727,0.000052248724,0.20988305],"genre_scores_gemma":[0.99831176,0.0000593767,0.00095791183,0.000031816904,0.000030696883,0.00010366473,0.0000025702732,0.0000032109092,0.0004989891],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9968878,0.00027848504,0.00037394706,0.0003029628,0.0016509026,0.00050590915],"domain_scores_gemma":[0.99574393,0.0023533013,0.0002998389,0.0001498462,0.0013923392,0.000060763603],"candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.021252371,0.000053977117,0.00013138658,0.0037229126,0.0007373274,0.00012717505,0.0013171823,0.00008916166,0.000020294525],"category_scores_gemma":[0.016369717,0.000044841614,0.000046731202,0.037386656,0.0034948688,0.0003790136,0.00010082518,0.00013758149,0.0000059923586],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010625704,0.00010102023,0.3793327,0.000025492049,0.0000034659859,8.078279e-9,0.006991916,0.000049884366,0.00036967103,0.6034181,0.0026782127,0.0070189144],"study_design_scores_gemma":[0.0006145745,0.00014101536,0.81205875,0.00004978304,0.0000027103836,2.0854695e-7,0.12980972,0.0019397804,0.00065529323,0.018448452,0.036105484,0.0001742262],"about_ca_topic_score_codex":0.00047745154,"about_ca_topic_score_gemma":0.00022879629,"teacher_disagreement_score":0.58496964,"about_ca_system_score_codex":0.0007141673,"about_ca_system_score_gemma":0.0016424361,"threshold_uncertainty_score":0.99921703},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2987668259","doi":"10.1007/s11192-019-03291-w","title":"Evaluating medical conferences: the emerging need for a quality metric","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":11,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Alberta Hip and Knee Clinic; University of Calgary","funders":"","keywords":"Impact factor; Scope (computer science); Quality (philosophy); Metric (unit); Publishing; Observational study; Computer science; Library science; Ranking (information retrieval); Political science; Medicine; Business; Information retrieval","score_opus":0.8632070423881478,"score_gpt":0.7154768984215607,"score_spread":0.14773014396658712,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2987668259","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9001105,0.0022637183,0.061688304,0.0062394417,0.0057880296,0.0021358018,0.00008964359,0.000100133155,0.021584466],"genre_scores_gemma":[0.9903953,0.00010061093,0.0031072923,0.00080921926,0.00020031589,0.000067092806,0.000009476813,0.000020811,0.005289852],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.94184977,0.0010309861,0.001922253,0.0017154092,0.051814627,0.0016669576],"domain_scores_gemma":[0.85030746,0.11738067,0.001814633,0.004292361,0.023896141,0.0023087128],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.25881523,0.00029724822,0.000694419,0.1489679,0.0008757854,0.005164391,0.008702159,0.00026744057,0.0037286726],"category_scores_gemma":[0.5399934,0.00016397025,0.0004924886,0.63241637,0.00042524864,0.0007880301,0.0017442633,0.0006269276,0.0013563843],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000057222805,0.0002451529,0.288818,0.000030487234,0.000047020792,0.0000025957752,0.00042357654,0.0004342368,0.00041838916,0.023080721,0.013487978,0.6729546],"study_design_scores_gemma":[0.004200752,0.0011559937,0.2884158,0.000039336097,0.000038644233,0.000016942202,0.0052769696,0.4901171,0.00079091394,0.03399798,0.17495292,0.0009966604],"about_ca_topic_score_codex":0.00022195427,"about_ca_topic_score_gemma":0.000024696876,"teacher_disagreement_score":0.67195797,"about_ca_system_score_codex":0.0002707261,"about_ca_system_score_gemma":0.0017491643,"threshold_uncertainty_score":0.9994212},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2989484692","doi":"10.1007/s11192-019-03301-x","title":"Do articles in open access journals have more frequent altmetric activity than articles in subscription-based journals? An investigation of the research output of Finnish universities","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":105,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Turun Yliopisto","keywords":"Altmetrics; Discipline; Scientometrics; Bibliometrics; Social media; Library science; Computer science; Sociology; Social science; World Wide Web","score_opus":0.838152280754456,"score_gpt":0.6455667087672443,"score_spread":0.1925855719872117,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2989484692","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9955313,0.0008126932,0.00020764513,0.0013037829,0.0004433122,0.0009106401,0.00006028094,0.000006311641,0.0007239901],"genre_scores_gemma":[0.99892265,0.00014185451,0.00034326222,0.00006284908,0.000023143297,0.000008554824,0.0000015016543,0.000013897187,0.00048227905],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.97376555,0.0024317822,0.0014986485,0.0010111089,0.020212078,0.0010808457],"domain_scores_gemma":[0.9783849,0.008512352,0.0012215439,0.0017518249,0.009483315,0.00064602785],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.108231656,0.0002111957,0.0006495731,0.26909086,0.0003236766,0.007054362,0.011431104,0.00019942832,0.00043864132],"category_scores_gemma":[0.043196917,0.0001450322,0.00019400302,0.61527854,0.00093990914,0.0059272307,0.0033438117,0.00092968275,0.000024694027],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000714087,0.0004190861,0.9661016,0.000020500667,0.000007404975,0.0000059402514,0.0014834232,0.003067941,0.007809614,0.00016598376,0.0003029797,0.020544104],"study_design_scores_gemma":[0.0010863772,0.00021378273,0.93484783,0.00011935488,0.0000043483183,0.0000016582044,0.008116978,0.019445434,0.029963762,0.005993111,0.00004977768,0.00015761389],"about_ca_topic_score_codex":0.004475062,"about_ca_topic_score_gemma":0.0009897391,"teacher_disagreement_score":0.34618768,"about_ca_system_score_codex":0.00096834963,"about_ca_system_score_gemma":0.0019846612,"threshold_uncertainty_score":0.9939764},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2998363348","doi":"10.1007/s11192-019-03339-x","title":"The role of Web of Science publications in China’s tenure system","year":2020,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":72,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"China; Web of science; Political science; World Wide Web; Library science; Computer science; Sociology; MEDLINE; Law","score_opus":0.40245736986760666,"score_gpt":0.5294233401406908,"score_spread":0.12696597027308415,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2998363348","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.91279554,0.0030526756,0.0008228233,0.003567369,0.0007272805,0.00056009553,0.00010113873,0.000040944054,0.078332156],"genre_scores_gemma":[0.999057,0.00009355951,0.0005411955,0.000021617127,0.000028471839,0.000009041754,8.665044e-7,0.0000069506223,0.0002412964],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.97615623,0.0001933559,0.0013749254,0.0008745887,0.020557126,0.00084375014],"domain_scores_gemma":[0.9851052,0.0038702176,0.0007967435,0.0013344686,0.0081375325,0.0007558106],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.052034017,0.00013782273,0.00038788063,0.14968626,0.000492373,0.0016090189,0.007038584,0.00008051716,0.000050920495],"category_scores_gemma":[0.17766507,0.00007896608,0.00015016088,0.8136093,0.0021330682,0.00086809427,0.001515258,0.0002997782,0.000070048955],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002422099,0.00023346185,0.64912796,0.000029106062,0.000008872534,0.000002185553,0.001111199,0.00022229193,0.019175202,0.15922688,0.001981582,0.16885705],"study_design_scores_gemma":[0.00052837376,0.00024458944,0.80620056,0.0000179467,0.0000055272567,0.0000046891205,0.0056019477,0.1132713,0.012683889,0.0019597968,0.059282925,0.00019845975],"about_ca_topic_score_codex":0.0001027676,"about_ca_topic_score_gemma":0.000015446512,"teacher_disagreement_score":0.6639231,"about_ca_system_score_codex":0.00024868004,"about_ca_system_score_gemma":0.0015980299,"threshold_uncertainty_score":0.9994274},"labels":[],"label_agreement":null},{"id":"W3000674115","doi":"10.1007/s11192-019-03340-4","title":"Mapping the dynamics of research networks in ecology and evolution using co-citation analysis (1975–2014)","year":2020,"lang":"en","type":"article","venue":"Scientometrics","topic":"Plant and animal studies","field":"Agricultural and Biological Sciences","cited_by":20,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Ecology; Population ecology; Evolutionary ecology; Citation; Population; Taxon; Field (mathematics); Sociology; Biology; Geography; Computer science; Demography","score_opus":0.2390556245279027,"score_gpt":0.34205590808683334,"score_spread":0.10300028355893065,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3000674115","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99804157,0.00046705315,0.0002626654,0.00096092426,0.0000220577,0.00008427484,0.00001649143,0.0000065697673,0.0001383789],"genre_scores_gemma":[0.9997099,0.00013113719,0.000055214496,0.000035275483,0.000043056756,0.0000012713213,0.000012537821,2.136006e-7,0.000011413162],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99905384,0.00011347208,0.00014162117,0.00017104793,0.0003157811,0.00020426784],"domain_scores_gemma":[0.99914986,0.000601481,0.00006244123,0.00001325778,0.00013015066,0.000042802632],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0015244168,0.000044499586,0.0001234257,0.00039075944,0.00023447727,0.000041237847,0.00014572649,0.00004525035,0.000008407884],"category_scores_gemma":[0.000474475,0.000016613658,0.000032016545,0.02322321,0.0001617154,0.000057873647,0.00009908445,0.00011330137,0.0000015359875],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000006571094,0.000011598729,0.99301225,0.0000036045997,0.000013696099,6.7068146e-7,0.00017044309,0.0008935241,0.0033713672,0.0003695289,0.00005129964,0.0020954686],"study_design_scores_gemma":[0.000028684577,0.00005118365,0.7748945,0.0000024889805,0.000007683791,2.5922358e-7,0.0010249049,0.22382647,0.0000074818563,0.00007095381,0.000056020468,0.00002936871],"about_ca_topic_score_codex":0.00032342976,"about_ca_topic_score_gemma":0.0033438916,"teacher_disagreement_score":0.22293295,"about_ca_system_score_codex":0.000053654032,"about_ca_system_score_gemma":0.0000054300012,"threshold_uncertainty_score":0.9975386},"labels":[],"label_agreement":null},{"id":"W3014941759","doi":"10.1007/s11192-020-03434-4","title":"What motivates ‘free’ revealing? Measuring outbound non-pecuniary openness, innovation types and expectations of future profit growth","year":2020,"lang":"en","type":"article","venue":"Scientometrics","topic":"Innovation and Knowledge Management","field":"Business, Management and Accounting","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"Centre of Excellence in Plant Energy Biology, Australian Research Council","keywords":"Openness to experience; Business; Variety (cybernetics); Profit (economics); Marketing; Open innovation; Product innovation; Industrial organization; Value (mathematics); Product (mathematics); Microeconomics; Economics; Computer science","score_opus":0.04621886148311979,"score_gpt":0.2533953459324179,"score_spread":0.20717648444929812,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3014941759","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9010448,0.001216822,0.0019098992,0.015486636,0.0033058184,0.0011177664,0.0000090607,0.00022028416,0.07568895],"genre_scores_gemma":[0.9966221,0.000046194247,0.0010045303,0.001391032,0.0006613485,0.000013796982,0.00004339258,0.000019406363,0.00019820257],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9984846,0.000008351245,0.00040393308,0.00033436943,0.00055737235,0.00021138122],"domain_scores_gemma":[0.99791735,0.000030474146,0.00032899587,0.00019648294,0.0015087577,0.000017930375],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.000748238,0.0001493451,0.00019071231,0.002824287,0.00022332923,0.0007939317,0.00036744002,0.000051987045,0.000032858883],"category_scores_gemma":[0.0012070693,0.00014072644,0.000028607295,0.023673708,0.00007046719,0.0020649033,0.00043027382,0.000113732414,0.00004876711],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006716123,0.00016469385,0.09314476,0.0022429086,0.00010148831,0.0000073807482,0.0031676372,0.0000417978,0.002997301,0.8674596,0.010096752,0.020508507],"study_design_scores_gemma":[0.017100148,0.0005240518,0.429652,0.0022096927,0.0009026101,0.000008685956,0.21196862,0.081157915,0.028843703,0.06320209,0.15813978,0.006290718],"about_ca_topic_score_codex":0.000012100878,"about_ca_topic_score_gemma":0.0000061112246,"teacher_disagreement_score":0.8042575,"about_ca_system_score_codex":0.000029971967,"about_ca_system_score_gemma":0.000025538757,"threshold_uncertainty_score":0.9970786},"labels":[],"label_agreement":null},{"id":"W3016661053","doi":"10.1007/s11192-020-03462-0","title":"Telescopic and panoramic views of library and information science research 2011–2018: a comparison of four weighting schemes for author co-citation analysis","year":2020,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":10,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Citation; Field (mathematics); Context (archaeology); Weighting; Information retrieval; Data science; Zoom; Citation analysis; Set (abstract data type); Citation impact; Co-citation; Rank (graph theory); World Wide Web; Mathematics; History","score_opus":0.7748089364909767,"score_gpt":0.6271083223531914,"score_spread":0.14770061413778535,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3016661053","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9662152,0.0015286604,0.029025653,0.0011532757,0.00014467938,0.00066496735,0.00013270107,0.000018176517,0.0011167134],"genre_scores_gemma":[0.98202115,0.00028347675,0.017491654,0.000055091306,0.000025044677,0.000011165055,0.000013667436,0.0000064441547,0.00009232618],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9846341,0.00026173753,0.0014539373,0.00077130715,0.012197578,0.0006813453],"domain_scores_gemma":[0.9849724,0.007680626,0.00090838154,0.0005375562,0.0051824884,0.000718504],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.037481986,0.0001479669,0.000641403,0.2297348,0.00055749627,0.002475807,0.0019164417,0.00011470327,0.00006722835],"category_scores_gemma":[0.07969864,0.00011206927,0.00012736057,0.5852225,0.0014210263,0.0068551707,0.0010478362,0.0002689579,0.000018987314],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000590702,0.000093784074,0.7657769,0.00016751145,0.00004017806,3.185602e-7,0.002180106,0.000059356393,0.0076214527,0.0056959307,0.0024081531,0.21589728],"study_design_scores_gemma":[0.000993167,0.0011231806,0.57573444,0.000026610289,0.00006947147,0.0000010738822,0.004363029,0.34944925,0.034261897,0.0018709338,0.03181486,0.00029211232],"about_ca_topic_score_codex":0.000014878793,"about_ca_topic_score_gemma":0.0000012590331,"teacher_disagreement_score":0.3554877,"about_ca_system_score_codex":0.000047820555,"about_ca_system_score_gemma":0.0005050422,"threshold_uncertainty_score":0.9985597},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W3033805182","doi":"10.1007/s11192-020-03537-y","title":"Gatekeepers in knowledge transfer between science and technology: an exploratory study in the area of gene editing","year":2020,"lang":"en","type":"article","venue":"Scientometrics","topic":"CRISPR and Genetic Engineering","field":"Biochemistry, Genetics and Molecular Biology","cited_by":24,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"National Natural Science Foundation of China","keywords":"Betweenness centrality; Knowledge flow; Social network analysis; Knowledge transfer; Computer science; Data science; Knowledge management; World Wide Web; Centrality; Social media","score_opus":0.042753495327502666,"score_gpt":0.33993385302328943,"score_spread":0.29718035769578677,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3033805182","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99789506,0.00071483303,0.00095330086,0.000102102036,0.000045266064,0.000185183,0.000004790701,0.0000055118935,0.00009395548],"genre_scores_gemma":[0.9997543,0.000020555628,0.00013657128,0.000025994825,0.000045157845,0.000008909782,0.0000023492007,0.0000052153864,9.2201543e-7],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99905384,0.00002747325,0.00015570378,0.00030212014,0.00026674752,0.0001941201],"domain_scores_gemma":[0.9996057,0.000015507389,0.000013909512,0.00018011156,0.000118960896,0.00006579202],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014569345,0.000072489864,0.00010215293,0.0010250715,0.00004263985,0.000022708065,0.0003788413,0.00004966682,5.025621e-7],"category_scores_gemma":[0.00057272287,0.00006049585,0.000010349742,0.009155121,0.00020761573,0.000008479092,0.000082615894,0.00009721177,2.9561454e-7],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000046156847,0.00013048958,0.47856554,0.000019279983,0.000004500421,0.000003288938,0.0086007295,0.00017938425,0.5029861,0.000011852299,0.000012154742,0.009482079],"study_design_scores_gemma":[0.0013220495,0.0012755768,0.32548314,0.000015309224,0.000017344553,0.0000027969986,0.033009402,0.0012873005,0.6367018,0.000011782927,0.0006045575,0.0002688837],"about_ca_topic_score_codex":0.0000027557724,"about_ca_topic_score_gemma":0.000045710585,"teacher_disagreement_score":0.1530824,"about_ca_system_score_codex":0.00001242543,"about_ca_system_score_gemma":0.000080116064,"threshold_uncertainty_score":0.43987322},"labels":[],"label_agreement":null},{"id":"W3037469314","doi":"10.1007/s11192-020-03590-7","title":"Coronavirus mapping in scientific publications: When science advances rapidly and collectively, is access to this knowledge open to society?","year":2020,"lang":"en","type":"article","venue":"Scientometrics","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":101,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Conselho Nacional de Desenvolvimento Científico e Tecnológico; Fundação Oswaldo Cruz; Comunidad de Madrid","keywords":"Coronavirus disease 2019 (COVID-19); Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); 2019-20 coronavirus outbreak; Coronavirus; Sociology of scientific knowledge; Open science; Data science; Library science; Political science; Computer science; Sociology; Social science; Virology; Medicine; Physics; Infectious disease (medical specialty)","score_opus":0.45005805485460654,"score_gpt":0.5235256239953888,"score_spread":0.07346756914078229,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3037469314","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6750314,0.0062668365,0.02588628,0.15146057,0.0046806876,0.006189587,0.00041772166,0.00028562293,0.1297813],"genre_scores_gemma":[0.93143624,0.00010881951,0.024573643,0.024912074,0.00018992799,0.00024842442,0.0000074937166,0.00003537053,0.01848801],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9908495,0.0001685832,0.0009522985,0.002610686,0.004456911,0.0009620174],"domain_scores_gemma":[0.992964,0.0015493155,0.00037048187,0.0009268783,0.0026386417,0.0015506545],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.026427127,0.00025571216,0.0004927849,0.009032258,0.0018543641,0.04255295,0.018140243,0.00012165242,0.0005081963],"category_scores_gemma":[0.056596834,0.00021494048,0.00007159627,0.1842111,0.0009560685,0.017907722,0.011699261,0.0004521765,0.0005798517],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019245914,0.0000959028,0.11098177,0.000012314876,0.0000056571316,9.0288165e-7,0.037639916,0.00006286946,0.00090889074,0.0007961236,0.6457933,0.20368308],"study_design_scores_gemma":[0.00048372927,0.0000671197,0.08170581,0.000040717376,0.000003946322,0.0000021098592,0.0033814867,0.0036227542,0.0011887293,0.0038858932,0.9052087,0.00040899043],"about_ca_topic_score_codex":0.00017030355,"about_ca_topic_score_gemma":0.000098440556,"teacher_disagreement_score":0.2594154,"about_ca_system_score_codex":0.00049209263,"about_ca_system_score_gemma":0.0018553223,"threshold_uncertainty_score":0.9994451},"labels":[],"label_agreement":null},{"id":"W3045327594","doi":"10.1007/s11192-020-03744-7","title":"Understanding the temporal evolution of COVID-19 research through machine learning and natural language processing","year":2020,"lang":"en","type":"preprint","venue":"Scientometrics","topic":"COVID-19 diagnosis using AI","field":"Medicine","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo; Research and Productivity Council; University of New Brunswick; Concordia University; National Research Council Canada","funders":"","keywords":"Coronavirus disease 2019 (COVID-19); Similarity (geometry); Contagious disease; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Outbreak; Diversity (politics); Data science; Disease; Computer science; Geography; Artificial intelligence; Medicine; Sociology; Virology; Infectious disease (medical specialty); Pathology","score_opus":0.3378602076196574,"score_gpt":0.4903504625910624,"score_spread":0.152490254971405,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3045327594","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25502032,0.22031733,0.25494152,0.25869906,0.0027009347,0.005622036,0.00019663776,0.00095472677,0.0015474223],"genre_scores_gemma":[0.99584496,0.00040567116,0.00174132,0.0014501839,0.0001880005,0.00001651195,0.00006957561,0.000038051243,0.00024572693],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99539155,0.00039701126,0.00042839127,0.00080854236,0.0024799902,0.0004945398],"domain_scores_gemma":[0.9968267,0.0016447923,0.00035575405,0.00040743424,0.00045980033,0.00030550972],"candidate_categories":["metaresearch","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.005486285,0.00024096607,0.000475453,0.0031455723,0.00063468824,0.00026903034,0.00041396156,0.00023111244,0.00002157489],"category_scores_gemma":[0.025585247,0.00017927111,0.000116959476,0.013209865,0.0008125959,0.00012301182,0.001532546,0.0025711097,0.0000039942556],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001200965,0.0010189676,0.6983858,0.036738254,0.0005745642,0.00066011876,0.18539836,0.006222739,0.010046804,0.005382322,0.025300136,0.029071033],"study_design_scores_gemma":[0.009628563,0.0020024271,0.034216344,0.0065113343,0.0012902624,0.00028314677,0.09130299,0.6646737,0.0019061061,0.021459404,0.16434242,0.0023833432],"about_ca_topic_score_codex":0.0016905001,"about_ca_topic_score_gemma":0.00007243236,"teacher_disagreement_score":0.74082464,"about_ca_system_score_codex":0.0033335818,"about_ca_system_score_gemma":0.0023386925,"threshold_uncertainty_score":0.99973},"labels":[],"label_agreement":null},{"id":"W3062515517","doi":"10.1007/s11192-020-03671-7","title":"Vocabulary sharing among subjects belonging to the hierarchy of sciences","year":2020,"lang":"en","type":"article","venue":"Scientometrics","topic":"linguistics and terminology studies","field":"Arts and Humanities","cited_by":30,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Brock University","funders":"","keywords":"Hierarchy; Vocabulary; Sociology; Linguistics; Computer science; Data science; Political science; Philosophy","score_opus":0.11905871706491908,"score_gpt":0.29160392100640864,"score_spread":0.17254520394148956,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3062515517","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9537126,0.00044044905,0.000109515764,0.0024097022,0.0008293785,0.00014469217,0.000015347337,0.000042455813,0.042295847],"genre_scores_gemma":[0.9981401,0.0000118045455,0.00015272752,0.0008109563,0.00039782605,0.0000039776223,6.494445e-7,0.0000053510194,0.00047660817],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99890095,0.0000124426,0.00015025337,0.00024971642,0.0004254033,0.00026124148],"domain_scores_gemma":[0.99942786,0.00014468233,0.000070041286,0.000109646666,0.00016573245,0.00008205002],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006237771,0.00007968607,0.00013070344,0.0007488649,0.00073176087,0.00020198437,0.0005774962,0.000014353241,0.000049373077],"category_scores_gemma":[0.0014488862,0.000051391424,0.0000442919,0.0020736107,0.00078974955,0.0000745679,0.0004024995,0.000088804736,0.000026175883],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000073878837,0.000047501333,0.28063682,0.00006955106,0.000061787505,0.000013077222,0.13188739,0.00024078168,0.000054666463,0.5749942,0.0057233325,0.0062634936],"study_design_scores_gemma":[0.0010719755,0.0019236819,0.31364506,0.00020780624,0.00020476076,0.0000047644344,0.029046163,0.020570869,0.004103156,0.011297728,0.6164929,0.0014311039],"about_ca_topic_score_codex":0.000105103056,"about_ca_topic_score_gemma":0.0001017444,"teacher_disagreement_score":0.6107696,"about_ca_system_score_codex":0.0000117866975,"about_ca_system_score_gemma":0.000022858645,"threshold_uncertainty_score":0.5628187},"labels":[],"label_agreement":null},{"id":"W3082051346","doi":"10.1007/s11192-020-03675-3","title":"Publishing volumes in major databases related to Covid-19","year":2020,"lang":"en","type":"article","venue":"Scientometrics","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":152,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thompson Rivers University","funders":"Wellcome Trust","keywords":"Scopus; Coronavirus disease 2019 (COVID-19); Web of science; Publishing; Pandemic; Library science; China; MEDLINE; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Bibliometrics; 2019-20 coronavirus outbreak; Analytics; History; Medicine; Database; Computer science; Political science; Disease; Infectious disease (medical specialty); Virology; Pathology","score_opus":0.2913286398736232,"score_gpt":0.4731993666241887,"score_spread":0.18187072675056554,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3082051346","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7162332,0.00092618755,0.026784074,0.22369587,0.002597289,0.00076855265,0.0003565596,0.00035612463,0.02828214],"genre_scores_gemma":[0.9618907,0.000012946734,0.0030630513,0.031712383,0.00012477559,0.000011498253,0.000034902143,0.00001719892,0.0031325317],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9908996,0.00034952245,0.0011256272,0.0013821194,0.005511123,0.00073198066],"domain_scores_gemma":[0.992232,0.0038245239,0.00035341026,0.0008349193,0.0005568161,0.0021982794],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.022288205,0.00019108283,0.00039978584,0.008143477,0.0002977279,0.009177243,0.0067982734,0.00016353326,0.0010007429],"category_scores_gemma":[0.5404868,0.00015316189,0.00008809343,0.11266095,0.000134435,0.010449136,0.0022979828,0.0008192403,0.0011625518],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002080313,0.000031810017,0.33567017,0.000006485586,0.0000040516475,0.0000351163,0.0030519322,0.0009001834,0.000047253783,0.0023021435,0.62909925,0.028830763],"study_design_scores_gemma":[0.00091403857,0.000051863615,0.045057885,0.000015146007,0.0000062718586,0.000008311176,0.0037728017,0.01013747,0.00011074282,0.0043494278,0.9351778,0.00039825216],"about_ca_topic_score_codex":0.000635189,"about_ca_topic_score_gemma":0.00011832625,"teacher_disagreement_score":0.5181986,"about_ca_system_score_codex":0.00029081025,"about_ca_system_score_gemma":0.0006656833,"threshold_uncertainty_score":0.9999125},"labels":[],"label_agreement":null},{"id":"W3085773434","doi":"10.1007/s11192-020-03701-4","title":"Classifications of science and their effects on bibliometric evaluations","year":2020,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":15,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bureau de Coopération Interuniversitaire; Université du Québec à Montréal; Université de Montréal","funders":"Social Sciences and Humanities Research Council of Canada; National Planning Office of Philosophy and Social Science","keywords":"Multidisciplinary approach; Bibliometrics; Discipline; Computer science; Order (exchange); Scientometrics; Data science; Management science; Sociology; Library science; Social science; Engineering; Business","score_opus":0.7109975348963583,"score_gpt":0.6052004678785627,"score_spread":0.10579706701779557,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3085773434","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9584671,0.0018010407,0.015508568,0.006252912,0.0009798473,0.00080519146,0.000109729306,0.00006972911,0.01600584],"genre_scores_gemma":[0.99732786,0.00022836008,0.0014276943,0.000704391,0.0000775168,0.00001650636,0.0000020265197,0.000012949771,0.00020271848],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.97100896,0.00027035672,0.0009563715,0.0014909594,0.02539883,0.00087450776],"domain_scores_gemma":[0.954363,0.020154089,0.0006719414,0.0015405224,0.021138715,0.0021317252],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.051733702,0.00023036069,0.00046877345,0.7541529,0.0008370977,0.0026323958,0.0044707567,0.00010571961,0.00013385898],"category_scores_gemma":[0.47328752,0.0001524147,0.00013477568,0.9758922,0.0022002042,0.0011055275,0.0012804038,0.0003460488,0.00030472918],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017930486,0.00023700253,0.04966729,0.000027419452,0.000012982932,0.0000019753488,0.00082057185,0.00007261092,0.03295281,0.0068160673,0.011856916,0.8975164],"study_design_scores_gemma":[0.00096274173,0.0012779959,0.8464807,0.000016948537,0.000014289269,0.0000042175593,0.00089704257,0.045960482,0.07213408,0.0041855546,0.027683962,0.00038194374],"about_ca_topic_score_codex":0.000010497166,"about_ca_topic_score_gemma":6.668176e-7,"teacher_disagreement_score":0.8971345,"about_ca_system_score_codex":0.00020955995,"about_ca_system_score_gemma":0.0010478008,"threshold_uncertainty_score":0.99840295},"labels":[],"label_agreement":null},{"id":"W3092009239","doi":"10.1007/s11192-020-03718-9","title":"Navigation-based candidate expansion and pretrained language models for citation recommendation","year":2020,"lang":"en","type":"article","venue":"Scientometrics","topic":"Topic Modeling","field":"Computer Science","cited_by":18,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Institute for Advanced Research; University of Waterloo","funders":"","keywords":"Computer science; Benchmark (surveying); Citation; Ranking (information retrieval); Vocabulary; Domain (mathematical analysis); Task (project management); Information retrieval; Artificial intelligence; Language model; Machine learning; Natural language processing; World Wide Web; Linguistics","score_opus":0.06364737314638658,"score_gpt":0.31168752864163235,"score_spread":0.24804015549524577,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3092009239","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09517525,0.000045794328,0.90101194,0.0031216105,0.0001771187,0.00024907367,0.000013397409,0.00009639718,0.00010940152],"genre_scores_gemma":[0.82742584,0.0000023238176,0.17170194,0.0007484847,0.000034256645,0.000014965668,0.00005169328,0.000004784515,0.000015745612],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9989907,0.000027742391,0.00016294951,0.00035353503,0.0003033163,0.0001617284],"domain_scores_gemma":[0.9992968,0.00017226106,0.00008036725,0.00015821244,0.000163771,0.00012860082],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00059309136,0.00007234412,0.00008019621,0.000525869,0.00011615316,0.0002035741,0.00024514887,0.000037534475,0.0000033954095],"category_scores_gemma":[0.00046942953,0.00007163172,0.000023553956,0.00330145,0.000014976526,0.00059973507,0.0000599403,0.00005289676,0.000002214574],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000032890483,0.000070365364,0.0010003905,0.00026141468,0.00001357071,0.000003016525,0.022496434,0.05438607,0.012389027,0.026520401,0.0011171895,0.8817092],"study_design_scores_gemma":[0.00042515685,0.000060696828,0.00015076602,0.000008393489,0.0000029777666,3.412212e-7,0.00007919788,0.99546087,0.00219711,0.001329466,0.00019824263,0.000086781176],"about_ca_topic_score_codex":0.000015014537,"about_ca_topic_score_gemma":0.0000015797244,"teacher_disagreement_score":0.9410748,"about_ca_system_score_codex":0.000044117547,"about_ca_system_score_gemma":0.000064489956,"threshold_uncertainty_score":0.2921057},"labels":[],"label_agreement":null},{"id":"W3095982673","doi":"10.1007/s11192-020-03727-8","title":"Mapping computer science research in Africa: using academic networking sites for assessing research activity","year":2020,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":17,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Cégep de Sorel-Tracy; McGill University; Concordia University; University of Calgary","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"Scopus; Bibliometrics; Web of science; Productivity; Regional science; Citation; Database; Library science; Data science; Computer science; Geography; Political science; MEDLINE; Economic growth","score_opus":0.9631006661554827,"score_gpt":0.7125945034653474,"score_spread":0.25050616269013537,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3095982673","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8637369,0.0015064485,0.12793884,0.002325575,0.001346555,0.0012222466,0.000019279663,0.000057520676,0.0018466375],"genre_scores_gemma":[0.97701013,0.0001251243,0.021480111,0.00012202501,0.00108467,0.000032435328,0.000001533551,0.000035682977,0.00010826884],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.93857133,0.0019790146,0.0014942025,0.003572211,0.04908288,0.005300357],"domain_scores_gemma":[0.9358036,0.039815404,0.00046126448,0.0014232349,0.02029973,0.002196801],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.35724887,0.0003184628,0.00071785616,0.3628614,0.003488654,0.014139009,0.009712496,0.00036351336,0.00004583602],"category_scores_gemma":[0.17549501,0.0002680495,0.00021061009,0.8699734,0.0029378529,0.0041725324,0.0069283866,0.003294362,0.0001600016],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007405911,0.00026224717,0.30351484,0.00008072893,0.000010980147,0.00004983738,0.0035727422,0.0039276974,0.07190828,0.00080173986,0.0074305055,0.6083663],"study_design_scores_gemma":[0.000716686,0.00024894424,0.07240373,0.00008029723,0.0000021088524,0.00000570211,0.0015001611,0.89858246,0.0022431887,0.005976413,0.017829005,0.00041127906],"about_ca_topic_score_codex":0.00008387188,"about_ca_topic_score_gemma":0.0000055212113,"teacher_disagreement_score":0.8946548,"about_ca_system_score_codex":0.0019534277,"about_ca_system_score_gemma":0.002632551,"threshold_uncertainty_score":0.9999772},"labels":[],"label_agreement":null},{"id":"W3125340956","doi":"10.1007/s11192-021-03868-4","title":"Tracking developments in artificial intelligence research: constructing and applying a new search strategy","year":2021,"lang":"en","type":"article","venue":"Scientometrics","topic":"Explainable Artificial Intelligence (XAI)","field":"Computer Science","cited_by":106,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Biotechnology and Biological Sciences Research Council; Directorate for Biological Sciences; Social Sciences and Humanities Research Council of Canada; National Research Foundation; National Natural Science Foundation of China; Engineering and Physical Sciences Research Council; Taishan Scholar Foundation of Shandong Province; National Institutes of Health; National Science Foundation","keywords":"Profiling (computer programming); Computer science; Artificial intelligence; Multidisciplinary approach; Benchmark (surveying); Domain (mathematical analysis); Applications of artificial intelligence; Data science; Social science; Sociology","score_opus":0.44533367302389276,"score_gpt":0.44675477312989736,"score_spread":0.0014211001060046025,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3125340956","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4258285,0.001148443,0.563968,0.00058387534,0.0007030542,0.0004989257,0.0000015609627,0.00012279603,0.0071448334],"genre_scores_gemma":[0.91180634,0.00009697664,0.087799795,0.000057121717,0.000058547506,0.000012242137,0.0000012108455,0.000011050228,0.00015672018],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99524826,0.00029845786,0.00058886973,0.0009880553,0.0017398477,0.0011365362],"domain_scores_gemma":[0.9971406,0.0010221417,0.000073650735,0.0005012864,0.0008709335,0.0003914236],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0063352813,0.0001665677,0.00022302454,0.0041238577,0.00045484508,0.0018655912,0.0011044566,0.0001061077,0.000026675922],"category_scores_gemma":[0.0036927965,0.00018558354,0.000033994926,0.036449064,0.00027891767,0.0011991512,0.0011087817,0.0006624271,0.00009653334],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000002841876,0.000053122632,0.003937588,0.000019187277,0.0000050750505,0.00017466037,0.0037697495,0.0006089016,0.0037508337,0.15712315,0.000013396907,0.8305415],"study_design_scores_gemma":[0.00020089508,0.00018124617,0.0062795705,0.00030863628,0.0000050740746,0.0002667342,0.049925268,0.11072734,0.6967097,0.13247816,0.001856644,0.001060734],"about_ca_topic_score_codex":0.00023027649,"about_ca_topic_score_gemma":0.00027157794,"teacher_disagreement_score":0.82948077,"about_ca_system_score_codex":0.00029225147,"about_ca_system_score_gemma":0.0014052644,"threshold_uncertainty_score":0.99917054},"labels":[],"label_agreement":null},{"id":"W3133429097","doi":"10.1007/s11192-020-03840-8","title":"Mapping the intellectual structure of GIS-T field (2008–2019): a dynamic co-word analysis","year":2021,"lang":"en","type":"article","venue":"Scientometrics","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":38,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"Kungliga Tekniska Högskolan","keywords":"Field (mathematics); Computer science; Geographic information system; Data science; Geography; Cartography; Mathematics","score_opus":0.15018153582932317,"score_gpt":0.4231165180604795,"score_spread":0.2729349822311563,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3133429097","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6214622,0.001964664,0.3621791,0.004536949,0.0016803071,0.00029985298,0.00055788463,0.000045558994,0.0072734714],"genre_scores_gemma":[0.9916468,0.00010185523,0.0017032721,0.00095345965,0.00002378392,0.0000011414206,0.00005577948,0.000004385886,0.005509496],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9950147,0.0003239149,0.0006925504,0.0005858497,0.003053088,0.00032989276],"domain_scores_gemma":[0.9925238,0.004927371,0.00031184353,0.0014142679,0.0007130405,0.00010968652],"candidate_categories":["metaresearch","bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0052230796,0.00012467551,0.00035881624,0.0055365353,0.00023456383,0.0005534725,0.001790265,0.00007235801,0.0032527316],"category_scores_gemma":[0.033832423,0.00007893023,0.00025649712,0.08218203,0.00019749178,0.00024303299,0.0006889948,0.00020203383,0.00012592415],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000038076498,0.00028328618,0.015477556,0.000047000347,0.000998196,0.00003617815,0.013152299,0.0027995051,0.0022349171,0.0054099695,0.34077552,0.6187475],"study_design_scores_gemma":[0.0007232401,0.00016640674,0.10466667,0.00002957616,0.00046791325,0.000014582295,0.030906355,0.032055072,0.010086856,0.01798619,0.802209,0.000688144],"about_ca_topic_score_codex":0.00006924326,"about_ca_topic_score_gemma":0.0003440056,"teacher_disagreement_score":0.61805934,"about_ca_system_score_codex":0.000052046158,"about_ca_system_score_gemma":0.00012594461,"threshold_uncertainty_score":0.99765843},"labels":[],"label_agreement":null},{"id":"W3135175820","doi":"10.1007/s11192-021-03911-4","title":"Digital technology helps remove gender bias in academia","year":2021,"lang":"en","type":"article","venue":"Scientometrics","topic":"Diversity and Career in Medicine","field":"Social Sciences","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Meritocracy; Gender bias; Altmetrics; Metric (unit); Gender disparity; Productivity; Psychology; Sociology; Political science; Computer science; Library science; Social psychology; Gender studies; Marketing","score_opus":0.14776339979880546,"score_gpt":0.3722449533529507,"score_spread":0.22448155355414526,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3135175820","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6080496,0.0014173817,0.00017904217,0.007199117,0.0012359006,0.00009945231,0.000014658104,0.000102360464,0.38170254],"genre_scores_gemma":[0.98354614,0.0002907062,0.00031664135,0.0010085497,0.00011867277,0.0000012264592,0.0000067123524,0.000004145167,0.014707184],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975072,0.00005023809,0.00013762298,0.0002797998,0.0016029441,0.00042217213],"domain_scores_gemma":[0.9991897,0.00015773994,0.000043739976,0.00016641602,0.00028025266,0.00016219562],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.0021629399,0.000064044776,0.00012490583,0.0024052747,0.0002100359,0.00020438373,0.00059090566,0.0004948156,0.00019046347],"category_scores_gemma":[0.008608742,0.0000665123,0.000039378578,0.029731145,0.00048045558,0.00031269714,0.00027419478,0.0006624186,0.00015443291],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000047071876,0.0001488684,0.7140703,0.00001634468,0.000013503262,0.00032591558,0.009759527,0.000011267201,0.00010504604,0.0654391,0.018204866,0.19190054],"study_design_scores_gemma":[0.00057506753,0.000031265852,0.03287601,0.000027771432,0.000011754355,0.000007662317,0.08312868,0.000019245083,0.000456676,0.009515674,0.8731153,0.00023490269],"about_ca_topic_score_codex":0.00003662736,"about_ca_topic_score_gemma":0.00007446185,"teacher_disagreement_score":0.85491043,"about_ca_system_score_codex":0.00022906903,"about_ca_system_score_gemma":0.00034025946,"threshold_uncertainty_score":0.99974215},"labels":[{"model":"gemma","categories":["metaresearch"],"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":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W3162449416","doi":"10.1007/s11192-021-04042-6","title":"Correction to: Prevalence of potentially predatory publishing in Scopus on the country level","year":2021,"lang":"en","type":"article","venue":"Scientometrics","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Russian Science Foundation; University of Waterloo; Russian Foundation for Basic Research","keywords":"Scopus; Publishing; Library science; Bibliometrics; Geography; Political science; Computer science; MEDLINE; Law","score_opus":0.1781125987131383,"score_gpt":0.4081491312190696,"score_spread":0.23003653250593128,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3162449416","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9234915,0.00064407435,0.009710692,0.0059168744,0.014713357,0.00039616376,0.00019555751,0.000039884137,0.04489189],"genre_scores_gemma":[0.98166186,0.00003792242,0.00046569813,0.0015722244,0.000101586156,0.0000078854955,0.0000037118018,0.000007961158,0.016141163],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.992035,0.0004423936,0.0007595057,0.0007368503,0.005632245,0.00039404488],"domain_scores_gemma":[0.9916574,0.0050140773,0.0004028667,0.0009485703,0.0017812862,0.00019581319],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.02506894,0.000121927,0.00022569987,0.003556388,0.0002017604,0.0043214527,0.003812931,0.00013673688,0.00034620528],"category_scores_gemma":[0.25376624,0.00008145078,0.00007088399,0.04789036,0.00012883636,0.003667208,0.0008349153,0.0007101733,0.000084118554],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000038088085,0.00020370845,0.2731969,0.000028954351,0.000009191506,0.00001519299,0.0013810033,0.0036377343,0.0008877895,0.003442632,0.5997053,0.11745354],"study_design_scores_gemma":[0.00040976718,0.000082518935,0.92978424,0.0003105086,0.000010380252,0.000016538435,0.0023723298,0.0054095048,0.007873607,0.0026546016,0.05078253,0.00029346842],"about_ca_topic_score_codex":0.00016753902,"about_ca_topic_score_gemma":0.00012687282,"teacher_disagreement_score":0.65658736,"about_ca_system_score_codex":0.00019504382,"about_ca_system_score_gemma":0.000730512,"threshold_uncertainty_score":0.99671215},"labels":[],"label_agreement":null},{"id":"W3179761068","doi":"10.1007/s11192-021-04022-w","title":"Who profits from the Canadian nanotechnology reward system? Implications for gender-responsible innovation","year":2021,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Polytechnique Montréal; Université du Québec à Montréal; Université de Montréal; Concordia University","funders":"Social Sciences and Humanities Research Council of Canada; Canada Research Chairs","keywords":"Privilege (computing); Productivity; Gender equality; Work (physics); Gender gap; Field (mathematics); Reward system; Gender disparity; Political science; Public relations; Sociology; Psychology; Demographic economics; Gender studies; Economic growth; Economics; Engineering; Law","score_opus":0.7142155899698618,"score_gpt":0.5720511809625689,"score_spread":0.14216440900729288,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3179761068","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.75225896,0.0053738216,0.12223459,0.08652956,0.0062336493,0.0030111545,0.0038453657,0.00028603058,0.020226851],"genre_scores_gemma":[0.9871114,0.000056965102,0.0074114073,0.0009952169,0.00017037883,0.00013601624,0.00010601946,0.000022839473,0.0039897715],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.989327,0.0003286738,0.0010745901,0.0013810998,0.0066530323,0.001235561],"domain_scores_gemma":[0.9643256,0.010315587,0.00045109878,0.0023425946,0.021968225,0.00059687515],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.033165045,0.00020754372,0.00037970368,0.13651237,0.0020832922,0.0051373127,0.004239121,0.00036119917,0.00013091152],"category_scores_gemma":[0.214897,0.00014171036,0.00016813431,0.7040871,0.00032989582,0.00051083567,0.00079573895,0.0004001644,0.00031384965],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017402523,0.00010357132,0.2631528,0.00001617454,0.0000510756,0.000012722141,0.00030389905,0.00007667996,0.0048008636,0.40350473,0.16648428,0.1614758],"study_design_scores_gemma":[0.0006402486,0.000085325424,0.5062875,0.0000151238755,0.00002088473,0.000024672576,0.0010532852,0.0030936648,0.006847499,0.055856086,0.42571914,0.00035653266],"about_ca_topic_score_codex":0.007166742,"about_ca_topic_score_gemma":0.023359679,"teacher_disagreement_score":0.56757474,"about_ca_system_score_codex":0.001132596,"about_ca_system_score_gemma":0.005349234,"threshold_uncertainty_score":0.9994446},"labels":[],"label_agreement":null},{"id":"W3184001797","doi":"10.1007/s11192-022-04418-2","title":"Reproducibility of COVID-19 pre-prints","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Coronavirus disease 2019 (COVID-19); Reproducibility; Code (set theory); Sample (material); Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Underpinning; Parsing; 2019-20 coronavirus outbreak; Computer science; Information retrieval; Biology; Artificial intelligence; Mathematics; Medicine; Statistics; Chemistry; Virology; Engineering; Chromatography; Pathology","score_opus":0.26096079931966837,"score_gpt":0.4948671563696969,"score_spread":0.23390635705002855,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3184001797","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9701893,0.0003979053,0.005572559,0.004711528,0.002424285,0.00032852218,0.00013979456,0.00006761154,0.016168525],"genre_scores_gemma":[0.98874646,0.0000057651205,0.0005836948,0.001068278,0.000049464463,0.00001633264,0.0000049579207,0.0000060320385,0.009518999],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9865512,0.0008323908,0.0010786146,0.0024248671,0.008695818,0.0004171072],"domain_scores_gemma":[0.9897251,0.0033252244,0.0008203592,0.004822536,0.00073851366,0.0005682761],"candidate_categories":["metaresearch","bibliometrics","open_science","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.13003221,0.00010099097,0.00030213813,0.004709946,0.00059362187,0.00051034905,0.0063821315,0.000056705332,0.0031720223],"category_scores_gemma":[0.40887487,0.00008051571,0.00012963617,0.0532653,0.000272146,0.0009091959,0.0036496562,0.00048486947,0.000062347644],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000046873527,0.00022036173,0.7151869,0.000010906078,0.00000791693,0.0000049947603,0.0027705932,0.0045670173,0.00015461682,0.0022645213,0.22324777,0.05151753],"study_design_scores_gemma":[0.00046039978,0.000103249986,0.24152063,0.0000015098858,0.000008046136,0.000017645221,0.002300317,0.0019302388,0.0008612798,0.049234673,0.70333195,0.00023003954],"about_ca_topic_score_codex":0.0002597722,"about_ca_topic_score_gemma":0.000005061824,"teacher_disagreement_score":0.4800842,"about_ca_system_score_codex":0.0003645854,"about_ca_system_score_gemma":0.0008023402,"threshold_uncertainty_score":0.9989938},"labels":[],"label_agreement":null},{"id":"W3190254501","doi":"10.1007/s11192-022-04408-4","title":"A systematic method for identifying references to academic research in grey literature","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"Optics and Image Analysis","field":"Business, Management and Accounting","cited_by":11,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Newcastle University; University of Oxford; Queen's University Belfast; University College London; University of Warwick; University of Southampton; Queen's University; London School of Economics and Political Science; Queen Mary University of London","keywords":"Grey literature; Government (linguistics); Computer science; Heuristics; Bibliometrics; Scientific literature; Information retrieval; Matching (statistics); Metadata; Legislation; Library science; Operations research; Political science; Statistics; MEDLINE; World Wide Web; Law; Mathematics","score_opus":0.22978697577162527,"score_gpt":0.4608031597729013,"score_spread":0.23101618400127605,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3190254501","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8053877,0.016699634,0.11362905,0.016017092,0.0066395337,0.011395914,0.00013733043,0.0003953409,0.029698428],"genre_scores_gemma":[0.98395604,0.000029159071,0.010793471,0.0012676894,0.00039824736,0.0005701258,0.000039304265,0.00002711079,0.0029188672],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9970072,0.00009153868,0.0004121609,0.0004529545,0.0015310243,0.0005051213],"domain_scores_gemma":[0.9979988,0.0004648759,0.00015242341,0.00029209148,0.0010637465,0.00002810577],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.016462646,0.000105790816,0.00027650013,0.016711,0.0005801294,0.00177151,0.00086093874,0.00004480869,0.000053815216],"category_scores_gemma":[0.006004028,0.000094102885,0.00009351392,0.06628922,0.000018061332,0.0010345611,0.0008416722,0.00047934576,0.00006559642],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00049135566,0.0018923947,0.096610576,0.17656948,0.0006882925,0.00036735006,0.016558737,0.016072009,0.016631782,0.42714816,0.15659966,0.0903702],"study_design_scores_gemma":[0.0031546762,0.0003250937,0.016156789,0.008543762,0.0006011491,0.000018492781,0.032782655,0.65284646,0.00030541423,0.09662567,0.18619768,0.002442141],"about_ca_topic_score_codex":0.000174191,"about_ca_topic_score_gemma":0.000033911972,"teacher_disagreement_score":0.6367745,"about_ca_system_score_codex":0.00015954697,"about_ca_system_score_gemma":0.000037002792,"threshold_uncertainty_score":0.9992647},"labels":[],"label_agreement":null},{"id":"W3195597029","doi":"10.1007/s11192-021-04118-3","title":"A credit-like rating system to determine the legitimacy of scientific journals and publishers","year":2021,"lang":"en","type":"article","venue":"Scientometrics","topic":"Private Equity and Venture Capital","field":"Business, Management and Accounting","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Vlaamse regering; Thompson Rivers University","keywords":"Economic Justice; Publishing; Credit rating; Quality (philosophy); Actuarial science; Business; Political science; Law; Epistemology","score_opus":0.05810792340576652,"score_gpt":0.28104658239953,"score_spread":0.22293865899376347,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3195597029","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98424613,0.00095975667,0.00094880897,0.0016735367,0.0038146183,0.000212593,0.000006199328,0.000049422288,0.008088919],"genre_scores_gemma":[0.99785703,0.0000028865377,0.00040048,0.00061342504,0.0005870709,0.0000054138177,0.000008027698,0.0000103487755,0.00051530113],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99797434,0.000018945431,0.00032997137,0.00034044456,0.000992567,0.00034373894],"domain_scores_gemma":[0.9982219,0.0001370169,0.00024672603,0.00035562876,0.0009899572,0.000048773796],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0031579235,0.00011557801,0.0001894476,0.0021185295,0.0005275709,0.005251329,0.0005123983,0.000041354557,0.00007406615],"category_scores_gemma":[0.0036048426,0.00008531166,0.000074289026,0.015839685,0.00016523359,0.0021878,0.00076987135,0.00013679003,0.0000357317],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010523804,0.0008888161,0.17856246,0.0044756667,0.0002951536,0.0004919669,0.0042163744,0.00065021653,0.08886877,0.08781134,0.3840067,0.24962729],"study_design_scores_gemma":[0.0031244333,0.00009924843,0.13256392,0.0012819883,0.00037633776,0.00022575136,0.020770026,0.035520986,0.006571556,0.0017103549,0.79624164,0.0015137505],"about_ca_topic_score_codex":0.000022210734,"about_ca_topic_score_gemma":0.000014807456,"teacher_disagreement_score":0.41223496,"about_ca_system_score_codex":0.000045009456,"about_ca_system_score_gemma":0.00007338904,"threshold_uncertainty_score":0.9957813},"labels":[],"label_agreement":null},{"id":"W3205605438","doi":"10.1007/s11192-022-04351-4","title":"Return migration of German-affiliated researchers: analyzing departure and return by gender, cohort, and discipline using Scopus bibliometric data 1996–2020","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"Spatial and Panel Data Analysis","field":"Economics, Econometrics and Finance","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia; University of Toronto; University of New Brunswick","funders":"Bundesministerium für Bildung und Forschung; Max-Planck-Institut für demografische Forschung; Deutscher Akademischer Austauschdienst","keywords":"Scopus; German; Emigration; Demographic economics; Tracking (education); Bibliometrics; Construct (python library); Political science; Sociology; Geography; Library science; Economics; MEDLINE; Computer science","score_opus":0.18269089971254934,"score_gpt":0.35642126913825983,"score_spread":0.1737303694257105,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3205605438","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9585166,0.032684207,0.0023155415,0.000227578,0.00020675054,0.00024163642,0.0056449673,0.000022127715,0.00014061351],"genre_scores_gemma":[0.9927131,0.0034772344,0.0012599784,0.000060738214,0.000054768887,0.000007005215,0.0022177973,0.000022032516,0.0001873255],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99728847,0.00009767007,0.0007610766,0.0009737025,0.00045326634,0.00042582865],"domain_scores_gemma":[0.997927,0.00024193992,0.0006032947,0.0008582227,0.00012472784,0.00024481278],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0048349607,0.00018703473,0.0005103449,0.035053097,0.0006071286,0.00040111464,0.00076092454,0.00009151579,0.00022376754],"category_scores_gemma":[0.0015621122,0.00020201069,0.00006162261,0.14920114,0.00020512195,0.00076021603,0.0014470036,0.00034050402,0.000003941193],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000026863565,0.0001064564,0.98934126,0.0000712031,0.00014142659,0.000007135346,0.00038146912,0.00009684908,0.0009472523,0.00009094687,0.00654584,0.002243316],"study_design_scores_gemma":[0.0012080264,0.0002384154,0.38131416,0.000020588552,0.00019767671,0.00003257418,0.0005107933,0.5807839,0.00026995662,0.0011540836,0.033524465,0.00074535236],"about_ca_topic_score_codex":0.0039838958,"about_ca_topic_score_gemma":0.00028851037,"teacher_disagreement_score":0.6080271,"about_ca_system_score_codex":0.00014763958,"about_ca_system_score_gemma":0.00005593422,"threshold_uncertainty_score":0.9758838},"labels":[],"label_agreement":null},{"id":"W3206649023","doi":"10.1007/s11192-021-04154-z","title":"Time series-based bibliometric analysis of a systematic review of multidisciplinary care for opioid dose reduction: exploring the origins of the North American opioid crisis","year":2021,"lang":"en","type":"review","venue":"Scientometrics","topic":"Opioid Use Disorder Treatment","field":"Medicine","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Public Health Ontario; Western University; Institute for Work & Health; Institute of Health Services and Policy Research; Lunenfeld-Tanenbaum Research Institute; University of Toronto","funders":"Canadian Institutes of Health Research","keywords":"Multidisciplinary approach; Metric (unit); Citation; Opioid; Health psychology; Chronic pain; Phenomenon; Bibliometrics; Citation analysis; Medicine; Psychology; Psychiatry; Public health; Sociology; Social science; Data mining; Computer science; Epistemology; Economics; Nursing; Library science","score_opus":0.08741432870258167,"score_gpt":0.39051394589559163,"score_spread":0.30309961719300993,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3206649023","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005031547,0.98618746,0.000039919014,0.00006997814,0.0004832776,0.0068609384,0.0012878651,0.000023536226,0.000015469244],"genre_scores_gemma":[0.0011346532,0.99520785,0.0012709119,0.000020356445,0.000045286226,0.001432639,0.0007475357,0.00006472517,0.000076028155],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.9934516,0.0004741708,0.0024515905,0.00077829865,0.0024082873,0.00043605742],"domain_scores_gemma":[0.9896333,0.0010954583,0.0034248207,0.0028452817,0.0028209288,0.00018021443],"candidate_categories":["metaepi_narrow","bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.0016605376,0.0005828314,0.006778519,0.101497814,0.00018690477,0.00003578346,0.0010892985,0.00009235128,0.000028837274],"category_scores_gemma":[0.0038783492,0.00031041412,0.0036527452,0.67566234,0.00043452304,0.00011332055,0.00033561327,0.00027882966,0.0000036837137],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010976034,0.00047362497,0.00064197625,0.96585226,0.00332309,0.0000026400423,0.00076471514,0.000037019952,5.71221e-7,0.0000072012795,0.00010332157,0.02878262],"study_design_scores_gemma":[0.0009799636,0.0012762167,0.004965016,0.74379164,0.21162716,0.000027885855,0.005081724,0.00012518956,0.00009350906,6.693843e-7,0.03129993,0.0007311255],"about_ca_topic_score_codex":0.0001454686,"about_ca_topic_score_gemma":0.000011811676,"teacher_disagreement_score":0.57416457,"about_ca_system_score_codex":0.0010259579,"about_ca_system_score_gemma":0.0019165225,"threshold_uncertainty_score":0.9999348},"labels":[],"label_agreement":null},{"id":"W4212870265","doi":"10.1007/s11192-022-04285-x","title":"Size of science team at university and internal co-publications: science policy implications","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Agence Universitaire de la Francophonie","keywords":"Science policy; Scientometrics; Sociology; Political science; Library science; Computer science; Public administration","score_opus":0.4464727771215093,"score_gpt":0.578309172696832,"score_spread":0.13183639557532273,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4212870265","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92864186,0.0002531483,0.00095357304,0.002390833,0.0004817899,0.00036589528,0.00037409493,0.00004346586,0.066495344],"genre_scores_gemma":[0.9871949,0.000108153305,0.001398882,0.00013002669,0.00002764337,0.0000066133834,0.000002517645,0.000007173505,0.011124109],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9755877,0.00018062767,0.0006791027,0.0015177748,0.020959802,0.0010749503],"domain_scores_gemma":[0.977461,0.005715579,0.0008815358,0.0021861007,0.012057661,0.0016981188],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","sts","open_science"],"category_scores_codex":[0.060403705,0.00015402885,0.000285068,0.30223393,0.0051577026,0.0018893303,0.010023129,0.000039847597,0.0006386898],"category_scores_gemma":[0.15605552,0.00013735727,0.000105501036,0.798468,0.009322685,0.0018991062,0.009329715,0.00033187043,0.000058663696],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003231434,0.0006916583,0.7255284,0.000008313908,0.000010805074,0.000002626248,0.0012704588,0.00014019314,0.07739825,0.104712255,0.023323601,0.066881076],"study_design_scores_gemma":[0.00045915923,0.00024788885,0.87894547,0.0000019106867,0.000004861083,0.000045760982,0.0016962192,0.0022191794,0.004234893,0.0035983382,0.10833031,0.00021603427],"about_ca_topic_score_codex":0.00024596942,"about_ca_topic_score_gemma":0.000007154295,"teacher_disagreement_score":0.49623403,"about_ca_system_score_codex":0.0025332859,"about_ca_system_score_gemma":0.00418789,"threshold_uncertainty_score":0.9991468},"labels":[],"label_agreement":null},{"id":"W4214582452","doi":"10.1007/s11192-022-04305-w","title":"Retracted articles in oncology in the last three decades: frequency, reasons, and themes","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":32,"is_retracted":true,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Université du Québec à Montréal","funders":"","keywords":"Clinical Oncology; Oncology; China; Bibliometrics; Editorial board; Radiation oncology; Medicine; Library science; Cancer; Internal medicine; Political science; Computer science","score_opus":0.06008256517620937,"score_gpt":0.3581804415893788,"score_spread":0.2980978764131694,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4214582452","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9491235,0.001157824,0.000026219524,0.0072438074,0.0003739551,0.00023474361,0.0000132957675,0.000021568349,0.04180507],"genre_scores_gemma":[0.998539,0.00027085727,0.00027602937,0.0006657301,0.000065985565,0.000024806011,0.0000027016174,0.0000035048513,0.00015140056],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9972347,0.0008084304,0.00022922484,0.00023783327,0.0010743808,0.00041541367],"domain_scores_gemma":[0.9979409,0.0017110539,0.00008737382,0.00013030635,0.0000583359,0.000072023766],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.013093946,0.00006926893,0.00012292733,0.0014885548,0.0006180672,0.000072149276,0.00067324407,0.00025857493,0.00023968336],"category_scores_gemma":[0.00599664,0.000054775224,0.00002472988,0.015163713,0.00038859234,0.00019721987,0.0001350698,0.0018018836,0.000008368239],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013702011,0.00019819522,0.7622908,0.000004406311,0.000003904941,0.000033462275,0.10526664,0.000011421104,0.00012227242,0.09981692,0.0047062966,0.027531998],"study_design_scores_gemma":[0.00051736407,0.00016823328,0.5869892,0.0000142407325,0.000009169418,0.00001466096,0.1369798,0.00030851545,0.000029208864,0.059009414,0.21573196,0.0002281808],"about_ca_topic_score_codex":0.0051332903,"about_ca_topic_score_gemma":0.020980092,"teacher_disagreement_score":0.21102567,"about_ca_system_score_codex":0.00035476265,"about_ca_system_score_gemma":0.00031102673,"threshold_uncertainty_score":0.99688447},"labels":[],"label_agreement":null},{"id":"W4220877524","doi":"10.1007/s11192-022-04338-1","title":"Interdisciplinarity and impact: the effects of the citation time window","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":42,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Université du Québec à Montréal","funders":"National Planning Office of Philosophy and Social Science","keywords":"Citation; Impact factor; Window (computing); Citation analysis; Scientometrics; Computer science; Bibliometrics; Data science; Citation impact; Statistics; Data mining; Mathematics; Library science; Political science; World Wide Web; Law","score_opus":0.2744588650714582,"score_gpt":0.5399536004006232,"score_spread":0.265494735329165,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4220877524","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9933278,0.0011457123,0.0004284798,0.0014297593,0.0013937205,0.0005191043,0.000058800422,0.000011354456,0.0016853003],"genre_scores_gemma":[0.9969249,0.000026688434,0.00007250484,0.0001155192,0.000029020994,0.00002020112,0.0000015103891,0.0000072704447,0.0028024302],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98200136,0.00076254556,0.0005336002,0.000574779,0.01561731,0.0005104088],"domain_scores_gemma":[0.98061275,0.015337603,0.00050726935,0.0012520904,0.0020106167,0.00027963938],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.039560106,0.00013243733,0.00024681468,0.06512869,0.0014785513,0.0013092889,0.0041198824,0.000044008237,0.0003657265],"category_scores_gemma":[0.09398838,0.000060919396,0.0002330613,0.54746395,0.00057902344,0.00035674177,0.0058919247,0.00045579096,0.000040432344],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000048550923,0.0004722862,0.69177157,0.000030287503,0.00004792501,0.000006802905,0.0034699466,0.00058611605,0.010552847,0.001568631,0.05059177,0.2408533],"study_design_scores_gemma":[0.0003983947,0.0003706958,0.9784735,0.0000040565405,0.000011274314,0.000013406973,0.0005296089,0.0066401265,0.0010800678,0.008498293,0.0038697487,0.00011083769],"about_ca_topic_score_codex":0.000051014402,"about_ca_topic_score_gemma":0.0000029873113,"teacher_disagreement_score":0.48233527,"about_ca_system_score_codex":0.0002523708,"about_ca_system_score_gemma":0.00023502152,"threshold_uncertainty_score":0.99982136},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4221058759","doi":"10.1007/s11192-022-04331-8","title":"Music information visualization and classical composers discovery: an application of network graphs, multidimensional scaling, and support vector machines","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"Music and Audio Processing","field":"Computer Science","cited_by":54,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Multidimensional scaling; Computer science; Visualization; Pairwise comparison; Similarity (geometry); Music theory; Musical; Musicology; Realm; Musical composition; Classical music; Artificial intelligence; Art; Machine learning; Visual arts; History","score_opus":0.01655784240464562,"score_gpt":0.2749442264079385,"score_spread":0.25838638400329283,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4221058759","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4528466,0.000085983665,0.54645336,0.00010860279,0.0002581769,0.00013938057,0.000010692327,0.000040276238,0.000056935452],"genre_scores_gemma":[0.98765916,0.000007756206,0.0117851645,0.00042113685,0.00002475194,0.000010045025,0.00007475657,0.0000037525342,0.0000134954325],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99863064,0.000057220677,0.0002409623,0.00024145072,0.00066306867,0.00016665985],"domain_scores_gemma":[0.9993433,0.00006809102,0.00021865926,0.00016858467,0.000108452245,0.000092917486],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00088946323,0.00008437691,0.00011800258,0.0008066913,0.000410253,0.00020120743,0.00023015021,0.000025740504,0.0000050243925],"category_scores_gemma":[0.00006086503,0.00007971075,0.000019109311,0.0051317993,0.000100276644,0.0016636271,0.00038690053,0.000080827456,5.833529e-7],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004341108,0.00033984592,0.098422796,0.0001812843,0.000021493723,0.0000016888224,0.0065832003,0.01119957,0.002654388,0.44355857,0.0023749534,0.4346188],"study_design_scores_gemma":[0.0003648667,0.00014247946,0.07519391,0.0000058426212,0.000007877733,0.000011327344,0.00007431513,0.91737264,0.0001514842,0.0013087072,0.005219218,0.00014730562],"about_ca_topic_score_codex":0.000016133581,"about_ca_topic_score_gemma":0.00000222879,"teacher_disagreement_score":0.9061731,"about_ca_system_score_codex":0.000030874977,"about_ca_system_score_gemma":0.000059742775,"threshold_uncertainty_score":0.325051},"labels":[],"label_agreement":null},{"id":"W4240546189","doi":"10.1007/s11192-006-0097-x","title":"Preface","year":2006,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Bibliometrics; Interpretation (philosophy); Citation; Macro; Data science; Citation analysis; Computer science; Scientometrics; Epistemology; Sociology; Management science; Social science; Library science; Philosophy; Economics","score_opus":0.6899883062376743,"score_gpt":0.6221987966117799,"score_spread":0.06778950962589436,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4240546189","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6951882,0.0019262349,0.03157675,0.0012007128,0.0024349948,0.0003681535,0.00005389798,0.0001569882,0.26709408],"genre_scores_gemma":[0.9545873,0.000030979187,0.0035014146,0.0001367848,0.00021210685,0.00000808466,0.0000057769366,0.00001446763,0.041503105],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9709998,0.00018438775,0.0009194348,0.0011601938,0.025572687,0.0011635032],"domain_scores_gemma":[0.98823357,0.004107932,0.0003108647,0.0013859835,0.005362674,0.0005989694],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.04056021,0.00019610455,0.0003215265,0.23710883,0.00045066926,0.005520524,0.004318613,0.00013791367,0.0013559798],"category_scores_gemma":[0.063997105,0.00013755019,0.00021930598,0.7174109,0.00030948242,0.00095442066,0.0010133083,0.0002797748,0.004211696],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010506521,0.0004090271,0.4820734,0.0000040695045,0.000006513895,0.000026811887,0.00004762019,0.00051724486,0.0019988555,0.023244454,0.30456445,0.18709704],"study_design_scores_gemma":[0.0004197239,0.00009214846,0.51089877,0.0000020891785,0.0000035159821,0.00001031488,0.000073439034,0.0044533517,0.002418191,0.030205755,0.45114565,0.00027702568],"about_ca_topic_score_codex":0.00018243773,"about_ca_topic_score_gemma":0.00001985051,"teacher_disagreement_score":0.4803021,"about_ca_system_score_codex":0.00022680432,"about_ca_system_score_gemma":0.00022269871,"threshold_uncertainty_score":0.9995569},"labels":[],"label_agreement":null},{"id":"W4240741669","doi":"10.1556/scient.67.2006.2.7","title":"Preface","year":2006,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Bibliometrics; Interpretation (philosophy); Citation; Scientometrics; Citation analysis; Macro; Computer science; Data science; Epistemology; Sociology; Library science; Philosophy","score_opus":0.6899883062376743,"score_gpt":0.6221987966117799,"score_spread":0.06778950962589436,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4240741669","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6951882,0.0019262349,0.03157675,0.0012007128,0.0024349948,0.0003681535,0.00005389798,0.0001569882,0.26709408],"genre_scores_gemma":[0.9545873,0.000030979187,0.0035014146,0.0001367848,0.00021210685,0.00000808466,0.0000057769366,0.00001446763,0.041503105],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9709998,0.00018438775,0.0009194348,0.0011601938,0.025572687,0.0011635032],"domain_scores_gemma":[0.98823357,0.004107932,0.0003108647,0.0013859835,0.005362674,0.0005989694],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.04056021,0.00019610455,0.0003215265,0.23710883,0.00045066926,0.005520524,0.004318613,0.00013791367,0.0013559798],"category_scores_gemma":[0.063997105,0.00013755019,0.00021930598,0.7174109,0.00030948242,0.00095442066,0.0010133083,0.0002797748,0.004211696],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010506521,0.0004090271,0.4820734,0.0000040695045,0.000006513895,0.000026811887,0.00004762019,0.00051724486,0.0019988555,0.023244454,0.30456445,0.18709704],"study_design_scores_gemma":[0.0004197239,0.00009214846,0.51089877,0.0000020891785,0.0000035159821,0.00001031488,0.000073439034,0.0044533517,0.002418191,0.030205755,0.45114565,0.00027702568],"about_ca_topic_score_codex":0.00018243773,"about_ca_topic_score_gemma":0.00001985051,"teacher_disagreement_score":0.4803021,"about_ca_system_score_codex":0.00022680432,"about_ca_system_score_gemma":0.00022269871,"threshold_uncertainty_score":0.9995569},"labels":[],"label_agreement":null},{"id":"W4248757258","doi":"10.1556/scient.67.2006.1.1","title":"Key Labs and Open Labs in the Chinese scientific research system: Their role in the national and international scientific arena","year":2006,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"China; Science Citation Index; Citation; Key (lock); Library science; Chinese academy of sciences; Quarter (Canadian coin); Political science; Resource (disambiguation); Computer science; Data science; World Wide Web; History; Law","score_opus":0.5945123837319141,"score_gpt":0.593806149860948,"score_spread":0.0007062338709660221,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4248757258","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9509983,0.0029052193,0.000052442043,0.0037106278,0.0012437458,0.0010476724,0.00016419026,0.000014192637,0.039863598],"genre_scores_gemma":[0.9951642,0.000044835433,0.00023932436,0.00012848998,0.00015072334,0.00008031988,0.000028633081,0.000011178877,0.0041522626],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9640816,0.0020847549,0.0012719936,0.0021770175,0.028954996,0.0014296323],"domain_scores_gemma":[0.97729456,0.0141298175,0.00030019623,0.0012071647,0.0067909774,0.0002772922],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.27039412,0.00024684524,0.00035170827,0.16124608,0.002049613,0.0528011,0.011581393,0.00013899885,0.00009072709],"category_scores_gemma":[0.050815355,0.00012271648,0.00007877634,0.5613865,0.002106619,0.0019925833,0.0038017046,0.0008250338,0.0001606389],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004910176,0.00092479656,0.85550433,0.000027699401,0.000011020989,0.000040698662,0.0072135213,0.00016969391,0.00113616,0.03542786,0.06537916,0.034115948],"study_design_scores_gemma":[0.00090382586,0.00007145682,0.8559312,0.000027064274,0.0000018033272,0.000050993287,0.008599245,0.018287303,0.000056334957,0.03324895,0.08260381,0.00021798928],"about_ca_topic_score_codex":0.000543033,"about_ca_topic_score_gemma":0.0012826539,"teacher_disagreement_score":0.40014046,"about_ca_system_score_codex":0.00043094708,"about_ca_system_score_gemma":0.0005685454,"threshold_uncertainty_score":0.9992496},"labels":[],"label_agreement":null},{"id":"W4254187858","doi":"10.1556/scient.67.2006.2.10","title":"Why are hyperlinks to business Websites created? A content analysis","year":2006,"lang":"en","type":"article","venue":"Scientometrics","topic":"Competitive and Knowledge Intelligence","field":"Business, Management and Accounting","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Hyperlink; Competitor analysis; Link analysis; Context (archaeology); Content analysis; Link (geometry); Business information; Business; Computer science; Sample (material); World Wide Web; Marketing; Advertising; Web page; Sociology; Geography","score_opus":0.06684286375822009,"score_gpt":0.2673774425083551,"score_spread":0.20053457875013503,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4254187858","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9026318,0.0005941939,0.030430913,0.0016549454,0.00090248266,0.00031192403,0.000015119179,0.00022157395,0.063237056],"genre_scores_gemma":[0.9905168,0.0000075026023,0.0003143253,0.0033144518,0.0008202094,0.000015951273,0.000040331262,0.00001729959,0.004953128],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9980248,0.000007919258,0.00032438728,0.0005064715,0.00069467473,0.0004417441],"domain_scores_gemma":[0.9971528,0.00010066581,0.00017762497,0.00034628855,0.0021859808,0.000036622452],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00054650544,0.00020020403,0.00031342648,0.008727916,0.0002323771,0.00083603035,0.0005043694,0.00006126828,0.0007808515],"category_scores_gemma":[0.0010442658,0.00017616138,0.0001701338,0.0686396,0.00007093893,0.00054384203,0.00029091683,0.00009642124,0.0021166948],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017365228,0.00028742093,0.9601762,0.000054691707,0.000082972736,0.000026448208,0.000017651972,0.0011124556,0.0008227435,0.0052732043,0.028541023,0.003587791],"study_design_scores_gemma":[0.00014072686,0.000006915317,0.6758373,0.000029087529,0.00023249796,6.4899035e-7,0.00016881494,0.00257296,0.00032432986,0.0002494198,0.3201486,0.0002887313],"about_ca_topic_score_codex":0.0027824936,"about_ca_topic_score_gemma":0.0024237528,"teacher_disagreement_score":0.29160756,"about_ca_system_score_codex":0.00007503645,"about_ca_system_score_gemma":0.000017214012,"threshold_uncertainty_score":0.99866027},"labels":[],"label_agreement":null},{"id":"W4281255308","doi":"10.1007/s11192-022-04387-6","title":"Measuring the disparity among scientific disciplines using Library of Congress Subject Headings","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Subject (documents); Library of congress; Computer science; Information retrieval; Bibliometrics; Scientific literature; Library science; Data science; Scientific communication","score_opus":0.6099867867029692,"score_gpt":0.5220149078759858,"score_spread":0.08797187882698343,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4281255308","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9879879,0.0015075497,0.0013361184,0.00052365946,0.0053552277,0.00053315965,0.00029660552,0.000052499592,0.002407278],"genre_scores_gemma":[0.9935462,0.000035789883,0.00087497046,0.000054386677,0.00008843888,0.000023520164,0.000013053303,0.000027988659,0.005335636],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9656685,0.0009097184,0.0013897247,0.0015005269,0.029245835,0.0012857212],"domain_scores_gemma":[0.984528,0.007917204,0.0011270529,0.0023133033,0.0033797661,0.00073464296],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","open_science"],"category_scores_codex":[0.07053809,0.00027002013,0.00055330124,0.15632807,0.0039363354,0.0077411365,0.00913916,0.000080132384,0.0012392598],"category_scores_gemma":[0.044533387,0.0001722012,0.0003984683,0.67784804,0.0020639056,0.002198376,0.008683388,0.00074123207,0.000038542956],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000030800664,0.0003192126,0.9601827,0.000016292148,0.000021679974,0.000013142909,0.0005438753,0.0035693957,0.0020777837,0.0012293599,0.0125460075,0.019449735],"study_design_scores_gemma":[0.0009178443,0.00026934507,0.7697348,0.000025490062,0.00003962694,0.00003573169,0.0031927584,0.12977989,0.010902231,0.007883293,0.07650656,0.0007124588],"about_ca_topic_score_codex":0.00021839826,"about_ca_topic_score_gemma":0.000025655732,"teacher_disagreement_score":0.52151996,"about_ca_system_score_codex":0.00029372665,"about_ca_system_score_gemma":0.0007736293,"threshold_uncertainty_score":0.9996737},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4285490980","doi":"10.1007/s11192-022-04466-8","title":"Scientometric portraits of recognized scientists: a structured literature review","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Tech University","funders":"Macquarie University","keywords":"Portrait; Scientometrics; Data science; Library science; Sociology; Information retrieval; History; Computer science; Art history","score_opus":0.4423497706215747,"score_gpt":0.5649995372104198,"score_spread":0.12264976658884513,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4285490980","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.45691952,0.46987295,0.0047461623,0.003640614,0.016643472,0.004522085,0.0039129737,0.00030529828,0.039436907],"genre_scores_gemma":[0.963683,0.014189549,0.0064422474,0.0016464108,0.00013723849,0.00013118959,0.00013681436,0.00006159011,0.0135719525],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9337935,0.0012845664,0.0028235707,0.002651866,0.057525113,0.0019214149],"domain_scores_gemma":[0.9713778,0.00540963,0.0021901133,0.0031631733,0.016285114,0.0015741899],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.11765935,0.00045941147,0.0012589478,0.4470861,0.00156462,0.0028881703,0.010750653,0.00014708855,0.006618031],"category_scores_gemma":[0.19142734,0.00035857785,0.00076208904,0.92149866,0.0007629203,0.0013587951,0.0052786535,0.001158069,0.00018072451],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000100950754,0.0013442521,0.06375939,0.00061255647,0.0000837021,0.0002205746,0.000637607,0.00024680689,0.0036903657,0.0037525299,0.35143647,0.5741148],"study_design_scores_gemma":[0.0043587787,0.0018786506,0.20685683,0.00073398324,0.00013493065,0.0007431741,0.00093674136,0.0039687413,0.0029563599,0.014246623,0.76117736,0.0020077955],"about_ca_topic_score_codex":0.000022791544,"about_ca_topic_score_gemma":0.0000038039282,"teacher_disagreement_score":0.572107,"about_ca_system_score_codex":0.0007086142,"about_ca_system_score_gemma":0.0011621525,"threshold_uncertainty_score":0.99988663},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4298394588","doi":"10.1007/s11192-022-04510-7","title":"Identifying science in the news: An assessment of the precision and recall of Altmetric.com news mention data","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria; Simon Fraser University","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"Computer science; Hyperlink; Recall; Information retrieval; Precision and recall; Data science; World Wide Web; Web page; Psychology","score_opus":0.7616169543367562,"score_gpt":0.6597713076679573,"score_spread":0.10184564666879892,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4298394588","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9860672,0.0011251702,0.006355869,0.00095172465,0.0017810067,0.00088316837,0.00018557416,0.000008784271,0.0026415267],"genre_scores_gemma":[0.99579805,0.00034495318,0.0033957348,0.000115652816,0.00002387667,0.000012611542,0.000010223336,0.00000858715,0.00029032238],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9509261,0.0020827895,0.0018104007,0.0018979338,0.04235292,0.00092985004],"domain_scores_gemma":[0.9816661,0.008195607,0.0014158279,0.005284611,0.0030581926,0.00037968747],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","open_science"],"category_scores_codex":[0.16984177,0.00017562123,0.00040827377,0.14689842,0.0011239502,0.0023540272,0.018554589,0.000058764163,0.0001821811],"category_scores_gemma":[0.07125832,0.00009843199,0.000113433125,0.69999915,0.0010668329,0.0022650086,0.01209999,0.0006025096,0.0000034254895],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001575604,0.0007958868,0.6743179,0.000014297672,0.000005897927,0.0000030820067,0.0008974765,0.0005168184,0.0035122053,0.0018481821,0.0033508716,0.3147216],"study_design_scores_gemma":[0.0006076434,0.00039994094,0.923167,0.000012490357,0.000013391507,0.000012459401,0.0060253874,0.05746128,0.00041469664,0.0045659267,0.007156623,0.00016316297],"about_ca_topic_score_codex":0.0007336157,"about_ca_topic_score_gemma":0.00011449173,"teacher_disagreement_score":0.5531007,"about_ca_system_score_codex":0.00036538567,"about_ca_system_score_gemma":0.0009598248,"threshold_uncertainty_score":0.9986816},"labels":[],"label_agreement":null},{"id":"W4298620868","doi":"10.1007/s11192-022-04511-6","title":"Different measures of international faculty and their impacts on global rankings","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"Higher Education Governance and Development","field":"Social Sciences","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Regional science; Political science; Econometrics; Library science; Sociology; Computer science; Economics","score_opus":0.07138703678488607,"score_gpt":0.37155696008756645,"score_spread":0.3001699233026804,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4298620868","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9816882,0.00017769296,0.00008922086,0.0014558401,0.0013654645,0.0000924712,0.000102511134,0.000015952834,0.0150126275],"genre_scores_gemma":[0.9988173,0.000065870336,0.00007874983,0.0004158726,0.000042794498,0.00000788982,0.000006751489,0.0000018840735,0.00056284526],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99817044,0.00005682182,0.00012060843,0.00015100223,0.0013387619,0.00016235329],"domain_scores_gemma":[0.9995204,0.00006578203,0.00010609695,0.000069323454,0.00014129477,0.000097067306],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009182845,0.000057129913,0.0000805179,0.0003709997,0.0003617736,0.0000670399,0.0002852448,0.000016523347,0.00022808499],"category_scores_gemma":[0.00050040527,0.00004532479,0.000030279869,0.00305835,0.000094727926,0.00007740767,0.00010559333,0.00005883704,0.0000035429343],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000032355973,0.00036507877,0.86413443,0.000005201538,0.000035701094,8.54761e-7,0.043971978,0.00004129264,0.000102524515,0.03222024,0.014862463,0.044227906],"study_design_scores_gemma":[0.00023145854,0.000040080842,0.6353461,0.0000045139186,0.0000018476854,4.7599863e-7,0.0031805355,0.000006668626,0.00016670478,0.00088764634,0.36006114,0.00007279615],"about_ca_topic_score_codex":0.00035170955,"about_ca_topic_score_gemma":0.000067014174,"teacher_disagreement_score":0.34519866,"about_ca_system_score_codex":0.00049188064,"about_ca_system_score_gemma":0.00020636804,"threshold_uncertainty_score":0.27825066},"labels":[],"label_agreement":null},{"id":"W4304145111","doi":"10.1007/s11192-022-04537-w","title":"Global impact or national accessibility? A paradox in China’s science","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bureau de Coopération Interuniversitaire; Université du Québec à Montréal; Université de Montréal","funders":"","keywords":"China; Publishing; Political science; Visibility; International community; Scientific communication; Scientific publishing; Regional science; Sociology; Library science; Geography; Law; Computer science","score_opus":0.6015316297703749,"score_gpt":0.6593529174440662,"score_spread":0.057821287673691324,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4304145111","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96778005,0.00044576125,0.0010270849,0.00066661445,0.001792452,0.00051073014,0.00043883096,0.00004926662,0.027289221],"genre_scores_gemma":[0.9970747,0.000023477674,0.0008977509,0.00021751187,0.00005873026,0.000051441988,0.0000068211075,0.000010046874,0.0016594699],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9300591,0.00053140003,0.0013379471,0.002253839,0.06387719,0.001940533],"domain_scores_gemma":[0.9861471,0.003779169,0.0006108497,0.0016521871,0.006363311,0.001447374],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.1443928,0.00027698264,0.00048168324,0.2889733,0.0016993027,0.006375298,0.011700628,0.00007597594,0.005159388],"category_scores_gemma":[0.21277158,0.0001858966,0.00026077838,0.8949101,0.0010199914,0.0026190467,0.0064890864,0.0005828656,0.00021566334],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.00011616002,0.00075682654,0.9146708,0.0000028617314,0.000005475771,0.000053153504,0.0002005591,0.005458027,0.0001011227,0.002572758,0.010197487,0.06586476],"study_design_scores_gemma":[0.0007811029,0.00037356195,0.9523097,0.0000018791225,0.0000016442417,0.00005336546,0.00032951173,0.024553644,0.000037926624,0.015159846,0.0061259572,0.00027185364],"about_ca_topic_score_codex":0.0005062098,"about_ca_topic_score_gemma":0.00008292424,"teacher_disagreement_score":0.6059368,"about_ca_system_score_codex":0.006556989,"about_ca_system_score_gemma":0.008900182,"threshold_uncertainty_score":0.99960035},"labels":[],"label_agreement":null},{"id":"W4306385889","doi":"10.1007/s11192-022-04542-z","title":"Evaluating the online impact of reporting guidelines for randomised trial reports and protocols: a cross-sectional web-based data analysis of CONSORT and SPIRIT initiatives","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital","funders":"","keywords":"Cross-sectional study; Consolidated Standards of Reporting Trials; Web application; Psychology; Medicine; Data science; Family medicine; Randomized controlled trial; Computer science; World Wide Web; Internal medicine","score_opus":0.9616850912565456,"score_gpt":0.7406745301888077,"score_spread":0.22101056106773787,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4306385889","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97914225,0.00044138185,0.005270359,0.0001405487,0.00017234524,0.013916329,0.0008588646,0.0000032921225,0.000054650787],"genre_scores_gemma":[0.97509795,0.0000048508796,0.02237772,0.0000629154,0.00007353396,0.002001016,0.00017943527,0.000008690706,0.00019385965],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9526206,0.0058999225,0.02761883,0.0017822615,0.0117691755,0.00030918527],"domain_scores_gemma":[0.86481845,0.038974416,0.074704215,0.006947408,0.014315553,0.00023992539],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.49236587,0.00019406497,0.0029372016,0.004509264,0.00047056586,0.00063744973,0.0011676918,0.000039915893,0.00088241015],"category_scores_gemma":[0.72637796,0.000089300156,0.0015468245,0.024448767,0.0002757463,0.0002682752,0.0006111838,0.00011992914,2.3123012e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002466868,0.00016525874,0.9799776,0.000069202644,0.0011401605,0.000002302271,0.00012727464,0.011178712,0.00030947878,0.000020089998,0.0011599244,0.0033830942],"study_design_scores_gemma":[0.015510489,0.0005408841,0.31029725,0.000013846289,0.0012002966,0.000014522778,0.00025253234,0.67072594,0.000012062886,0.0002869272,0.0010330064,0.00011225859],"about_ca_topic_score_codex":0.000036629124,"about_ca_topic_score_gemma":0.000016125303,"teacher_disagreement_score":0.6696804,"about_ca_system_score_codex":0.00004540324,"about_ca_system_score_gemma":0.0008191973,"threshold_uncertainty_score":0.99628705},"labels":[],"label_agreement":null},{"id":"W4309043512","doi":"10.1007/s11192-022-04546-9","title":"Researcher geographic mobility and publication productivity: an investigation into individual and institutional characteristics and the roles of academicians","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"Spatial and Panel Data Analysis","field":"Economics, Econometrics and Finance","cited_by":9,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Productivity; Institution; China; Ranking (information retrieval); Rank (graph theory); Association (psychology); Regional science; Sociology; Political science; Psychology; Social science; Economic growth; Economics; Computer science; Mathematics","score_opus":0.07782835240499963,"score_gpt":0.2776165607269312,"score_spread":0.19978820832193156,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4309043512","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99545413,0.0015698045,0.00022264391,0.0018336577,0.000057399015,0.0001919963,0.00060267665,0.000007873582,0.000059820457],"genre_scores_gemma":[0.9989508,0.00033535794,0.00023755209,0.0001329515,0.000031731855,0.000034390552,0.00024054469,0.0000037779878,0.00003288835],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99888515,0.00009833682,0.00031853147,0.0003720807,0.00019941822,0.00012649778],"domain_scores_gemma":[0.9992143,0.00012592794,0.00025079376,0.00021794306,0.00010328734,0.00008777951],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0062737693,0.00006795804,0.00018589139,0.0017486702,0.00046227287,0.00014730339,0.00021256763,0.000042268955,0.000027936157],"category_scores_gemma":[0.0017511306,0.0000613078,0.000020321975,0.0042529674,0.00089030777,0.00046699197,0.00035150853,0.0002162551,8.3842536e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019541883,0.000054718657,0.9275023,0.000026330988,0.000025003228,1.0487951e-7,0.0027506254,0.000010036493,0.00004183549,0.05201375,0.000028025748,0.017527772],"study_design_scores_gemma":[0.00031485464,0.000054519955,0.97166234,0.000001597373,0.000014432263,0.000002754416,0.00029488173,0.006248978,0.000016028629,0.019022215,0.0022871124,0.000080309714],"about_ca_topic_score_codex":0.0009666621,"about_ca_topic_score_gemma":0.00009915616,"teacher_disagreement_score":0.04416006,"about_ca_system_score_codex":0.000032738764,"about_ca_system_score_gemma":0.000034957626,"threshold_uncertainty_score":0.3555476},"labels":[],"label_agreement":null},{"id":"W4311046123","doi":"10.1007/s11192-022-04565-6","title":"Retraction of Predatory publishing in Scopus: evidence on cross-country differences lacks justification","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":12,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Scopus; Publishing; Bibliometrics; Library science; Data science; Political science; Regional science; Computer science; Geography; MEDLINE; Law","score_opus":0.7729591827362152,"score_gpt":0.6206454994995081,"score_spread":0.15231368323670713,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311046123","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98859817,0.0014403409,0.0007649863,0.0004990958,0.0038719669,0.00041493453,0.00016691037,0.000033950568,0.004209634],"genre_scores_gemma":[0.9975622,0.00019991281,0.0002498348,0.00013407263,0.000082184,0.000044307388,0.000012112484,0.000012527898,0.0017027968],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9628152,0.0007973398,0.0015642622,0.0014876176,0.032456428,0.00087914756],"domain_scores_gemma":[0.97594756,0.01508291,0.0012771401,0.0017986144,0.005421689,0.0004720582],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.09781933,0.00019763832,0.00040519002,0.21557449,0.00068646536,0.0067095743,0.005449575,0.00020236606,0.0011561909],"category_scores_gemma":[0.27711844,0.00016431129,0.00013692804,0.6009262,0.00039972298,0.004809255,0.0015294885,0.0013916155,0.0000779208],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008790491,0.00045448792,0.9269142,0.00002293511,0.0000046827026,0.0000074609156,0.00032113513,0.0016653131,0.0012292454,0.0013475316,0.00771228,0.060232814],"study_design_scores_gemma":[0.00038789544,0.0003916381,0.9791538,0.000032638396,0.0000036056194,0.0000030835872,0.0012208901,0.008783312,0.0012112206,0.0018374604,0.00676536,0.00020910651],"about_ca_topic_score_codex":0.00031571757,"about_ca_topic_score_gemma":0.000034142344,"teacher_disagreement_score":0.38535172,"about_ca_system_score_codex":0.0010373471,"about_ca_system_score_gemma":0.00064940355,"threshold_uncertainty_score":0.9999314},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4311696862","doi":"10.1007/s11192-022-04601-5","title":"Women and key positions in scientific collaboration networks: analyzing central scientists’ profiles in the artificial intelligence ecosystem through a gender lens","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"National Research Council Canada; Concordia University","funders":"National Research Council Canada","keywords":"Field (mathematics); Key (lock); Set (abstract data type); Social network analysis; Computer science; Knowledge management; Data science; Artificial intelligence; Sociology; Social media; World Wide Web","score_opus":0.48391299523641157,"score_gpt":0.5021268655735966,"score_spread":0.01821387033718508,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311696862","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97938555,0.0018397933,0.013095802,0.0011150598,0.0023724546,0.0010446557,0.00021344294,0.00002588451,0.0009073597],"genre_scores_gemma":[0.9984841,0.00011122205,0.0007736357,0.00015044484,0.00007637425,0.00017817356,0.000026678059,0.000013157159,0.00018622835],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.97856736,0.0016033637,0.0015151615,0.0018008965,0.01457199,0.0019412442],"domain_scores_gemma":[0.9926561,0.0036395784,0.00045228653,0.0010712587,0.0018354107,0.0003453441],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.07692394,0.00024699527,0.00040856548,0.11827082,0.0024363468,0.011614599,0.0035691017,0.000095240175,0.00033027705],"category_scores_gemma":[0.018034926,0.00018549395,0.000102293045,0.67391855,0.00058062305,0.0018834096,0.0015783552,0.00073306484,0.00003684617],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013544314,0.0025156573,0.62508416,0.00005329959,0.000032139385,0.00022951963,0.07579497,0.12303875,0.00097079837,0.09302519,0.008583428,0.07053662],"study_design_scores_gemma":[0.00069603784,0.0005560647,0.3459907,0.000025311949,0.000013403166,0.0000829178,0.09002556,0.47855607,0.00023332378,0.05700652,0.025885332,0.00092879485],"about_ca_topic_score_codex":0.00008312241,"about_ca_topic_score_gemma":0.0004091224,"teacher_disagreement_score":0.55564773,"about_ca_system_score_codex":0.0014887754,"about_ca_system_score_gemma":0.00073818193,"threshold_uncertainty_score":0.9988623},"labels":[],"label_agreement":null},{"id":"W4311809706","doi":"10.1007/s11192-022-04599-w","title":"Self-citation behavior within the health allied professions’ scientific sector in Italy: a bibliometric analysis","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":7,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Scopus; Citation; Multivariate statistics; Psychology; Multivariate analysis; Significant difference; Health science; Statistics; Actuarial science; Library science; Medicine; Medical education; Political science; Economics; Computer science; MEDLINE; Mathematics","score_opus":0.5549541081458416,"score_gpt":0.5712223387063445,"score_spread":0.016268230560502972,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311809706","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9847992,0.0029119267,0.0023108649,0.0025942747,0.004160847,0.0018948745,0.00025904676,0.00012399448,0.0009449645],"genre_scores_gemma":[0.99260366,0.00006214924,0.0026270084,0.00039850236,0.00005112787,0.00043619744,0.00007115578,0.000029669818,0.003720514],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9426335,0.0029895275,0.00266303,0.0027363512,0.04690688,0.0020706882],"domain_scores_gemma":[0.9779707,0.010319959,0.0018716918,0.0030895416,0.0054915403,0.0012565252],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.19466531,0.00037483676,0.0008532909,0.9414215,0.004405165,0.009076762,0.008245592,0.00012733281,0.0014159869],"category_scores_gemma":[0.0628017,0.00025056483,0.0005580157,0.9951064,0.00050542323,0.0012648377,0.003659904,0.0013522289,0.0002804402],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000036471698,0.0019473281,0.9294596,0.000015420428,0.00010125872,0.000032768137,0.0035091075,0.0026101484,0.00018938976,0.001902874,0.0124226175,0.04777307],"study_design_scores_gemma":[0.00079234224,0.00041612546,0.9358577,0.0000033481517,0.000073552634,0.00000996617,0.004490503,0.04143238,0.00005926352,0.0018011688,0.014677948,0.00038568073],"about_ca_topic_score_codex":0.00040185012,"about_ca_topic_score_gemma":0.0003534444,"teacher_disagreement_score":0.13186361,"about_ca_system_score_codex":0.0020795134,"about_ca_system_score_gemma":0.0023714209,"threshold_uncertainty_score":0.99999464},"labels":[],"label_agreement":null},{"id":"W4311812368","doi":"10.1007/s11192-022-04589-y","title":"Measuring h-index and scholarly productivity in academic dermatology in Canada","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"Diversity and Career in Medicine","field":"Social Sciences","cited_by":15,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University; University of Alberta; University of Saskatchewan; Université de Montréal; Université Laval; University of Toronto; University of British Columbia; McGill University Health Centre","funders":"National Institute of Diabetes and Digestive and Kidney Diseases","keywords":"Quartile; Index (typography); Productivity; Medicine; Logistic regression; Ranking (information retrieval); Demography; Family medicine; Psychology; Sociology; Internal medicine; Computer science; Economics","score_opus":0.09083608700775822,"score_gpt":0.3045341623938356,"score_spread":0.21369807538607735,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4311812368","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9923745,0.0007834026,0.0000057901416,0.0031207686,0.0006772896,0.00012419056,0.0000042362108,0.000008275081,0.0029015942],"genre_scores_gemma":[0.99937606,0.000060030736,0.000011965157,0.00030771908,0.000022571097,0.000005301905,6.7997155e-7,0.0000018701818,0.00021379637],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99726003,0.00025756483,0.00011360671,0.00023203912,0.001789328,0.00034741496],"domain_scores_gemma":[0.99957705,0.00012369838,0.000039301172,0.000081631886,0.000060027014,0.00011828082],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0046963464,0.0000447927,0.00010911196,0.0017220646,0.00039050967,0.000054738146,0.00048617486,0.00003366358,0.00003840628],"category_scores_gemma":[0.0025098918,0.000052896165,0.0000075401044,0.011570111,0.00014439388,0.00041214132,0.00032003855,0.00061869127,9.263806e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000032781181,0.000011405651,0.9861902,0.0000040531295,8.421702e-7,0.000044375614,0.0058528697,0.00007106032,0.000010989184,0.0004468987,0.00042011024,0.006943933],"study_design_scores_gemma":[0.0004625737,0.000010819138,0.91189486,0.0000072133575,0.0000026883756,0.000007531387,0.047197916,0.00007469455,0.000024852246,0.00039584804,0.0398091,0.000111904585],"about_ca_topic_score_codex":0.837354,"about_ca_topic_score_gemma":0.921206,"teacher_disagreement_score":0.08385198,"about_ca_system_score_codex":0.0016056855,"about_ca_system_score_gemma":0.0016686517,"threshold_uncertainty_score":0.55590546},"labels":[],"label_agreement":null},{"id":"W4313400493","doi":"10.1007/s11192-022-04627-9","title":"An artificial intelligence-based framework for data-driven categorization of computer scientists: a case study of world’s Top 10 computing departments","year":2022,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":11,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Regina","funders":"","keywords":"Computer science; Ranking (information retrieval); Context (archaeology); Cluster analysis; Metric (unit); Rank (graph theory); Artificial intelligence; Categorization; Rand index; Data mining; Index (typography); Artificial neural network; Machine learning; Information retrieval; Mathematics","score_opus":0.6906770264083343,"score_gpt":0.6059824450165449,"score_spread":0.08469458139178943,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313400493","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5279557,0.00004851952,0.4693736,0.00003158338,0.0013503247,0.0008449333,0.00034860446,0.0000185628,0.000028158422],"genre_scores_gemma":[0.93569124,0.0000013311768,0.06394059,0.000039287566,0.00010002341,0.000026294007,0.000117975076,0.000018889867,0.000064373744],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.97984564,0.00074494473,0.0019464039,0.0017304464,0.01491634,0.0008161958],"domain_scores_gemma":[0.981778,0.008041147,0.0013350894,0.0025694706,0.0057766484,0.0004995993],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.04189893,0.00023251964,0.00061195344,0.15208605,0.0011941541,0.0016282244,0.006528906,0.000072215014,0.00050893],"category_scores_gemma":[0.020868858,0.00020341035,0.00014776221,0.5139614,0.00033669392,0.0007391094,0.0030368762,0.00033075255,0.000012115927],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021351819,0.012161433,0.19205838,0.000055699304,0.00008048384,0.00023123216,0.0034107855,0.20950857,0.00024213974,0.0067056236,0.002706028,0.5726261],"study_design_scores_gemma":[0.00046563346,0.0025226306,0.0027829874,0.0000056620083,0.000030774285,0.000022222444,0.0060411007,0.9828791,0.00064165227,0.0034278124,0.00092927064,0.0002511447],"about_ca_topic_score_codex":0.0003263329,"about_ca_topic_score_gemma":0.00014337621,"teacher_disagreement_score":0.77337056,"about_ca_system_score_codex":0.00030491292,"about_ca_system_score_gemma":0.0006097962,"threshold_uncertainty_score":0.9994082},"labels":[],"label_agreement":null},{"id":"W4321611868","doi":"10.1007/s11192-023-04657-x","title":"Research mobility to the United States: a bibliometric analysis","year":2023,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"Canada Research Chairs","keywords":"China; Presidency; Political science; Politics; Cohort; Demographic economics; Medicine; Economics; Law","score_opus":0.8030659521882623,"score_gpt":0.6690166332988601,"score_spread":0.1340493188894022,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4321611868","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9701576,0.00049703417,0.013830964,0.008682446,0.0013718301,0.0010978669,0.00028493197,0.00025150858,0.003825785],"genre_scores_gemma":[0.985088,0.0007876728,0.000467117,0.0007191326,0.0001571317,0.00012455505,0.00009741748,0.000029945786,0.012529053],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.92013186,0.002395172,0.0017447245,0.0027264983,0.06943381,0.0035679506],"domain_scores_gemma":[0.885332,0.066013634,0.0004302859,0.006325702,0.038647898,0.0032504827],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.34870318,0.00033672655,0.00072068407,0.98879325,0.0022532472,0.013326163,0.013679617,0.00022967416,0.0011767289],"category_scores_gemma":[0.3806984,0.00019512812,0.0006172929,0.9995631,0.0008058267,0.00075782574,0.0060271337,0.0011527071,0.019717226],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000033390083,0.0003906965,0.2956139,0.000008855273,0.00018888144,0.000038679857,0.0010999434,0.045648254,0.00005268309,0.0008521748,0.50724,0.14883254],"study_design_scores_gemma":[0.00017655609,0.00018766867,0.5808696,0.0000021306591,0.000026603646,0.000001241232,0.002007274,0.08368258,0.00007061494,0.0020613857,0.33070713,0.00020720091],"about_ca_topic_score_codex":0.0016366033,"about_ca_topic_score_gemma":0.00018649384,"teacher_disagreement_score":0.2852557,"about_ca_system_score_codex":0.00067706814,"about_ca_system_score_gemma":0.0004983443,"threshold_uncertainty_score":0.9997363},"labels":[],"label_agreement":null},{"id":"W4366824053","doi":"10.1007/s11192-023-04696-4","title":"Scientific production of an oral implantology journal: a 5-year bibliometric study","year":2023,"lang":"en","type":"article","venue":"Scientometrics","topic":"Dental Implant Techniques and Outcomes","field":"Dentistry","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Bibliometrics; Observational study; Impact factor; Scope (computer science); Scopus; China; Library science; European union; Citation; Demography; Geography; Social science; Medicine; MEDLINE; Regional science; Political science; Sociology; Computer science; Pathology; Law; Archaeology","score_opus":0.11277669678380281,"score_gpt":0.4224531671033209,"score_spread":0.3096764703195181,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4366824053","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9855817,0.00007616972,0.0000656235,0.000022477196,0.013210723,0.00040118,0.000039882085,0.00032621875,0.00027600472],"genre_scores_gemma":[0.9949057,0.000025527494,0.000618318,0.000010469318,0.00010050778,0.000007650377,0.000035542813,0.000022380213,0.004273922],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9963589,0.00016110108,0.0005580693,0.00052887853,0.0018288228,0.00056426506],"domain_scores_gemma":[0.99804616,0.00011651403,0.0003096202,0.000600866,0.0006893909,0.00023742252],"candidate_categories":["bibliometrics"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.005954213,0.00015826618,0.00033541035,0.18378668,0.00038503465,0.00069886935,0.000819264,0.000092369526,0.0001139055],"category_scores_gemma":[0.0021781737,0.00013387357,0.0001140355,0.42940873,0.00019507765,0.0007619682,0.00028943384,0.00024477544,0.00060934434],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000058838345,0.0012976361,0.87960106,0.000056498382,0.000064431275,0.0010698583,0.00066457887,0.000023226874,0.0059263497,0.0001597445,0.05629547,0.0547823],"study_design_scores_gemma":[0.00070220954,0.00068403146,0.9842704,0.000010978381,0.00004825767,0.0028318877,0.0017700467,0.00010546738,0.0043398766,0.00034395515,0.00467641,0.00021650568],"about_ca_topic_score_codex":0.000057763606,"about_ca_topic_score_gemma":0.00008025549,"teacher_disagreement_score":0.24562205,"about_ca_system_score_codex":0.000087439585,"about_ca_system_score_gemma":0.000075853895,"threshold_uncertainty_score":0.8254644},"labels":[],"label_agreement":null},{"id":"W4378838751","doi":"10.1007/s11192-023-04729-y","title":"Assessing the quest of SMEs in pivoting for new technological ventures: comparing the patenting indexes of seven developed cities","year":2023,"lang":"en","type":"article","venue":"Scientometrics","topic":"Innovation and Knowledge Management","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canadian Institute for Advanced Research","funders":"Ministry of Science and Technology, Taiwan","keywords":"Commit; Business; Upgrade; Silicon valley; Industrial organization; Cluster (spacecraft); Competitive advantage; Marketing; Sorting; Entrepreneurship; Computer science; Database; Finance","score_opus":0.1530829460319046,"score_gpt":0.34833974741526863,"score_spread":0.19525680138336404,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4378838751","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9771605,0.000122215,0.0022789205,0.001098804,0.0005575853,0.00049067696,8.8997194e-7,0.000117101896,0.018173274],"genre_scores_gemma":[0.9989611,0.0000051792617,0.00042059837,0.00013666581,0.00012143284,0.00001494601,0.000008649998,0.000009499333,0.0003219272],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99864393,0.00001501053,0.00044626437,0.00017769446,0.000436766,0.00028035062],"domain_scores_gemma":[0.9987996,0.00031396825,0.00039722293,0.00017063602,0.00031371377,0.0000048555576],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004134758,0.000099175166,0.00018257336,0.0027038858,0.00026125944,0.0003512331,0.00054550165,0.000045596236,0.0000051574343],"category_scores_gemma":[0.0026948983,0.000060641858,0.000052147756,0.019341856,0.00014701669,0.00034026307,0.0005171952,0.00012550662,0.00000723328],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000778744,0.000062314044,0.61164194,0.0004801083,0.000028100987,0.0000013114358,0.00079609046,0.00046507968,0.00037520862,0.3373769,0.0015474847,0.047217663],"study_design_scores_gemma":[0.0014970622,0.000020349842,0.8334113,0.0006246877,0.000069679016,6.748012e-7,0.041223735,0.060728457,0.0014746678,0.024966061,0.03550832,0.00047504032],"about_ca_topic_score_codex":0.0000697396,"about_ca_topic_score_gemma":0.00006270277,"teacher_disagreement_score":0.31241083,"about_ca_system_score_codex":0.0000316102,"about_ca_system_score_gemma":0.000038825227,"threshold_uncertainty_score":0.92931205},"labels":[],"label_agreement":null},{"id":"W4379015792","doi":"10.1007/s11192-023-04741-2","title":"Letter to the editor: Laxdal (2023) “The sex gap in sports and exercise medicine research: who does research on females?”","year":2023,"lang":"en","type":"letter","venue":"Scientometrics","topic":"Sex and Gender in Healthcare","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Natural Sciences and Engineering Research Council of Canada; Ontario Ministry of Health and Long-Term Care","keywords":"Sports medicine; Sports science; Psychology; Medical education; Medicine; Physical therapy; Physiology","score_opus":0.4665820957653342,"score_gpt":0.5100070099196117,"score_spread":0.04342491415427746,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379015792","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013878934,0.0020314276,0.0000012084269,0.964552,0.013629559,0.002653979,0.0000662501,0.00006704397,0.0031195881],"genre_scores_gemma":[0.015372968,0.0033639555,0.000039693157,0.6741864,0.15915804,0.0005713466,0.00019887496,0.0002097801,0.14689894],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9825613,0.0011704391,0.00071776984,0.0016481774,0.0117426375,0.0021596712],"domain_scores_gemma":[0.9892594,0.006599749,0.00010464849,0.002270161,0.0012543559,0.00051167206],"candidate_categories":["metaresearch","bibliometrics","research_integrity"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.038173437,0.0004041941,0.0007972761,0.01687226,0.0008167193,0.00022007077,0.0012763651,0.0009285813,0.00018741569],"category_scores_gemma":[0.008950209,0.00019294511,0.00009726056,0.039260965,0.0013513466,0.00006289109,0.00084254274,0.009406091,0.0004139298],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000062746374,0.000030889772,0.007036155,0.0005987459,0.000012557058,0.0010997073,0.004168635,0.0000020097718,0.0000024384913,0.000010755193,0.9742101,0.0127652595],"study_design_scores_gemma":[0.00038164394,0.00049276085,0.022170404,0.0013894839,0.000020790512,0.000018606826,0.0031918453,0.000021834781,0.00002153876,0.0003398039,0.9717469,0.0002043879],"about_ca_topic_score_codex":0.0012060336,"about_ca_topic_score_gemma":0.00026504687,"teacher_disagreement_score":0.2903656,"about_ca_system_score_codex":0.0006027049,"about_ca_system_score_gemma":0.00050395005,"threshold_uncertainty_score":0.9993978},"labels":[],"label_agreement":null},{"id":"W4381053087","doi":"10.1007/s11192-023-04758-7","title":"The marginalizing effect of journal submission fees in Accounting and Finance","year":2023,"lang":"en","type":"article","venue":"Scientometrics","topic":"Auditing, Earnings Management, Governance","field":"Business, Management and Accounting","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Nipissing University; Saint Mary's University; Memorial University of Newfoundland","funders":"","keywords":"Publishing; Accounting; Elite; Unintended consequences; Business; Quality (philosophy); Principal (computer security); Finance; Economics; Political science; Law","score_opus":0.012715760344684279,"score_gpt":0.255071190898144,"score_spread":0.24235543055345973,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4381053087","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9950468,0.0005925637,0.0010986855,0.00042370203,0.0005671092,0.00015597705,4.3358176e-7,0.00004376631,0.0020709804],"genre_scores_gemma":[0.9983019,0.00056215416,0.000057021312,0.00007461601,0.00023227352,0.00000494217,0.0000012213543,0.000015547423,0.0007502909],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9981573,0.000023226452,0.0003030511,0.00023457644,0.00084338535,0.00043841053],"domain_scores_gemma":[0.9948634,0.00044036447,0.004382385,0.00016269251,0.00014143495,0.000009728679],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0063254973,0.00012667026,0.00017453176,0.0022385933,0.00049624714,0.00068821665,0.00044530065,0.000035276826,0.000009063279],"category_scores_gemma":[0.028024904,0.000089702226,0.000045923483,0.014879984,0.000103817285,0.0008640023,0.00051932887,0.0002313203,0.000054906555],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019203508,0.000009387876,0.6219653,0.00017250278,0.0000057756947,0.000018811554,0.00006596251,0.00045230065,0.00022623343,0.0019164904,0.0032549775,0.37189308],"study_design_scores_gemma":[0.0007087073,0.000023282184,0.7758186,0.00023095269,0.000017012677,0.0000031025859,0.00016694568,0.009697028,0.00015292648,0.0008231438,0.21218304,0.00017525551],"about_ca_topic_score_codex":0.000071358685,"about_ca_topic_score_gemma":0.000013601524,"teacher_disagreement_score":0.3717178,"about_ca_system_score_codex":0.00004918452,"about_ca_system_score_gemma":0.000013908776,"threshold_uncertainty_score":0.98016244},"labels":[],"label_agreement":null},{"id":"W4381682726","doi":"10.1007/s11192-023-04768-5","title":"On the lack of women researchers in the Middle East and North Africa","year":2023,"lang":"en","type":"article","venue":"Scientometrics","topic":"Diversity and Career in Medicine","field":"Social Sciences","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Middle East; Seniority; Productivity; Political science; Gender equality; Publication; Geography; Economic growth; Development economics; Demographic economics; Sociology; Gender studies; Economics; Law","score_opus":0.5835821107292325,"score_gpt":0.3947975949750255,"score_spread":0.18878451575420702,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4381682726","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95263225,0.000060444036,7.7917167e-7,0.004384831,0.00010334799,0.00011820382,0.0000059824483,0.000009599151,0.042684585],"genre_scores_gemma":[0.9976373,0.00013087924,0.0000035490118,0.0002982838,0.000023163086,0.0000040473196,7.0539346e-7,0.0000013403044,0.0019007829],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99739385,0.00021645689,0.00006463558,0.000106224266,0.0018674532,0.00035139165],"domain_scores_gemma":[0.99865735,0.0010061286,0.000025679165,0.00013582257,0.00008592686,0.00008908182],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.012807353,0.00003170125,0.000060175742,0.00074481714,0.00029515775,0.0000814048,0.0007111531,0.000024390136,0.000054304932],"category_scores_gemma":[0.005584139,0.000017927156,0.000015310967,0.022112831,0.0007131898,0.000053927226,0.000115546856,0.00013119276,0.00004170048],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001350004,0.000062093764,0.046971854,0.00001523658,0.0000063343928,0.000010026639,0.88085663,0.000018516968,0.0000035644857,0.032771766,0.033517007,0.0057534506],"study_design_scores_gemma":[0.00020599335,0.00012017297,0.09812135,0.000015493313,0.0000029493776,8.282647e-8,0.8212192,0.000046293975,0.0000014766723,0.0027417387,0.07747403,0.000051185925],"about_ca_topic_score_codex":0.00024717464,"about_ca_topic_score_gemma":0.00025261132,"teacher_disagreement_score":0.05963742,"about_ca_system_score_codex":0.000098025805,"about_ca_system_score_gemma":0.00007849692,"threshold_uncertainty_score":0.9986726},"labels":[],"label_agreement":null},{"id":"W4386402131","doi":"10.1007/s11192-023-04818-y","title":"The Matthew Effect in China’s social sciences and humanities research: a comparative analysis of CSSCI and SSCI","year":2023,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":7,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Planning Office of Philosophy and Social Science","keywords":"Elite; China; Humanities; Social science; Citation analysis; Political science; Sociology; Citation; Philosophy; Law; Politics","score_opus":0.8933759979167228,"score_gpt":0.6993538977340414,"score_spread":0.1940221001826814,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386402131","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98807365,0.0024008544,0.000027933145,0.0010264065,0.00023022701,0.00035622797,0.000050694627,0.000020692538,0.0078133205],"genre_scores_gemma":[0.9977904,0.0008165282,0.00006689679,0.000028725173,0.000025472054,0.000019708847,0.000002392713,0.0000048460092,0.0012450548],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.97931486,0.0014578373,0.0008258275,0.0010195652,0.016147384,0.0012345301],"domain_scores_gemma":[0.9631076,0.033782538,0.00029161482,0.0004913129,0.0020990123,0.00022795447],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics","sts"],"category_scores_codex":[0.16835909,0.00017386013,0.0006612338,0.34124807,0.002451258,0.0050068367,0.0025928724,0.00009899321,0.000043391454],"category_scores_gemma":[0.054379962,0.000099471035,0.00014606814,0.80526614,0.005054522,0.00059656997,0.0021120864,0.00042851074,0.000044365595],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005090946,0.00010226754,0.87923145,0.000024762016,0.0001277269,0.000012168233,0.008029198,0.0001813181,0.00031745507,0.023858776,0.010844247,0.077219695],"study_design_scores_gemma":[0.00029308698,0.00031634653,0.95793754,0.0000064820692,0.000022880628,7.830575e-7,0.0045360043,0.025352975,0.00013026592,0.007529927,0.0037489533,0.00012477401],"about_ca_topic_score_codex":0.00046130814,"about_ca_topic_score_gemma":0.00046325274,"teacher_disagreement_score":0.46401805,"about_ca_system_score_codex":0.0001129867,"about_ca_system_score_gemma":0.00020266067,"threshold_uncertainty_score":0.9988474},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4387014740","doi":"10.1007/s11192-023-04837-9","title":"An analysis of international mobility and research productivity in computer science","year":2023,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Regional science; China; Scopus; Economic geography; Demographic economics; Political science; Geography; Economics; Law","score_opus":0.707261089982269,"score_gpt":0.6701808773690084,"score_spread":0.037080212613260555,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387014740","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9943623,0.000078540725,0.00355722,0.0003848863,0.0007339784,0.00025723924,0.00005476693,0.00002749727,0.0005436039],"genre_scores_gemma":[0.99813175,0.00009510789,0.0015479162,0.000011316608,0.000044815613,0.0000082691395,0.0000065723652,0.000004858407,0.00014939543],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9659516,0.0005454489,0.0008441667,0.0018661858,0.029756557,0.0010360458],"domain_scores_gemma":[0.9801772,0.005865741,0.00019778052,0.001525916,0.011632155,0.000601194],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.2598549,0.00011562317,0.00041521498,0.6201922,0.0003628971,0.0026918354,0.004839997,0.0000811031,0.00011104836],"category_scores_gemma":[0.11829817,0.00008886992,0.00009881279,0.91921633,0.0023682306,0.0018393276,0.0025145065,0.00036501465,0.00005810061],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009395817,0.00029837713,0.8502404,0.0000028803445,0.000013361817,0.0000046341247,0.00044300497,0.003165475,0.0020081324,0.0005085403,0.00014112827,0.14316466],"study_design_scores_gemma":[0.00011186018,0.00008748834,0.65527844,0.0000015478528,0.0000039105526,4.505742e-7,0.00018844593,0.34199014,0.0007556351,0.0012678823,0.00025025877,0.000063908876],"about_ca_topic_score_codex":0.0003420087,"about_ca_topic_score_gemma":0.00010272436,"teacher_disagreement_score":0.33882466,"about_ca_system_score_codex":0.00035268656,"about_ca_system_score_gemma":0.00047275695,"threshold_uncertainty_score":0.99834347},"labels":[],"label_agreement":null},{"id":"W4387937804","doi":"10.1007/s11192-023-04846-8","title":"Interdisciplinarity affects the technological impact of scientific research","year":2023,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":17,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"","keywords":"Variety (cybernetics); Scope (computer science); Balance (ability); Term (time); Technological change; Scientific literature; Computer science; Psychology; Physics; Artificial intelligence; Biology","score_opus":0.8337135587900643,"score_gpt":0.7017474832136088,"score_spread":0.13196607557645545,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387937804","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9867854,0.00039416045,0.0007096247,0.0016842579,0.0012752577,0.0006159081,0.00008252689,0.00013788593,0.008314973],"genre_scores_gemma":[0.9934497,0.000066512664,0.00017004227,0.000008185309,0.00005915931,0.000027771594,0.000007639429,0.000014834086,0.006196159],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9527063,0.0014190105,0.0010379587,0.0017600128,0.040771548,0.0023052157],"domain_scores_gemma":[0.9486988,0.03315534,0.00043448378,0.0036091383,0.013287753,0.00081448787],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","sts","open_science"],"category_scores_codex":[0.2891755,0.0002291591,0.0004907206,0.36658043,0.0020940544,0.0065724277,0.011121862,0.00026756973,0.0004950324],"category_scores_gemma":[0.3148681,0.00011146743,0.00058400154,0.8970405,0.005796964,0.0006987143,0.008312862,0.0011635971,0.002714852],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000040872503,0.0004709522,0.4092692,0.000014735343,0.000039664214,0.000041519215,0.0006525233,0.0004404039,0.008068422,0.009028022,0.26109558,0.31083807],"study_design_scores_gemma":[0.00035433663,0.00062199193,0.9154066,0.000014331066,0.000004763411,0.000010986892,0.0012663421,0.016744405,0.002787363,0.048007637,0.01455087,0.00023038022],"about_ca_topic_score_codex":0.000058802612,"about_ca_topic_score_gemma":0.000015150567,"teacher_disagreement_score":0.53046006,"about_ca_system_score_codex":0.000500513,"about_ca_system_score_gemma":0.00079459377,"threshold_uncertainty_score":0.9997077},"labels":[],"label_agreement":null},{"id":"W4388106222","doi":"10.1007/s11192-023-04854-8","title":"Publish and flourish: investigating publication requirements for PhD students in China","year":2023,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":26,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"National Social Science Fund of China","keywords":"China; Publication; Index (typography); Citation; Science Citation Index; Library science; Sample (material); Citation index; Order (exchange); Publishing; Political science; Computer science; World Wide Web; Business; Law","score_opus":0.7616037686334712,"score_gpt":0.6142861649481008,"score_spread":0.1473176036853704,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388106222","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98978925,0.00024303145,0.0012130716,0.0030925188,0.00120353,0.0009063801,0.00007433125,0.000102951584,0.0033749586],"genre_scores_gemma":[0.9861603,0.00012530477,0.00264154,0.0002848763,0.00010488092,0.000077887256,0.00005327011,0.000023597488,0.010528338],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.97883165,0.0002242631,0.0011757296,0.0014338029,0.01704975,0.0012847835],"domain_scores_gemma":[0.9879489,0.004820972,0.00047884253,0.00095948146,0.0048569287,0.00093488424],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.09208526,0.0002011515,0.0003639384,0.26271707,0.00047587432,0.011739424,0.0037273553,0.0001577629,0.00008848673],"category_scores_gemma":[0.32471627,0.00016441196,0.000100846126,0.6774588,0.0002417299,0.0026092643,0.0020070826,0.0002819982,0.00024410627],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004543041,0.000110589506,0.8474698,0.0000133753,0.000006148119,0.0000024578753,0.0003682068,0.000025876503,0.00023867274,0.00072903006,0.057898887,0.093132384],"study_design_scores_gemma":[0.0010357741,0.0000939184,0.9385756,0.000012735093,0.000003228749,0.0000012312928,0.00034574856,0.022264011,0.00017976765,0.010142177,0.027127868,0.00021788769],"about_ca_topic_score_codex":0.000095900294,"about_ca_topic_score_gemma":0.00003293428,"teacher_disagreement_score":0.41474175,"about_ca_system_score_codex":0.00027173758,"about_ca_system_score_gemma":0.00021876216,"threshold_uncertainty_score":0.9892865},"labels":[],"label_agreement":null},{"id":"W4389686822","doi":"10.1007/s11192-023-04876-2","title":"The oligopoly of open access publishing","year":2023,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":33,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bureau de Coopération Interuniversitaire; Université du Québec à Montréal; Université de Montréal","funders":"National Natural Science Foundation of China","keywords":"Publishing; Oligopoly; Open access publishing; Business; Computer science; Political science; Library science; Law","score_opus":0.9066864170215745,"score_gpt":0.7241775974229555,"score_spread":0.18250881959861898,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4389686822","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7497893,0.0024032756,0.0032381653,0.019262308,0.008582635,0.0014711408,0.00021479266,0.00020250896,0.21483585],"genre_scores_gemma":[0.97830695,0.00081431237,0.00033218096,0.00020694082,0.00010084205,0.000025624311,0.0000073275023,0.00001793415,0.020187883],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9669818,0.00041244665,0.0013644554,0.0011218992,0.028655633,0.0014637352],"domain_scores_gemma":[0.95562476,0.029013252,0.0008841434,0.0026327188,0.010955939,0.0008891784],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","open_science"],"category_scores_codex":[0.1524852,0.00016712047,0.0003863936,0.2333492,0.0012678143,0.12072442,0.0473551,0.00012539931,0.00029290468],"category_scores_gemma":[0.41767758,0.00009280769,0.00019840438,0.87874407,0.0006257634,0.010709695,0.024871178,0.00039567484,0.00084839825],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013767795,0.00006238958,0.13712224,0.0000040808736,0.000013315817,0.000006398117,0.00012559544,0.00009296977,0.00023998287,0.0143311685,0.44987464,0.39811343],"study_design_scores_gemma":[0.00048141307,0.000085784435,0.45964444,0.000007703983,0.000004261295,0.0000028797638,0.00074663264,0.007529027,0.0019171395,0.0464995,0.48287025,0.00021098809],"about_ca_topic_score_codex":0.0004285221,"about_ca_topic_score_gemma":0.000052505944,"teacher_disagreement_score":0.64539486,"about_ca_system_score_codex":0.00014828765,"about_ca_system_score_gemma":0.000686739,"threshold_uncertainty_score":0.99992955},"labels":[],"label_agreement":null},{"id":"W4390694881","doi":"10.1007/s11192-023-04904-1","title":"Country names in journal titles: shaping researchers’ perception of journals quality","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Charles Sturt University","keywords":"Audience measurement; Publishing; Mainstream; Quality (philosophy); Internationalization; Perception; Diversity (politics); Scope (computer science); Political science; Public relations; Sociology; Library science; Social science; Psychology; Law; Business","score_opus":0.8876352380282057,"score_gpt":0.7020697218535584,"score_spread":0.18556551617464723,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390694881","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95221394,0.020578688,0.008039581,0.0011783344,0.0025479198,0.00024360736,0.00007097687,0.00003327197,0.01509367],"genre_scores_gemma":[0.99225575,0.0029277292,0.0013864174,0.00006694272,0.0001937835,0.000002918113,0.00000215824,0.000015967044,0.0031483644],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.96851635,0.0008523004,0.0018911517,0.0008389259,0.02685336,0.0010479033],"domain_scores_gemma":[0.9811025,0.009849984,0.00047399415,0.00081143546,0.006869085,0.0008929824],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.18481858,0.00017276518,0.00047672898,0.29310152,0.0002688633,0.008265501,0.0031185395,0.00019779355,0.0028484464],"category_scores_gemma":[0.15918395,0.000120467055,0.0002687126,0.5876537,0.0004422907,0.0017904092,0.00069507695,0.0011598056,0.00047775457],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000033222528,0.0002691926,0.32668442,0.000083949555,0.00002998971,0.00011104711,0.0022017413,0.0002084611,0.009192443,0.0020208957,0.030971346,0.6281933],"study_design_scores_gemma":[0.0007006803,0.0002830818,0.8733782,0.00029234096,0.0000090080475,0.00009832066,0.008077649,0.020268546,0.00042541072,0.018496,0.07756481,0.00040590312],"about_ca_topic_score_codex":0.00017149442,"about_ca_topic_score_gemma":0.000027062393,"teacher_disagreement_score":0.6277874,"about_ca_system_score_codex":0.000811816,"about_ca_system_score_gemma":0.0010500075,"threshold_uncertainty_score":0.9980631},"labels":[],"label_agreement":null},{"id":"W4391175495","doi":"10.1007/s11192-024-04930-7","title":"Exploring relationships among eminent psychologists using co-occurrence analysis","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"Mental Health Research Topics","field":"Psychology","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Brock University","funders":"","keywords":"Psychology; Sociology; Data science; Computer science","score_opus":0.7976071504131107,"score_gpt":0.5730943597756223,"score_spread":0.2245127906374884,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4391175495","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97463334,0.0015079525,0.009898913,0.00014086674,0.0033893678,0.0002958656,0.00013734693,0.00018116664,0.009815179],"genre_scores_gemma":[0.9968381,0.00009635101,0.001244376,0.000039226758,0.00013172405,0.0000604126,0.00007643272,0.000013340432,0.0015000475],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9966876,0.00029373335,0.000432894,0.00072139635,0.0011359481,0.000728425],"domain_scores_gemma":[0.99809766,0.0006726662,0.00007980044,0.00061340065,0.00012129664,0.0004151496],"candidate_categories":["bibliometrics","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.004063511,0.00015137512,0.00020993488,0.010053158,0.00032637076,0.0002589217,0.00043688747,0.000100307334,0.0010217414],"category_scores_gemma":[0.00092117564,0.00014300652,0.00014622704,0.04536474,0.00023003835,0.00043715828,0.00009798823,0.00066334475,0.0006957272],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000137270235,0.000188624,0.95384127,0.00013291002,0.00025234904,0.00018283584,0.0031880417,0.0002516458,0.000033292243,0.0048031057,0.003725286,0.033386923],"study_design_scores_gemma":[0.00018792531,0.000073138435,0.9709216,0.000054268243,0.00014526774,0.000013694206,0.0007894354,0.011449455,0.000057188357,0.00014722385,0.015899353,0.00026148843],"about_ca_topic_score_codex":0.0002012554,"about_ca_topic_score_gemma":0.000023863551,"teacher_disagreement_score":0.035311583,"about_ca_system_score_codex":0.00045954873,"about_ca_system_score_gemma":0.00008063863,"threshold_uncertainty_score":0.99989146},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4392109487","doi":"10.1007/s11192-024-04948-x","title":"Regional disparities in Web of Science and Scopus journal coverage","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":145,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Scopus; Web of science; Scientometrics; Data science; Regional science; Bibliometrics; World Wide Web; Computer science; Library science; Information retrieval; Political science; Sociology; MEDLINE","score_opus":0.5043046486588112,"score_gpt":0.5676367040364447,"score_spread":0.06333205537763353,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392109487","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9719712,0.015067001,0.0012932578,0.0013977997,0.0023335377,0.00017038498,0.000036853522,0.000020855432,0.007709087],"genre_scores_gemma":[0.9956211,0.0023825346,0.00065829355,0.00010093946,0.00007221149,0.0000027056094,5.1781035e-7,0.000008807585,0.0011528878],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9708697,0.00014674026,0.0009705138,0.0009175018,0.026178908,0.0009166171],"domain_scores_gemma":[0.9857115,0.005847993,0.0002210204,0.0005897503,0.0068010613,0.00082868844],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.097425595,0.00015729885,0.0003509867,0.38178134,0.00041558163,0.008781928,0.0029635502,0.00009248902,0.00015958688],"category_scores_gemma":[0.1065742,0.00010930417,0.00010431078,0.7053051,0.0030285027,0.0024069815,0.0010997886,0.0005469694,0.00006264549],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003005771,0.00024672673,0.43028668,0.00006868945,0.000013493342,0.00017750062,0.0006907793,0.0001280257,0.006268474,0.035109,0.033124793,0.49385577],"study_design_scores_gemma":[0.0011584149,0.000528455,0.7285021,0.00021279513,0.000011545889,0.00040895274,0.00095600705,0.07250423,0.0021664097,0.05443011,0.13854542,0.00057556224],"about_ca_topic_score_codex":0.00005421605,"about_ca_topic_score_gemma":0.000014931583,"teacher_disagreement_score":0.49328023,"about_ca_system_score_codex":0.00045302146,"about_ca_system_score_gemma":0.0032279708,"threshold_uncertainty_score":0.9996847},"labels":[],"label_agreement":null},{"id":"W4394768444","doi":"10.1007/s11192-024-04981-w","title":"Is interdisciplinarity more likely to produce novel or disruptive research?","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":17,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Social Science Fund of China","keywords":"Novelty; Scopus; Computer science; Data science; Quality (philosophy); Citation; Creativity; Psychology; Epistemology; Political science; Social psychology; MEDLINE; World Wide Web","score_opus":0.89847556445197,"score_gpt":0.7359858661900647,"score_spread":0.1624896982619053,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4394768444","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7864584,0.0041509597,0.113623306,0.053777814,0.011330716,0.0030846884,0.0011825678,0.00049649767,0.025895108],"genre_scores_gemma":[0.92848015,0.000105886276,0.0052921106,0.00060701737,0.0004593851,0.00007905862,0.0000062897993,0.000049973307,0.06492012],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9515532,0.00034565243,0.001255796,0.0034354443,0.04102551,0.0023843846],"domain_scores_gemma":[0.9627688,0.016287642,0.00016072259,0.0028902027,0.01527341,0.0026192549],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.11287813,0.00038807414,0.00057937205,0.38840893,0.0011103536,0.018075868,0.007795368,0.00025483654,0.0016488236],"category_scores_gemma":[0.22622605,0.00023825964,0.0003503373,0.8426282,0.0008968755,0.0018212951,0.0076874127,0.0014015146,0.00772296],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015043531,0.00080072996,0.015969815,0.00007373959,0.00005597473,0.00020734915,0.011031082,0.00007333726,0.0037540898,0.0033578305,0.56636477,0.39816087],"study_design_scores_gemma":[0.0005463494,0.0014245929,0.10843466,0.00016273528,0.000017131479,0.00007058738,0.005119682,0.024558984,0.009207766,0.0081401635,0.8414085,0.0009088741],"about_ca_topic_score_codex":0.000119358214,"about_ca_topic_score_gemma":0.000041417497,"teacher_disagreement_score":0.45421925,"about_ca_system_score_codex":0.0010590259,"about_ca_system_score_gemma":0.0014067339,"threshold_uncertainty_score":0.9992638},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4396237870","doi":"10.1007/s11192-024-05017-z","title":"CFMf topic-model: comparison with LDA and Top2Vec","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":10,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Social Sciences and Humanities Research Council of Canada; Fonds de Recherche du Québec-Société et Culture; Canada Research Chairs","keywords":"Computer science; Topic model; Data science; Information retrieval","score_opus":0.08423497244561039,"score_gpt":0.45627159766623665,"score_spread":0.37203662522062625,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4396237870","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.47743636,0.00519275,0.35261887,0.004740103,0.0010165056,0.00023299077,0.0000078155035,0.00029760587,0.15845698],"genre_scores_gemma":[0.97278535,0.000053758984,0.019266767,0.00008880592,0.00011581141,0.0000026590858,0.0000020427965,0.000004370612,0.0076804063],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99863535,0.00006568899,0.00011341393,0.00024109121,0.00074927876,0.00019514769],"domain_scores_gemma":[0.99927115,0.00037041778,0.000023685385,0.000076871605,0.00012318254,0.0001346823],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0014654126,0.00006142944,0.00011136589,0.0010454407,0.00029196448,0.00042980697,0.0001420467,0.000034869132,0.000069614245],"category_scores_gemma":[0.0002837087,0.000046751673,0.000035239995,0.008901986,0.0001891716,0.00017888051,0.000045767392,0.0000870036,0.000018975235],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000585526,0.00006416789,0.039355103,0.000036970647,0.000048760427,0.0000112644375,0.015625361,0.0030580387,0.000023241513,0.41864195,0.0028870916,0.5202422],"study_design_scores_gemma":[0.00022055978,0.000095594325,0.022076758,0.00004342265,0.00009523424,0.000002982934,0.0022024696,0.4350793,0.00006114396,0.03086521,0.5088897,0.00036765245],"about_ca_topic_score_codex":0.00022054222,"about_ca_topic_score_gemma":0.0003273469,"teacher_disagreement_score":0.5198746,"about_ca_system_score_codex":0.00006472406,"about_ca_system_score_gemma":0.00015748545,"threshold_uncertainty_score":0.42771092},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4399053656","doi":"10.1007/s11192-024-05047-7","title":"Visualizing hidden communities of interest: A case-study analysis of topic-based social networks in astrobiology","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"Climate Change Communication and Perception","field":"Social Sciences","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"Social Sciences and Humanities Research Council of Canada; Fonds de Recherche du Québec-Société et Culture; Canada Research Chairs","keywords":"Social network analysis; Data science; Computer science; Astrobiology; World Wide Web; Social media; Biology","score_opus":0.733346814239927,"score_gpt":0.5580407318549871,"score_spread":0.17530608238493983,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399053656","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.997601,0.00022577074,0.00042995892,0.00016185365,0.00021171916,0.0001314938,0.00001642952,0.000024523964,0.0011972688],"genre_scores_gemma":[0.9996763,0.000110147026,0.00008327678,0.000025412759,0.000030742263,0.0000068014424,0.000028297052,0.0000034426594,0.000035594996],"study_design_codex":"observational","study_design_gemma":"qualitative","domain_scores_codex":[0.9986774,0.0005752897,0.00029465344,0.000100182726,0.00019144613,0.0001610844],"domain_scores_gemma":[0.99889535,0.00064938806,0.00009115787,0.00018237138,0.00015019623,0.000031568925],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024677576,0.00005637458,0.000220858,0.0043037925,0.00018150794,0.000062940686,0.00032091644,0.00007420303,0.0002867536],"category_scores_gemma":[0.00016430025,0.000056294528,0.00010374864,0.018229883,0.00031378827,0.00007563196,0.00011037071,0.00013723242,0.0000012329192],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011834899,0.00043700807,0.6215482,0.000044853612,0.00014664045,0.000020112522,0.33374733,0.0003374179,0.000059846458,0.0041453033,0.00006297537,0.039438475],"study_design_scores_gemma":[0.00045903385,0.0002461026,0.18335621,0.00007705518,0.00038405936,0.0000017957053,0.7496889,0.06318849,0.000009835727,0.000071099676,0.0022984452,0.00021897239],"about_ca_topic_score_codex":0.012509514,"about_ca_topic_score_gemma":0.14041363,"teacher_disagreement_score":0.438192,"about_ca_system_score_codex":0.00012931487,"about_ca_system_score_gemma":0.000057555713,"threshold_uncertainty_score":0.9940663},"labels":[],"label_agreement":null},{"id":"W4400867599","doi":"10.1007/s11192-024-05078-0","title":"Navigating geopolitical storms: assessing the robustness of Canada’s 5G research network in the wake of the Huawei conflict","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Geopolitics; Wake; Storm; Robustness (evolution); Political science; Regional science; Computer science; Sociology; Meteorology; Geography; Politics; Engineering; Law; Aerospace engineering","score_opus":0.07914980132810835,"score_gpt":0.4294486770892403,"score_spread":0.35029887576113194,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400867599","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9796622,0.0010562342,0.0046816347,0.0014689607,0.00047440088,0.00040495757,0.000019911908,0.000017468868,0.012214234],"genre_scores_gemma":[0.99927074,0.0000025196946,0.00022305503,0.000029854727,0.00033240768,0.000014537394,0.0000037448444,0.000009990279,0.00011315556],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9966992,0.00047243695,0.00039669752,0.00023264225,0.001582244,0.0006167621],"domain_scores_gemma":[0.9965236,0.0024228303,0.000104804436,0.0005795568,0.00032044598,0.000048724345],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007954184,0.0001040139,0.00019689192,0.00026375972,0.000422625,0.0002464863,0.0011418901,0.00002987918,0.000044545523],"category_scores_gemma":[0.00026600447,0.000051674037,0.00011787037,0.017730646,0.00043248813,0.000087315755,0.00029566197,0.00085527205,3.908463e-7],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000053309946,0.00026039162,0.33150715,0.00018248837,0.0002183579,0.0000072970097,0.004944264,0.05564729,0.00022884787,0.4824545,0.049424566,0.07511951],"study_design_scores_gemma":[0.00043612777,0.00010651613,0.2943071,0.0019886321,0.00021629508,0.0000081686985,0.022207232,0.5745411,0.002799066,0.03144473,0.07123482,0.0007101666],"about_ca_topic_score_codex":0.077270195,"about_ca_topic_score_gemma":0.013166003,"teacher_disagreement_score":0.51889384,"about_ca_system_score_codex":0.00015086615,"about_ca_system_score_gemma":0.00089544104,"threshold_uncertainty_score":0.9288743},"labels":[],"label_agreement":null},{"id":"W4403143722","doi":"10.1007/s11192-024-05145-6","title":"Interdisciplinarity and expert rating: an analysis based on faculty opinions","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"Interdisciplinary Research and Collaboration","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Social Science Fund of China","keywords":"Sociology; Computer science; Data science; Psychology; Management science; Engineering","score_opus":0.23329725374897026,"score_gpt":0.5451679613832665,"score_spread":0.31187070763429625,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403143722","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7771541,0.00041942636,0.1842975,0.01574253,0.002445636,0.0005896598,0.0009403064,0.00030877872,0.018102078],"genre_scores_gemma":[0.99620783,0.000001859446,0.0013454149,0.00012659661,0.00012371266,0.000019817913,0.0001850868,0.000007927264,0.0019817287],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99466985,0.00025400627,0.0004262364,0.0009364045,0.0033458297,0.00036766348],"domain_scores_gemma":[0.9970459,0.001089293,0.00006586374,0.00067283586,0.0007175007,0.00040861577],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0063840263,0.00015298738,0.00024302842,0.010413038,0.0005646345,0.0033168618,0.0007378678,0.00007602713,0.0003259655],"category_scores_gemma":[0.0030775568,0.00010331301,0.00016191666,0.05474503,0.0001891067,0.00086270866,0.00038787958,0.00022212332,0.00016676322],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00067503477,0.0023931505,0.1585543,0.000083973006,0.00058640965,0.00030338627,0.035203133,0.03080096,0.0020678479,0.056046557,0.25261265,0.46067262],"study_design_scores_gemma":[0.00015513918,0.00054143777,0.040945623,0.000014620872,0.000022496219,0.0000015226243,0.0030293595,0.9425716,0.00021994363,0.002455977,0.009872912,0.00016940701],"about_ca_topic_score_codex":0.000014351169,"about_ca_topic_score_gemma":0.00017134691,"teacher_disagreement_score":0.9117706,"about_ca_system_score_codex":0.00014761907,"about_ca_system_score_gemma":0.00020691042,"threshold_uncertainty_score":0.9977178},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4403437307","doi":"10.1007/s11192-024-05163-4","title":"Open access improves the dissemination of science: insights from Wikipedia","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"Wikis in Education and Collaboration","field":"Social Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"China Scholarship Council","keywords":"Open science; Computer science; World Wide Web; Data science; Scholarly communication; Information Dissemination; Library science; Information retrieval; Political science; Mathematics; Publishing","score_opus":0.059974193410401964,"score_gpt":0.49664389550811855,"score_spread":0.4366697020977166,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403437307","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8614618,0.0013918277,0.00041926253,0.0044501107,0.007974719,0.00053645816,0.000020005307,0.000061706814,0.123684116],"genre_scores_gemma":[0.995787,0.00013620254,0.00019731354,0.000091492075,0.00027720333,0.000019367877,0.000005801835,0.0000040749146,0.0034815466],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99790144,0.000075331016,0.00017520168,0.00028334107,0.0013917077,0.00017299867],"domain_scores_gemma":[0.9982241,0.00063683826,0.00008682575,0.00024297017,0.0007102558,0.000099021585],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0025396778,0.000054990785,0.000070175854,0.0015785006,0.00087783125,0.004777106,0.0027446079,0.000040903647,0.00016583758],"category_scores_gemma":[0.00497998,0.00003773084,0.000022547236,0.042745747,0.0013678354,0.0032205938,0.00045372723,0.0000794632,0.00002780982],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000064600526,0.00017324499,0.0032002542,0.000018916526,0.00001398984,7.642912e-7,0.1472842,0.000009648487,0.012386561,0.53750825,0.039349306,0.26004842],"study_design_scores_gemma":[0.00015228982,0.00006408727,0.20732687,0.00015125601,0.000053598502,2.1167624e-7,0.072780125,0.0025291825,0.019237628,0.055326335,0.64199126,0.00038716404],"about_ca_topic_score_codex":0.001268919,"about_ca_topic_score_gemma":0.00048282827,"teacher_disagreement_score":0.60264194,"about_ca_system_score_codex":0.00021930768,"about_ca_system_score_gemma":0.0019187795,"threshold_uncertainty_score":0.99625605},"labels":[],"label_agreement":null},{"id":"W4403560014","doi":"10.1007/s11192-024-05168-z","title":"Systematic bias in citing practices: evidence from accounting journal additions to the FT list","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"Auditing, Earnings Management, Governance","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"Accounting; Bibliometrics; Library science; Information retrieval; Computer science; Political science; Econometrics; Economics","score_opus":0.09018432702100378,"score_gpt":0.3219311249758001,"score_spread":0.23174679795479636,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403560014","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.75651455,0.007552529,0.19628838,0.014694324,0.008421171,0.002020273,0.000048208592,0.00059037126,0.013870189],"genre_scores_gemma":[0.99324673,0.00009050997,0.0008219676,0.0021142464,0.0026843778,0.000060006012,0.000010259658,0.000041764557,0.0009301383],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9960569,0.00007570312,0.0007707351,0.000574612,0.0019175801,0.0006044462],"domain_scores_gemma":[0.98237646,0.003622199,0.012836122,0.0005596422,0.00055575755,0.000049811726],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.009542187,0.00023181,0.00028613888,0.0040430613,0.0006902312,0.009080347,0.0012819344,0.000053830718,0.00045403707],"category_scores_gemma":[0.30963364,0.00017290865,0.00012159685,0.026403897,0.00005723011,0.0048586642,0.0007475901,0.00067382125,0.001928857],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000055414363,0.0004913887,0.35567585,0.044539023,0.00063009124,0.0014841715,0.011300909,0.018660964,0.0009982144,0.024707973,0.35840946,0.18304653],"study_design_scores_gemma":[0.00062480744,0.000040029045,0.10977657,0.08267191,0.00070677965,0.00005276409,0.011119433,0.13086225,0.00003523561,0.0024244704,0.660156,0.001529743],"about_ca_topic_score_codex":0.0012226813,"about_ca_topic_score_gemma":0.0004946513,"teacher_disagreement_score":0.30174655,"about_ca_system_score_codex":0.0003230785,"about_ca_system_score_gemma":0.0000862139,"threshold_uncertainty_score":0.99884826},"labels":[],"label_agreement":null},{"id":"W4405389827","doi":"10.1007/s11192-024-05215-9","title":"The use of emotion words by the sciences and other subjects","year":2024,"lang":"en","type":"article","venue":"Scientometrics","topic":"Emotions and Moral Behavior","field":"Psychology","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Brock University","funders":"","keywords":"Amusement; Pleasure; Salient; Psychology; Affective science; Discipline; Social science; Multidisciplinary approach; Social psychology; Sociology; Cognitive psychology; Emotion classification; History","score_opus":0.18986084024820976,"score_gpt":0.4081956613056643,"score_spread":0.21833482105745455,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405389827","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.989844,0.0044030487,0.00041587674,0.00074400613,0.0015914366,0.0001314593,0.000027030179,0.000026458096,0.0028166394],"genre_scores_gemma":[0.9898964,0.000055101093,0.00010517745,0.0001123998,0.000027039861,0.0000057552656,5.901622e-7,0.000005442462,0.009792067],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99914837,0.0000631867,0.00011031886,0.00018465806,0.0003169005,0.00017653783],"domain_scores_gemma":[0.99927247,0.0004355248,0.000031748834,0.00018438452,0.000041881543,0.000033998276],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0012739281,0.000051467887,0.000047099453,0.0005521988,0.00029093033,0.00032424624,0.00019394899,0.000032175016,0.000084901425],"category_scores_gemma":[0.00017999188,0.000024691846,0.000028610963,0.007476855,0.00060121034,0.00009882113,0.000046488087,0.00007934927,0.000030740743],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010609914,0.00018727189,0.10643646,0.000021531023,0.000036049885,0.0000043567347,0.005715713,0.000011391368,0.001849719,0.14808056,0.1633953,0.57425106],"study_design_scores_gemma":[0.00020582911,0.0002913422,0.15633228,0.000039112114,0.00005101849,0.000027838767,0.0018007612,0.000642628,0.0005882248,0.0006125107,0.8392454,0.00016307902],"about_ca_topic_score_codex":0.00015723669,"about_ca_topic_score_gemma":0.000015439477,"teacher_disagreement_score":0.67585003,"about_ca_system_score_codex":0.000013544704,"about_ca_system_score_gemma":0.000021146148,"threshold_uncertainty_score":0.35923812},"labels":[],"label_agreement":null},{"id":"W4409158554","doi":"10.1007/s11192-025-05296-0","title":"Examining the role of co-first authorship in scientific collaboration: a quantitative study","year":2025,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Bibliometrics; Scientific communication; Scientific literature; Sociology; Data science; Library science; Political science; Computer science","score_opus":0.6419803795842838,"score_gpt":0.6109582313954125,"score_spread":0.03102214818887128,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409158554","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9737291,0.002277418,0.0010200188,0.000552004,0.0015757377,0.0010741974,0.00004569259,0.00002183709,0.019704027],"genre_scores_gemma":[0.9959499,0.000021306492,0.00060180295,0.00003518371,0.000016835727,0.000044809814,0.0000030720735,0.000007961398,0.0033191284],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9786101,0.001037848,0.0015672792,0.0014062485,0.016488777,0.0008897453],"domain_scores_gemma":[0.96724516,0.02040634,0.00065889966,0.0019897446,0.009360002,0.0003398561],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.11787893,0.0002090679,0.0005021271,0.25902575,0.0010153905,0.006315107,0.005061687,0.00012747665,0.00016184636],"category_scores_gemma":[0.18387233,0.00013298022,0.0001060675,0.8153915,0.0010143381,0.0008863263,0.0010929861,0.0004463833,0.00014228036],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000033593085,0.00070548547,0.95565206,0.000006278547,0.000015839103,0.000005232169,0.0058157467,0.0002897301,0.0004416861,0.011862708,0.002773428,0.022398202],"study_design_scores_gemma":[0.0009462487,0.00050375727,0.86705846,0.000025755513,0.000012378182,8.3037054e-7,0.07146922,0.012080025,0.0019118444,0.015516148,0.030259809,0.00021554019],"about_ca_topic_score_codex":0.00016020647,"about_ca_topic_score_gemma":0.0006408005,"teacher_disagreement_score":0.5563657,"about_ca_system_score_codex":0.00032068003,"about_ca_system_score_gemma":0.0010832623,"threshold_uncertainty_score":0.9947164},"labels":[],"label_agreement":null},{"id":"W4411174920","doi":"10.1007/s11192-025-05339-6","title":"Mapping the unseen in practice: comparing latent Dirichlet allocation and BERTopic for navigating topic spaces","year":2025,"lang":"en","type":"article","venue":"Scientometrics","topic":"Computational and Text Analysis Methods","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung","keywords":"Latent Dirichlet allocation; Topic model; Computer science; Data science; Information retrieval","score_opus":0.1049383044834109,"score_gpt":0.4485186821781035,"score_spread":0.3435803776946926,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4411174920","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.901986,0.0017241986,0.05515245,0.029150039,0.0005224028,0.00046123247,8.239328e-7,0.000030804626,0.0109720575],"genre_scores_gemma":[0.98817855,0.000081456885,0.010442397,0.00029005884,0.00005612604,0.000015615866,0.0000025232155,0.0000016463445,0.0009315995],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.998965,0.00018203023,0.00016934107,0.00017305724,0.0003374914,0.00017306051],"domain_scores_gemma":[0.99741244,0.0021946859,0.000085160034,0.00007353548,0.00020068146,0.000033466615],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0039688884,0.00004864371,0.000099797,0.0005892284,0.0005238798,0.0002593134,0.00017033644,0.000030310504,0.0000024543497],"category_scores_gemma":[0.0041031595,0.00003823644,0.000029575644,0.008232093,0.00009853038,0.0001724799,0.0000670994,0.00009618097,8.5875706e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000460698,0.000053738055,0.51764363,0.00003339139,0.00003704091,6.158145e-7,0.020085895,0.0012738958,0.0000446103,0.12967089,0.00014692578,0.33100474],"study_design_scores_gemma":[0.00057632884,0.000020733973,0.7455396,0.00012441327,0.000062318744,4.968405e-7,0.020281563,0.06776394,0.00004457236,0.023427872,0.14196102,0.0001971257],"about_ca_topic_score_codex":0.0010982257,"about_ca_topic_score_gemma":0.0006284598,"teacher_disagreement_score":0.33080763,"about_ca_system_score_codex":0.00010129224,"about_ca_system_score_gemma":0.00008501285,"threshold_uncertainty_score":0.49121615},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4415960658","doi":"10.1007/s11192-025-05453-5","title":"A method for assessing interdisciplinarity through co-authors of scientific publications","year":2025,"lang":"en","type":"article","venue":"Scientometrics","topic":"Interdisciplinary Research and Collaboration","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Natural Science Foundation of China","keywords":"Discipline; Diversity (politics); Scope (computer science); Interdisciplinarity; Work (physics); Broad spectrum; Sociology of scientific knowledge","score_opus":0.29238629110060305,"score_gpt":0.6097440840706423,"score_spread":0.31735779297003924,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415960658","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07026662,0.00011345494,0.8761277,0.005564172,0.001646972,0.00064765307,0.00015926501,0.000041175605,0.045432962],"genre_scores_gemma":[0.8375642,0.0000012261645,0.12345051,0.00006712593,0.000055954857,0.000098107615,0.000047762216,0.000008369976,0.03870675],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9945188,0.00034663317,0.0009382405,0.0008831802,0.002814989,0.00049816206],"domain_scores_gemma":[0.98848087,0.0035394582,0.00041131384,0.0010814234,0.006329952,0.00015700892],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.02255633,0.00013669059,0.00033439655,0.008941981,0.0013087996,0.0033263979,0.0018469073,0.000112034766,0.00012723549],"category_scores_gemma":[0.019801645,0.0001065217,0.00021589626,0.06828894,0.00053815113,0.0017592153,0.00071889156,0.00018606333,0.000040927345],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001441088,0.0008805473,0.025116144,0.00010637206,0.000069903814,0.0000011723431,0.004595557,0.0004251094,0.011406214,0.3748752,0.40141967,0.18096001],"study_design_scores_gemma":[0.0010971293,0.0002887066,0.040784754,0.00008525,0.000045867815,0.0000036081806,0.012046258,0.10229427,0.031914484,0.5364819,0.2745754,0.00038231834],"about_ca_topic_score_codex":0.0000064876904,"about_ca_topic_score_gemma":0.000056240508,"teacher_disagreement_score":0.76729757,"about_ca_system_score_codex":0.00019626984,"about_ca_system_score_gemma":0.0010617229,"threshold_uncertainty_score":0.99999136},"labels":[],"label_agreement":null},{"id":"W4415963683","doi":"10.1007/s11192-025-05454-4","title":"Exploring citation behavior of national self-references","year":2025,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Université du Québec à Montréal","funders":"National Social Science Fund of China; China Association for Science and Technology","keywords":"Scarcity; Diversification (marketing strategy); Citation; Norm (philosophy); Scientific communication; Scientific literature; Process (computing); Citation analysis","score_opus":0.9021341018802934,"score_gpt":0.6353747233350729,"score_spread":0.26675937854522047,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415963683","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9445034,0.00048809196,0.0051457994,0.00029339036,0.0021146338,0.00029332392,0.000045143996,0.000053872544,0.04706232],"genre_scores_gemma":[0.9933284,0.00022687599,0.004123298,0.000062128136,0.00003659438,0.000052040272,0.0000046964137,0.0000052344535,0.0021607112],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.977176,0.00017827883,0.0009974019,0.0007611617,0.020352291,0.00053485157],"domain_scores_gemma":[0.97846484,0.006223836,0.00038878236,0.0005898213,0.014047294,0.00028543413],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.035325885,0.00014516068,0.00031364223,0.30197138,0.00028975127,0.0014558379,0.0027081121,0.00008786057,0.00022962842],"category_scores_gemma":[0.09542064,0.000110957604,0.00016684853,0.6710516,0.00020678963,0.0014334747,0.0006702067,0.0002096313,0.00016715277],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013803305,0.0007082309,0.60989475,0.000022860902,0.000031537693,0.0000032128714,0.00046674686,0.00012362843,0.0018990248,0.036331072,0.008715275,0.34178984],"study_design_scores_gemma":[0.0004929491,0.00012400521,0.9564057,0.000015417325,0.000013739445,0.0000012996403,0.0007169114,0.0026593816,0.0069878306,0.0112402495,0.021148512,0.00019397971],"about_ca_topic_score_codex":0.000039603172,"about_ca_topic_score_gemma":0.0000060099815,"teacher_disagreement_score":0.36908025,"about_ca_system_score_codex":0.0002837151,"about_ca_system_score_gemma":0.0007206229,"threshold_uncertainty_score":0.99958074},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4415966073","doi":"10.1007/s11192-025-05467-z","title":"Research evaluation metrics comparison: does citation or citation context count differ from citation context similarity metric?","year":2025,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University; University of Lethbridge","funders":"","keywords":"Citation; Context (archaeology); Similarity (geometry); Weighting; Citation analysis; Metric (unit); Rank (graph theory)","score_opus":0.7775964114144409,"score_gpt":0.6498567782278969,"score_spread":0.12773963318654402,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415966073","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.83331746,0.009150801,0.12861688,0.004444565,0.009397165,0.0036922796,0.00075615064,0.00019881885,0.010425891],"genre_scores_gemma":[0.9893134,0.0005486813,0.003297392,0.0007480381,0.00024301819,0.00021491655,0.0003036454,0.00004132524,0.005289548],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.91566736,0.005294491,0.003926734,0.003721192,0.06891334,0.0024768908],"domain_scores_gemma":[0.7973707,0.11136612,0.0018848148,0.0033820663,0.08443266,0.0015636894],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.17903253,0.00063522917,0.0013299187,0.40338805,0.0021548683,0.012124805,0.005414058,0.0007547828,0.0022627285],"category_scores_gemma":[0.5252329,0.00042713582,0.00044258236,0.7771955,0.0010150763,0.0028778382,0.0015759665,0.0016446108,0.0009677044],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00034318052,0.0009901978,0.19776839,0.00003753059,0.00012919048,0.000010143468,0.0016137064,0.0004229377,0.0005629576,0.006156327,0.08596887,0.7059966],"study_design_scores_gemma":[0.0044994703,0.0006619761,0.5980206,0.00009208247,0.00015662822,0.0000016197276,0.021659508,0.23769788,0.0041778637,0.05553897,0.076635644,0.00085771695],"about_ca_topic_score_codex":0.001550698,"about_ca_topic_score_gemma":0.0013418221,"teacher_disagreement_score":0.70513886,"about_ca_system_score_codex":0.0035694921,"about_ca_system_score_gemma":0.0024472033,"threshold_uncertainty_score":0.9999671},"labels":[],"label_agreement":null},{"id":"W4417456948","doi":"10.1007/s11192-025-05508-7","title":"Data as scholarly output: addressing challenges in data citation tracking through natural language processing automation","year":2025,"lang":"en","type":"article","venue":"Scientometrics","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"Concordia University","keywords":"Citation; Python (programming language); Tracking (education); Automation; Citation analysis; Bibliometrics","score_opus":0.5489135483319633,"score_gpt":0.5066334825260266,"score_spread":0.042280065805936684,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417456948","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02095882,0.07295966,0.8615018,0.019182384,0.0018574061,0.0009937583,0.00011790856,0.00063025666,0.021798013],"genre_scores_gemma":[0.89278054,0.0017862727,0.10369338,0.0003255274,0.00006176101,0.000009469851,0.00057404774,0.000011591893,0.0007574348],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9955098,0.00029611925,0.00039001252,0.0013947452,0.0019035203,0.00050581765],"domain_scores_gemma":[0.9950242,0.00061224605,0.00026770696,0.00371026,0.0003206882,0.00006488798],"candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.00965426,0.00015329696,0.000166585,0.004683469,0.00031759051,0.015774947,0.011174264,0.0000678391,0.0000023266198],"category_scores_gemma":[0.02352634,0.00014871701,0.0000146832845,0.020756474,0.0000713031,0.19550458,0.007535483,0.00050268666,0.000025036405],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004617935,0.000101838545,0.00057679584,0.00022717546,0.000013116235,0.00003234542,0.0017411838,0.000038916867,0.00015090128,0.014261432,0.0007226224,0.98212904],"study_design_scores_gemma":[0.0007446575,0.000029515732,0.062848486,0.00054265955,0.000020904683,0.0000066753664,0.0031216645,0.88873446,0.00018750323,0.0011883901,0.04221455,0.00036055176],"about_ca_topic_score_codex":0.00009892069,"about_ca_topic_score_gemma":0.00015773492,"teacher_disagreement_score":0.9817685,"about_ca_system_score_codex":0.00018114627,"about_ca_system_score_gemma":0.0003494503,"threshold_uncertainty_score":0.99727976},"labels":[],"label_agreement":null},{"id":"W654809136","doi":"10.1007/s11192-015-1610-x","title":"Statistical tests for ‘related records’ search results","year":2015,"lang":"en","type":"article","venue":"Scientometrics","topic":"Statistics Education and Methodologies","field":"Mathematics","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Ranking (information retrieval); Computer science; Information retrieval; Subject (documents); Set (abstract data type); Measure (data warehouse); Negative binomial distribution; Raw data; Statistics; Data mining; Mathematics; Library science","score_opus":0.7040664114245264,"score_gpt":0.5759315830768087,"score_spread":0.12813482834771772,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W654809136","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.043558415,0.00014864157,0.91505843,0.0012544538,0.007629447,0.0010514054,0.0016708696,0.00033645728,0.029291859],"genre_scores_gemma":[0.043030504,0.0000123012205,0.9400946,0.000073236784,0.00009111114,0.000029224777,0.00009584359,0.000023049693,0.016550137],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9979845,0.00017484209,0.0004004607,0.00032460326,0.00071868877,0.00039687345],"domain_scores_gemma":[0.98651886,0.01172871,0.00010912856,0.00034072768,0.0009563878,0.00034615683],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.007557162,0.000109381384,0.00019814378,0.0013341297,0.00009481707,0.00010129992,0.00027214404,0.0000881114,0.00004854311],"category_scores_gemma":[0.20380263,0.00009452792,0.00003555599,0.0045024035,0.00014216629,0.000060080245,0.000074958254,0.00013652715,0.00012476383],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000088358414,0.00021911053,0.0010886531,0.000057439225,0.000017385242,0.000003619946,0.0018509795,0.000011025661,0.000025004847,0.29300123,0.6759457,0.027691457],"study_design_scores_gemma":[0.0030487522,0.0005810854,0.007504543,0.000023047654,0.000056346915,0.000018508168,0.002026144,0.003480154,0.00070877816,0.8694619,0.112693906,0.00039686356],"about_ca_topic_score_codex":0.000014147316,"about_ca_topic_score_gemma":0.000004842519,"teacher_disagreement_score":0.5764606,"about_ca_system_score_codex":0.0001573553,"about_ca_system_score_gemma":0.00037728393,"threshold_uncertainty_score":0.80290407},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"methods","study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"}]}