{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":407,"total_is_capped":false,"direct_labels_cover":407,"predictions_cover":407,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"query_hash":"b735a3723fba","filters":{"category":"scholarly_communication"}},"results":[{"id":"W1981886524","doi":"10.1002/asi.22688","title":"<scp>SciMAT</scp>: A new science mapping analysis software tool","year":2012,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":1187,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Universidad de Granada; McGill University","keywords":"Wizard; Workflow; Computer science; Software; Key (lock); Preprocessor; Software engineering; Data science; World Wide Web; Database; Programming language; Operating system","retraction":null,"screen_n_in":null,"score":{"opus":0.05196923175334746,"gpt":0.3484261076968199,"spread":0.2964568759434724,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","sts"],"consensus_categories":[],"category_scores_codex":[0.01940332,0.0001085865,0.0003237092,0.002261807,0.001253985,0.0009835976,0.003130979,0.00003275443,0.000003290966],"category_scores_gemma":[0.02019862,0.00006406442,0.0003252192,0.03076893,0.003825553,0.005885516,0.001090073,0.0001813557,0.00001468475],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001522741,"about_ca_system_score_gemma":0.0007120204,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001136739,"about_ca_topic_score_gemma":7.47178e-7,"domain_scores_codex":[0.9956865,0.00001783388,0.0008506853,0.0002313574,0.002671872,0.0005417836],"domain_scores_gemma":[0.994259,0.0005955732,0.0020256,0.0008068256,0.002126514,0.000186479],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000004835402,0.0000452368,0.08801094,0.000008388066,0.0001500285,1.424548e-7,0.006928504,0.0004734667,0.001367213,0.008810039,0.1303992,0.763802],"study_design_scores_gemma":[0.0005843117,0.0001937111,0.1426622,0.00002744104,0.0002482357,0.0001180648,0.07462858,0.02250362,0.001643411,0.01109563,0.7461097,0.0001850314],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7817136,0.00007199423,0.2138005,0.003152357,0.0007336466,0.0001680681,0.00000909911,0.00003898368,0.0003117651],"genre_scores_gemma":[0.9304643,0.00002129438,0.06772479,0.001478792,0.00007412121,0.000002877725,4.595117e-7,0.000002543762,0.0002308691],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.763617,"threshold_uncertainty_score":0.9988855,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"simulation_or_modeling","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4200395585","doi":"10.1007/s40037-021-00695-4","title":"Bibliometrics: Methods for studying academic publishing","year":2021,"lang":"en","type":"article","venue":"Perspectives on Medical Education","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":938,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Royal College of Physicians and Surgeons of Canada; University of Ottawa","funders":"","keywords":"Bibliometrics; Terminology; Publishing; Data science; Strengths and weaknesses; Population health; Computer science; Population; Library science; Medicine; Psychology; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.652587060367001,"gpt":0.6967628458869507,"spread":0.04417578551994972,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.05948411,0.0001703509,0.0003682181,0.1369572,0.000413167,0.004798167,0.002418063,0.0003491818,0.002225628],"category_scores_gemma":[0.7677514,0.0001292189,0.0002238737,0.416792,0.0001774939,0.00162808,0.00049591,0.001073089,0.000106925],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004534263,"about_ca_system_score_gemma":0.004114283,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003250789,"about_ca_topic_score_gemma":0.000003341181,"domain_scores_codex":[0.9856709,0.0009725204,0.0008070684,0.001311657,0.01059632,0.000641596],"domain_scores_gemma":[0.9095557,0.06000977,0.0007112044,0.001913936,0.02511396,0.002695381],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00001709267,0.000630289,0.005164388,0.000007109139,0.0000195984,0.00000169737,0.002081413,0.000003215156,0.0002061208,0.0243053,0.05017921,0.9173846],"study_design_scores_gemma":[0.001984875,0.0005301767,0.09929708,0.0001258356,0.00003780989,0.00005040143,0.2043805,0.01754517,0.002924726,0.1837441,0.4886123,0.0007670138],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1275684,0.03384215,0.6953861,0.1058889,0.007056116,0.0009652635,0.00002620779,0.0001245586,0.02914232],"genre_scores_gemma":[0.8358486,0.006136074,0.1420518,0.006544131,0.002420725,0.0003218003,0.00004027083,0.00005396955,0.006582601],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9166176,"threshold_uncertainty_score":0.9986865,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2518646621","doi":"10.1002/asi.23833","title":"Scholarly use of social media and altmetrics: A review of the literature","year":2017,"lang":"en","type":"review","venue":"Journal of the Association for Information Science and Technology","topic":"Social Media in Health Education","field":"Social Sciences","cited_by":571,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Université du Québec à Montréal; Université de Montréal","funders":"Social Sciences and Humanities Research Council of Canada; Alfred P. Sloan Foundation","keywords":"Altmetrics; Scholarship; Social media; Scholarly communication; Promotion (chess); Sociology; Data science; Computer science; Social science; Political science; World Wide Web; Publishing","retraction":null,"screen_n_in":null,"score":{"opus":0.1787038404795363,"gpt":0.4544080046811156,"spread":0.2757041642015793,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.009269384,0.0001005428,0.0007071443,0.0008829859,0.001114251,0.0003139401,0.001087802,0.0004702621,0.000001239783],"category_scores_gemma":[0.2233331,0.00005878304,0.0001985243,0.003506695,0.0008932215,0.003824422,0.0001459278,0.0005352979,4.356691e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005799502,"about_ca_system_score_gemma":0.0028066,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001720351,"about_ca_topic_score_gemma":0.00001685298,"domain_scores_codex":[0.997097,0.0002190992,0.001100124,0.00007623927,0.001315915,0.0001916604],"domain_scores_gemma":[0.9822153,0.002102267,0.009452125,0.000217174,0.005962281,0.00005085108],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000001643679,0.00001128313,0.0007042354,0.007511693,0.00003177635,1.753499e-8,0.008199901,1.013396e-8,3.654889e-7,0.02109617,0.001654578,0.9607883],"study_design_scores_gemma":[0.00009192092,0.00001738894,0.0009423506,0.01117186,0.0001887937,0.000003793366,0.0004514296,2.488927e-7,0.000002831029,0.001044316,0.9860325,0.00005253019],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.0002980301,0.9815807,0.000002462864,0.01274519,0.00353915,0.001553081,0.00007040392,0.000006808357,0.0002042054],"genre_scores_gemma":[0.0005202389,0.9989569,0.0001467182,0.0001280153,0.0001776179,0.00004195437,0.000001838229,0.000003427715,0.00002335471],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.984378,"threshold_uncertainty_score":0.8570033,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","scholarly_communication"],"domain":"evaluation","study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2028758832","doi":"10.1016/s0140-6736(10)60903-8","title":"An international registry of systematic-review protocols","year":2010,"lang":"en","type":"article","venue":"The Lancet","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":521,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Ottawa Hospital","funders":"National Institute for Health and Care Research","keywords":"Medicine; Computer science","retraction":null,"screen_n_in":null,"score":{"opus":0.7505475693393026,"gpt":0.5892674142758912,"spread":0.1612801550634113,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","open_science","insufficient_payload"],"consensus_categories":["metaresearch","insufficient_payload"],"category_scores_codex":[0.09903283,0.0001218203,0.003438641,0.00005135027,0.00004429752,0.0004708082,0.007977618,0.0000359237,0.006925179],"category_scores_gemma":[0.04988371,0.00004036322,0.0006346548,0.0003702229,0.00006140624,0.0001627304,0.00007727091,0.0001637879,0.00112761],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000004045208,"about_ca_system_score_gemma":0.00002854121,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000003480945,"about_ca_topic_score_gemma":0.00001383572,"domain_scores_codex":[0.9873357,0.003868084,0.00520132,0.0003307846,0.003144667,0.0001194666],"domain_scores_gemma":[0.9823982,0.001172291,0.004618836,0.01098145,0.0007763844,0.00005288919],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00002106834,0.000207084,0.03141287,0.1485324,0.000640763,0.00000524328,0.000656404,0.00001146289,0.003569071,0.09331319,0.7190724,0.002558021],"study_design_scores_gemma":[0.001036418,0.0001395896,0.02423659,0.07192647,0.001651501,0.0001811948,0.0005905969,0.007907512,0.0007359895,0.0299885,0.8609088,0.0006968519],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.1408229,0.01093169,0.01967539,0.07371772,0.004405637,0.2561955,0.0008109328,0.0001142744,0.4933259],"genre_scores_gemma":[0.9168285,0.0003760864,0.02333628,0.007845412,0.002427917,0.02198249,0.00002812314,0.00003987004,0.02713529],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7760056,"threshold_uncertainty_score":0.9996501,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"reproducibility","study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W3024722698","doi":"10.1038/s41597-020-0486-7","title":"The TRUST Principles for digital repositories","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Research Data Management Practices","field":"Computer Science","cited_by":390,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Ocean Networks Canada Society; University of Victoria","funders":"National Institute of Biomedical Imaging and Bioengineering; National Institute of Diabetes and Digestive and Kidney Diseases; Economic and Social Research Council","keywords":"World Wide Web; Computer science; Metadata; Scholarly communication; Data science; Internet privacy; Knowledge management; Publishing; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.2872569636919766,"gpt":0.3757555810503705,"spread":0.08849861735839387,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.002147771,0.00007645638,0.00006558477,0.00003154868,0.0009497717,0.03837242,0.01526573,0.00001304801,0.000001967453],"category_scores_gemma":[0.006170863,0.00005060746,0.00002346914,0.0006003391,0.0002592145,0.04000018,0.01025894,0.00008066493,0.00007849642],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000011219,"about_ca_system_score_gemma":0.0001221455,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000003639181,"about_ca_topic_score_gemma":0.000009160065,"domain_scores_codex":[0.9978414,0.00003923405,0.0002062888,0.0009667238,0.0006246713,0.0003216393],"domain_scores_gemma":[0.9943673,0.0004363721,0.0001152713,0.004846899,0.0001048473,0.0001293331],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0000182429,0.00002992879,0.0001968107,0.00003707585,0.0000348169,0.00001005414,0.0002488848,0.000007382684,0.0001887925,0.4258383,0.5179107,0.05547902],"study_design_scores_gemma":[0.00007565916,0.00002441888,0.0001751642,0.000003098945,0.000003187548,0.000001173154,0.00007910952,0.04619589,0.0002360177,0.0005813912,0.9525578,0.00006707858],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0005563246,0.0004394346,0.9210995,0.06060238,0.003790907,0.0009658859,0.002989911,0.0003062158,0.009249465],"genre_scores_gemma":[0.3362477,0.000683527,0.3351711,0.002322791,0.004216848,0.0004623708,0.02062322,0.000130461,0.300142],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.5859284,"threshold_uncertainty_score":0.9977459,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2132615533","doi":"10.1002/asi.20349","title":"The place of serials in referencing practices: Comparing natural sciences and engineering with social sciences and humanities","year":2006,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":294,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Université du Québec à Montréal","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"Natural science; Citation index; Citation; Science Citation Index; Social science; Index (typography); The arts; Digital humanities; Bibliometrics; Citation analysis; Social Sciences Citation Index; Library science; Sociology; Humanities; Computer science; Political science; Epistemology; Philosophy; World Wide Web","retraction":null,"screen_n_in":null,"score":{"opus":0.2802135795098643,"gpt":0.4847581285676076,"spread":0.2045445490577433,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["sts"],"category_scores_codex":[0.02117882,0.00007076069,0.0002323258,0.005429512,0.001389506,0.00147905,0.001206029,0.00002790762,2.377291e-7],"category_scores_gemma":[0.009161904,0.00003278094,0.00003626956,0.0419622,0.007014023,0.002738237,0.0003946512,0.0001952862,7.257341e-8],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005978278,"about_ca_system_score_gemma":0.0004433976,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001018022,"about_ca_topic_score_gemma":0.00003930776,"domain_scores_codex":[0.9966777,0.00002705425,0.0005481834,0.0001352605,0.00233248,0.0002793426],"domain_scores_gemma":[0.9942899,0.001960979,0.001977154,0.00009659831,0.001650492,0.00002485512],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0002569174,0.00009866897,0.551029,0.0001214073,0.00008087121,0.00000123283,0.01170226,0.001858135,0.03741667,0.1973272,0.003427963,0.1966797],"study_design_scores_gemma":[0.002141481,0.002501498,0.6369707,0.0001661279,0.00004123221,0.0004478701,0.2087575,0.09776947,0.008233329,0.01351365,0.02896735,0.0004897672],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9953043,0.0004004287,0.0001653765,0.003739407,0.00007384374,0.0001135893,0.000001856142,0.000003987117,0.0001971972],"genre_scores_gemma":[0.9968065,0.0001924588,0.002888852,0.00008264338,0.00001743036,0.000002274214,3.497029e-8,0.000001173749,0.000008609317],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1970553,"threshold_uncertainty_score":0.9999105,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1987707043","doi":"10.1136/bmj.h210","title":"Firm action needed on predatory journals","year":2015,"lang":"en","type":"editorial","venue":"BMJ","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":243,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Action (physics); Business; Public relations; Law and economics; Economics; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.7969926508729327,"gpt":0.6778639477958661,"spread":0.1191287030770666,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.07453775,0.0003023434,0.0006567832,0.08824164,0.0002552495,0.005935813,0.00401319,0.001299479,0.00191309],"category_scores_gemma":[0.3744678,0.0002046667,0.0003103461,0.1151029,0.0001050054,0.000630898,0.0007479962,0.001859084,0.005137558],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005909696,"about_ca_system_score_gemma":0.002213753,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001251702,"about_ca_topic_score_gemma":0.00002060145,"domain_scores_codex":[0.9520504,0.0006182799,0.0012438,0.001019096,0.04437969,0.0006887168],"domain_scores_gemma":[0.9646193,0.01673616,0.001270443,0.001828136,0.01453772,0.001008263],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00007796607,0.00007901764,0.0000950033,0.000006105177,0.00002103017,0.00002432573,0.00002812518,0.00001827629,0.00001642444,0.000005224414,0.9534311,0.04619735],"study_design_scores_gemma":[0.0004521565,0.0002922913,0.0004840363,0.00004061568,0.000007792441,0.000002787902,0.0001961423,0.000192195,0.00005272311,0.001847212,0.9961783,0.0002537403],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"editorial","genre_gemma":"editorial","genre_scores_codex":[0.001093037,0.001137071,0.0001123373,0.001476189,0.9631559,0.0004378086,0.0003500536,0.00004000721,0.03219762],"genre_scores_gemma":[0.001018158,0.0005184346,0.0001356163,0.0001064794,0.9289952,0.00003192541,0.00005132973,0.00003637673,0.0691065],"genre_candidate":"editorial","genre_consensus":"editorial","teacher_disagreement_score":0.2999301,"threshold_uncertainty_score":0.999997,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","scholarly_communication"],"domain":"evaluation","study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","research_integrity","scholarly_communication"],"domain":"methods","study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2048204427","doi":"10.1038/513030a","title":"Satellites: Make Earth observations open access","year":2014,"lang":"en","type":"article","venue":"Nature","topic":"Space exploration and regulation","field":"Physics and Astronomy","cited_by":241,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Natural Resources Canada; University of British Columbia; Canadian Forest Service","funders":"","keywords":"Astrobiology; Earth (classical element); Remote sensing; Geology; Astronomy; Physics","retraction":null,"screen_n_in":null,"score":{"opus":0.04629024388331381,"gpt":0.3451974733286163,"spread":0.2989072294453025,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00007844525,0.00006347222,0.00007076935,0.00002726895,0.0001063259,0.0003732105,0.000363649,0.0001491576,0.0004693685],"category_scores_gemma":[0.000008705993,0.00005231784,0.00002961037,0.0001824257,0.00001117567,0.0005016145,0.0001162586,0.0003655432,0.00006453383],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000003744557,"about_ca_system_score_gemma":0.00002261041,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002419942,"about_ca_topic_score_gemma":0.00002342132,"domain_scores_codex":[0.9996005,0.00002572285,0.0000759833,0.0001246528,0.0000891569,0.00008401326],"domain_scores_gemma":[0.9996487,0.00002559939,0.00004929944,0.0001711727,0.00006846929,0.00003669447],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000009855603,0.00007512356,0.1322406,0.000005513496,0.00003189562,1.196825e-7,0.0002529102,0.0004554982,0.001333495,0.7742662,0.02635192,0.06497684],"study_design_scores_gemma":[0.0003592836,0.00000950007,0.1913068,0.00001288567,0.000007808739,8.965802e-8,0.00008822516,0.001605714,0.003053364,0.02158269,0.7818468,0.0001268595],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5131612,0.0002377815,0.04151083,0.03030393,0.002384463,0.001365396,0.00007145126,0.0002282875,0.4107367],"genre_scores_gemma":[0.9875864,0.000001241165,0.001827039,0.001034299,0.000441317,0.00001638756,0.0002120882,0.000009187534,0.008872085],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7554948,"threshold_uncertainty_score":0.5139256,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2544237122","doi":"10.1002/asi.23781","title":"Developments in research data management in academic libraries: Towards an understanding of research data service maturity","year":2017,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"Research Data Management Practices","field":"Computer Science","cited_by":228,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"RDM; Maturity (psychological); Data management; Data curation; Service (business); Capability Maturity Model; Data management plan; Service delivery framework; Data as a service; Knowledge management; Public relations; Business; Computer science; Data science; Political science; Marketing; Database","retraction":null,"screen_n_in":null,"score":{"opus":0.6200252280254551,"gpt":0.5405688026875997,"spread":0.07945642533785546,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["metaresearch","scholarly_communication","open_science"],"category_scores_codex":[0.06803358,0.00005300273,0.0001324746,0.002812137,0.000814478,0.002376301,0.02052973,0.0001086709,2.900394e-7],"category_scores_gemma":[0.01730596,0.00004062695,0.000006319377,0.003838976,0.0003863393,0.1247002,0.01511551,0.0008249169,0.000001421708],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006760397,"about_ca_system_score_gemma":0.0007814651,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00007911096,"about_ca_topic_score_gemma":0.0001519965,"domain_scores_codex":[0.9960632,0.0001923033,0.0006022178,0.0002382734,0.002497933,0.0004061253],"domain_scores_gemma":[0.9951118,0.0003608016,0.001065851,0.002167942,0.00124575,0.00004790666],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.00003845915,0.00005308373,0.04153537,0.0001152018,0.00002687567,0.000002051529,0.001385444,0.00001116872,0.00008882827,0.9248568,0.001492347,0.03039438],"study_design_scores_gemma":[0.003480372,0.0002787584,0.4660254,0.0006785833,0.0000155625,0.00002047478,0.02749799,0.07241851,0.001763561,0.2980595,0.1294511,0.0003102602],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.4511512,0.0003097227,0.0445619,0.4518813,0.002121032,0.005528682,0.0002743323,0.0001025493,0.04406925],"genre_scores_gemma":[0.9881899,0.0007942202,0.01086177,0.00007756848,0.00001208838,0.000006873842,0.000008521286,0.000001991367,0.0000471233],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.6267973,"threshold_uncertainty_score":0.9986593,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2112084168","doi":"10.1111/j.1365-2958.2005.04703.x","title":"Weapons of mass retraction","year":2005,"lang":"en","type":"review","venue":"Molecular Microbiology","topic":"Bacterial Genetics and Biotechnology","field":"Biochemistry, Genetics and Molecular Biology","cited_by":203,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"SickKids Foundation; Hospital for Sick Children; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Cystic Fibrosis Foundation","keywords":"Pilus; Periplasmic space; Biology; Flagellum; Bacterial conjugation; Fimbriae Proteins; Secretion; Cell envelope; Function (biology); Virulence; Bacteria; Cell biology; Computational biology; Microbiology; Escherichia coli; Genetics; Gene; Biochemistry","retraction":null,"screen_n_in":null,"score":{"opus":0.01276240747529528,"gpt":0.276006894292041,"spread":0.2632444868167457,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.0001738301,0.0004653053,0.001146449,0.0001956455,0.0000348697,0.000009956259,0.0005338066,0.00261975,0.00006274263],"category_scores_gemma":[0.00005380343,0.0004202442,0.0006137945,0.000181961,0.0002484591,0.000001087599,0.0002356322,0.0004096088,0.00005769203],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000309577,"about_ca_system_score_gemma":0.0001938115,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000004015714,"about_ca_topic_score_gemma":0.000004624183,"domain_scores_codex":[0.9979463,0.0002375021,0.000676965,0.0006951343,0.00004430651,0.0003997478],"domain_scores_gemma":[0.9985268,0.00001226384,0.0005063119,0.0008110366,0.00008554736,0.0000580583],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00001042373,0.00003462236,3.258141e-7,0.0004491713,0.0001886146,0.000005689156,8.806743e-7,6.427669e-7,0.6833119,0.00005747508,0.0005505143,0.3153897],"study_design_scores_gemma":[0.0001496655,0.0002820218,3.104896e-7,0.0002877309,0.0002671203,0.0001987175,0.000001561551,1.430498e-7,0.09964076,0.00001001882,0.8988104,0.0003515158],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.0005608145,0.9971569,0.0004830715,0.00003669879,0.0006000762,0.0004334854,0.0002678121,0.00001948669,0.0004416635],"genre_scores_gemma":[0.0003900949,0.9950572,0.001768835,0.00006940095,0.0003071318,0.0000359661,0.001893266,0.0000801655,0.0003980005],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.8982599,"threshold_uncertainty_score":0.9998249,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"evaluation","study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication","research_integrity"],"domain":"evaluation","study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2926567176","doi":"10.1136/bmjopen-2018-026516","title":"Knowledge and motivations of researchers publishing in presumed predatory journals: a survey","year":2019,"lang":"en","type":"article","venue":"BMJ Open","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":160,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo; University of Ottawa; Jewish General Hospital; Institut du Savoir Montfort; McGill University; Ottawa Hospital","funders":"Canadian Institutes of Health Research; Ottawa Hospital Anesthesia Alternate Funds Association","keywords":"Publishing; Medicine; Psychological intervention; Law; Psychiatry; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.9177965981725121,"gpt":0.7046737378033378,"spread":0.2131228603691744,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.1739013,0.00007749222,0.0003312395,0.03416999,0.00008212571,0.008267575,0.004144824,0.0001052637,0.0009398088],"category_scores_gemma":[0.2431802,0.00005554944,0.00003906314,0.1161056,0.0001075579,0.004209267,0.003837105,0.0003892387,0.0001154252],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007752397,"about_ca_system_score_gemma":0.0008272337,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002134968,"about_ca_topic_score_gemma":0.001305276,"domain_scores_codex":[0.9906144,0.00190393,0.0009143365,0.0005604162,0.005557302,0.0004496672],"domain_scores_gemma":[0.9776171,0.01597779,0.0003915449,0.001064246,0.004543257,0.0004060529],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00006925716,0.0000938103,0.9502512,0.00001184831,0.000007349843,0.000001551836,0.0003827097,0.00001574663,0.0001796146,0.0002351482,0.02334965,0.02540216],"study_design_scores_gemma":[0.0006909508,0.00005686073,0.9891419,0.00005437819,4.717402e-7,0.000001350397,0.0005031046,0.002807582,0.0000988395,0.001610964,0.004959528,0.00007401575],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9533338,0.0006387595,0.00007307276,0.0006674282,0.0001485,0.002743282,0.0000650462,0.000003218559,0.04232689],"genre_scores_gemma":[0.9908869,0.00006337643,0.0003441161,0.00002451333,0.00001569336,0.00005940539,0.000004024901,0.000007059421,0.008594934],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.08193564,"threshold_uncertainty_score":0.9999735,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2089501922","doi":"10.1177/0963662510384461","title":"Academic staff and public communication: a survey of popular science publishing across 13 countries","year":2010,"lang":"en","type":"article","venue":"Public Understanding of Science","topic":"Climate Change Communication and Perception","field":"Social Sciences","cited_by":159,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Publishing; Newspaper; Public opinion; Political science; Scientific publishing; Social science; Public relations; Sociology; Media studies; Library science; Law; Politics","retraction":null,"screen_n_in":null,"score":{"opus":0.5880613633378229,"gpt":0.4714212013915512,"spread":0.1166401619462717,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts","scholarly_communication"],"consensus_categories":["sts"],"category_scores_codex":[0.02393834,0.0001161917,0.00021641,0.0005909778,0.004543836,0.002190525,0.003150151,0.0001489298,0.0001513585],"category_scores_gemma":[0.01090549,0.0001174375,0.00002862774,0.004663763,0.03387401,0.008681203,0.0009104891,0.0005417857,0.000001979384],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004369404,"about_ca_system_score_gemma":0.001813661,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.006822334,"about_ca_topic_score_gemma":0.04973115,"domain_scores_codex":[0.9965827,0.0002528199,0.0004585473,0.0003711032,0.001625712,0.0007090934],"domain_scores_gemma":[0.996187,0.0004851818,0.0004896327,0.0007940534,0.001660549,0.0003836359],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","study_design_scores_codex":[0.000008453426,0.00008170586,0.2208374,0.00002828919,0.000004355897,7.950872e-8,0.08147538,4.286748e-7,0.006424665,0.6868927,0.000120138,0.004126458],"study_design_scores_gemma":[0.0009319595,0.000118219,0.3167082,0.0001779187,0.00001340786,0.000009770049,0.6331336,0.00124672,0.0005157638,0.02772743,0.01871856,0.0006983869],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9676279,0.000198001,0.0004754704,0.01473757,0.0003321641,0.0002156681,0.00005249087,0.00006378094,0.01629699],"genre_scores_gemma":[0.9971629,0.002216073,0.000456293,0.00007483323,0.00002362248,0.000005861642,0.00001372631,0.000007280381,0.00003944189],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6591653,"threshold_uncertainty_score":0.9997913,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4249889786","doi":"10.12688/f1000research.15256.1","title":"What is a predatory journal? A scoping review","year":2018,"lang":"en","type":"review","venue":"F1000Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":154,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Ottawa Hospital; University of Ottawa","funders":"Ottawa Hospital Anesthesia Alternate Funds Association; Ottawa Hospital Research Institute; University of Ottawa","keywords":"PsycINFO; Grey literature; MEDLINE; CINAHL; Web of science; Medicine; Information retrieval; Computer science; Biology","retraction":null,"screen_n_in":null,"score":{"opus":0.9152973410968032,"gpt":0.7496890964298902,"spread":0.1656082446669129,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.1526493,0.0006879686,0.003806303,0.1078312,0.0009572492,0.02663212,0.0164496,0.000695651,0.02420215],"category_scores_gemma":[0.1051354,0.0004125765,0.002047736,0.2761776,0.000774161,0.003172324,0.005682404,0.003014539,0.02809673],"about_ca_system_candidate":true,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006484403,"about_ca_system_score_gemma":0.008336073,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000009779443,"about_ca_topic_score_gemma":0.000005209153,"domain_scores_codex":[0.9498433,0.004448206,0.003775579,0.00223753,0.0374523,0.002243045],"domain_scores_gemma":[0.9553101,0.01898348,0.001865007,0.004794673,0.0165202,0.002526517],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00000159701,0.0000402071,0.000004013262,0.02608021,0.00004341976,0.00006166766,0.00002013719,6.281368e-9,2.590152e-8,0.000008139489,0.3715579,0.6021827],"study_design_scores_gemma":[0.00008043241,0.0001371202,0.000001225176,0.3154487,0.00005466002,0.0001965904,0.00003515115,0.000009269235,6.34924e-7,0.0004700719,0.6832628,0.0003033261],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[5.255653e-7,0.9891187,0.0001062777,0.0008828244,0.002619705,0.003716286,0.00009015624,0.00003979166,0.003425746],"genre_scores_gemma":[5.298901e-8,0.9639038,0.0002296602,0.0005779005,0.0011173,0.0002231885,0.00001602876,0.0001024554,0.03382955],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.6018794,"threshold_uncertainty_score":0.9998326,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","scholarly_communication"],"domain":"evaluation","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["research_integrity","scholarly_communication"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4244401793","doi":"10.1136/ebm.11.6.162-a","title":"Of studies, syntheses, synopses, summaries, and systems: the \"5S\" evolution of information services for evidence-based healthcare decisions","year":2006,"lang":"en","type":"article","venue":"Evidence-Based Medicine","topic":"Electronic Health Records Systems","field":"Health Professions","cited_by":151,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McMaster University","funders":"","keywords":"Computer science; Health care; Information system; Data science; Healthcare system; Real world evidence; Knowledge management; Information retrieval; Medicine; Engineering; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.1694791498193305,"gpt":0.4373667516424011,"spread":0.2678876018230706,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.01073963,0.0003766303,0.001276686,0.0004898841,0.0009640751,0.00001285161,0.0005016816,0.0003536931,0.00002347983],"category_scores_gemma":[0.01531904,0.0002372271,0.0001058012,0.000941756,0.0005293983,0.0008654734,0.00008095555,0.0005771557,0.000007976392],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00150285,"about_ca_system_score_gemma":0.005194651,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.05487271,"about_ca_topic_score_gemma":0.008105821,"domain_scores_codex":[0.9920223,0.00225459,0.003197612,0.0004087636,0.001220989,0.0008957735],"domain_scores_gemma":[0.9457588,0.04783053,0.002411547,0.0009032889,0.002870474,0.0002253453],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"systematic_review","study_design_scores_codex":[0.00867843,0.0004011415,0.4153732,0.3437712,0.0004728163,0.000008761508,0.01148305,0.004912131,0.003693707,0.09361184,0.08587935,0.03171439],"study_design_scores_gemma":[0.007689958,0.007483931,0.05546145,0.6836772,0.001099828,0.00001701768,0.06129482,0.03672558,0.0008192812,0.002661108,0.1420353,0.001034542],"study_design_candidate":"systematic_review","study_design_consensus":null,"genre_codex":"review","genre_gemma":"empirical","genre_scores_codex":[0.1064174,0.7393286,0.02637754,0.1109552,0.003444178,0.01281148,0.0003482019,0.0002210851,0.00009630222],"genre_scores_gemma":[0.990871,0.004739583,0.0005748216,0.001400279,0.0008551782,0.001356079,0.00006913455,0.00003367047,0.0001002492],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8844536,"threshold_uncertainty_score":0.9929754,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2889499003","doi":"10.1503/cmaj.180154","title":"How predatory journals leak into PubMed","year":2018,"lang":"en","type":"article","venue":"Canadian Medical Association Journal","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":146,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":true,"about_ca":false},"ca_institutions":"Ottawa Hospital","funders":"University of Ottawa","keywords":"Key (lock); Data science; Computer science; MEDLINE; World Wide Web; Computer security; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.3334433946008996,"gpt":0.4961553382486549,"spread":0.1627119436477553,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.08107378,0.0001390894,0.0003110713,0.03402918,0.001081469,0.01064864,0.002797762,0.0005501052,0.01038845],"category_scores_gemma":[0.3063317,0.00009917778,0.0002036654,0.06545956,0.0002076342,0.001307828,0.0001640679,0.001497262,0.001171165],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001971778,"about_ca_system_score_gemma":0.005237541,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007693563,"about_ca_topic_score_gemma":0.03113579,"domain_scores_codex":[0.9711182,0.0007254727,0.0008256675,0.0003913632,0.02564573,0.001293546],"domain_scores_gemma":[0.9762717,0.002183804,0.0006431378,0.0003524112,0.01287234,0.007676636],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000002213984,0.00001248318,0.1015512,5.924e-7,0.00002489375,0.0001193744,0.0001234219,2.134901e-7,0.00000590086,0.0001070481,0.4496251,0.4484275],"study_design_scores_gemma":[0.0003753556,0.00006154756,0.1490853,0.00001317011,0.000003368051,0.0001801671,0.0005085059,0.0004152021,0.00003005905,0.005552176,0.8436528,0.0001223041],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.41811,0.004612911,0.009230076,0.4997303,0.0150188,0.0004934419,0.00009336133,0.00005080419,0.0526603],"genre_scores_gemma":[0.9742511,0.000494603,0.0002422246,0.006159651,0.00583498,0.000007096136,0.000001337181,0.00001413167,0.01299491],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.556141,"threshold_uncertainty_score":0.9996065,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","scholarly_communication","research_integrity"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2167083241","doi":"10.1525/jer.2010.5.4.23","title":"Researchers' Perspectives on Collective/Community Co-Authorship in Community-Based Participatory Indigenous Research","year":2010,"lang":"en","type":"article","venue":"Journal of Empirical Research on Human Research Ethics","topic":"Mental Health and Patient Involvement","field":"Health Professions","cited_by":143,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Dalhousie University","funders":"Canadian Institutes of Health Research","keywords":"Indigenous; Community-based participatory research; Participatory action research; Citizen journalism; Thematic analysis; Sociology; Public relations; Value (mathematics); Qualitative research; Political science; Social science; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.9721388023156835,"gpt":0.7766190470881619,"spread":0.1955197552275216,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","sts","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","sts","research_integrity","insufficient_payload"],"category_scores_codex":[0.5064174,0.0004420521,0.00100109,0.008870157,0.02797791,0.000245241,0.004061491,0.002178831,0.001091743],"category_scores_gemma":[0.09583548,0.0003656672,0.0002754447,0.00601737,0.006127353,0.0003091239,0.001235926,0.2490152,0.0008444211],"about_ca_system_candidate":true,"about_ca_system_consensus":true,"about_ca_system_score_codex":0.009017,"about_ca_system_score_gemma":0.02715219,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007181549,"about_ca_topic_score_gemma":0.01635125,"domain_scores_codex":[0.7033116,0.2745714,0.002100617,0.000597438,0.01459729,0.004821596],"domain_scores_gemma":[0.7895505,0.1909528,0.0004292162,0.002170802,0.01395684,0.00293978],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.01455392,0.02158808,0.1580751,0.00459101,0.0001206924,0.0005297887,0.7379432,0.00005178028,0.004892328,0.01697019,0.03978676,0.0008971339],"study_design_scores_gemma":[0.009262708,0.06372354,0.1392913,0.008578548,0.00001039576,0.000006718061,0.607281,0.0001304724,0.003034831,0.1121344,0.05580376,0.0007423294],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9536884,0.0001763211,0.000004443382,0.01033011,0.0003301762,0.00365988,0.00006137898,0.00003547268,0.03171386],"genre_scores_gemma":[0.993802,0.0002749264,0.0001426507,0.001682727,0.0007813583,0.000637506,0.00004885977,0.0001179952,0.002511961],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4105819,"threshold_uncertainty_score":0.9999335,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"methods","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["research_integrity","scholarly_communication"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2811055236","doi":"10.12688/f1000research.15256.2","title":"What is a predatory journal? A scoping review","year":2018,"lang":"en","type":"review","venue":"F1000Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":141,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Ottawa Hospital; University of Ottawa","funders":"Ottawa Hospital Anesthesia Alternate Funds Association; Ottawa Hospital Research Institute; University of Ottawa","keywords":"PsycINFO; MEDLINE; Grey literature; Web of science; CINAHL; Medicine; Psychology; Meta-analysis; Political science; Pathology","retraction":null,"screen_n_in":null,"score":{"opus":0.9152973410968032,"gpt":0.7496890964298902,"spread":0.1656082446669129,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","insufficient_payload"],"category_scores_codex":[0.1526493,0.0006879686,0.003806303,0.1078312,0.0009572492,0.02663212,0.0164496,0.000695651,0.02420215],"category_scores_gemma":[0.1051354,0.0004125765,0.002047736,0.2761776,0.000774161,0.003172324,0.005682404,0.003014539,0.02809673],"about_ca_system_candidate":true,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006484403,"about_ca_system_score_gemma":0.008336073,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000009779443,"about_ca_topic_score_gemma":0.000005209153,"domain_scores_codex":[0.9498433,0.004448206,0.003775579,0.00223753,0.0374523,0.002243045],"domain_scores_gemma":[0.9553101,0.01898348,0.001865007,0.004794673,0.0165202,0.002526517],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00000159701,0.0000402071,0.000004013262,0.02608021,0.00004341976,0.00006166766,0.00002013719,6.281368e-9,2.590152e-8,0.000008139489,0.3715579,0.6021827],"study_design_scores_gemma":[0.00008043241,0.0001371202,0.000001225176,0.3154487,0.00005466002,0.0001965904,0.00003515115,0.000009269235,6.34924e-7,0.0004700719,0.6832628,0.0003033261],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[5.255653e-7,0.9891187,0.0001062777,0.0008828244,0.002619705,0.003716286,0.00009015624,0.00003979166,0.003425746],"genre_scores_gemma":[5.298901e-8,0.9639038,0.0002296602,0.0005779005,0.0011173,0.0002231885,0.00001602876,0.0001024554,0.03382955],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.6018794,"threshold_uncertainty_score":0.9998326,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity","scholarly_communication"],"domain":"evaluation","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","research_integrity"],"domain":"methods","study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2114059546","doi":"10.1371/journal.pone.0132557","title":"How Long Is Too Long in Contemporary Peer Review? Perspectives from Authors Publishing in Conservation Biology Journals","year":2015,"lang":"en","type":"article","venue":"PLoS ONE","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":133,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Carleton University","funders":"Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Publishing; Peer review; Psychology; Prestige; Medical education; Medicine; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.9005902623703669,"gpt":0.5770311754804135,"spread":0.3235590868899534,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.05285542,0.0001510766,0.0006341162,0.03258255,0.00007051982,0.0060812,0.00169602,0.0002105732,0.0002723464],"category_scores_gemma":[0.3054913,0.0001138513,0.00008443333,0.09215616,0.0001407435,0.004230018,0.0004963569,0.000729456,0.00007465547],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003042699,"about_ca_system_score_gemma":0.0004879683,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001544737,"about_ca_topic_score_gemma":0.0002892992,"domain_scores_codex":[0.9872732,0.0009475035,0.0009202887,0.000786347,0.009579585,0.000493126],"domain_scores_gemma":[0.9815771,0.004174719,0.0005702157,0.000722851,0.01240806,0.0005470389],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00002005873,0.0004298584,0.8860663,0.00001281124,0.00002418139,0.00003690425,0.001263011,4.50182e-7,0.0003536692,0.00007832016,0.107857,0.003857478],"study_design_scores_gemma":[0.001419378,0.0001610362,0.965643,0.0007661827,0.00001149283,0.000002686961,0.005903068,0.002132257,0.000921315,0.01619785,0.006521943,0.0003198074],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7590696,0.05525811,0.0003843568,0.182354,0.0001525056,0.0003910153,0.00006019462,0.00001456317,0.002315648],"genre_scores_gemma":[0.9893864,0.002638737,0.0008545988,0.001457885,0.0001219813,0.00001930853,0.00002646802,0.00001092441,0.00548363],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2526359,"threshold_uncertainty_score":0.9949506,"prediction_status":"machine_predicted_unvalidated"},"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","scholarly_communication"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2001314811","doi":"10.1038/nchembio0709-436","title":"Open access chemical and clinical probes to support drug discovery","year":2009,"lang":"en","type":"article","venue":"Nature Chemical Biology","topic":"Pharmacogenetics and Drug Metabolism","field":"Pharmacology, Toxicology and Pharmaceutics","cited_by":132,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Structural Genomics Consortium; University of Toronto","funders":"Canadian Institutes of Health Research; Wellcome Trust","keywords":"Drug discovery; Pharmaceutical industry; Productivity; Data science; Business; Drug industry; Computational biology; Computer science; Biology; Biotechnology; Bioinformatics; Economics","retraction":null,"screen_n_in":null,"score":{"opus":0.134351923613858,"gpt":0.5619974727123195,"spread":0.4276455490984615,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","research_integrity"],"consensus_categories":["research_integrity"],"category_scores_codex":[0.0009726431,0.0003523838,0.0006642779,0.00008946157,0.0001007943,0.0001813136,0.00224311,0.001389735,0.000446682],"category_scores_gemma":[0.0004870805,0.0002885174,0.0001467608,0.0002815676,0.0003460156,0.0003064798,0.001916398,0.002484904,0.00007566068],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003542399,"about_ca_system_score_gemma":0.0001333867,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000006378462,"about_ca_topic_score_gemma":0.000001946804,"domain_scores_codex":[0.9973547,0.0003024321,0.00061134,0.0009277374,0.0001134371,0.0006903628],"domain_scores_gemma":[0.9982563,0.0004858274,0.000136933,0.000309757,0.0001048558,0.0007063344],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0009919782,0.000704637,0.01416758,0.00001522682,0.0001395384,0.00002857764,0.0000732976,0.000001018624,0.7873629,0.005056832,0.05536571,0.1360927],"study_design_scores_gemma":[0.002005464,0.0001150356,0.002729224,0.000006801678,0.0001322516,0.00003150999,0.000006475084,0.00003343862,0.5688101,0.007195745,0.4185392,0.0003947646],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9767361,0.001414241,0.00002079866,0.01110415,0.001459136,0.0007347016,0.0001496914,0.00007400609,0.008307201],"genre_scores_gemma":[0.9497935,0.0006766926,0.0007257066,0.04703065,0.001094645,0.00003906109,0.0001442027,0.00002081119,0.0004747339],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3631735,"threshold_uncertainty_score":0.9999567,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2161075463","doi":"10.1186/1748-5908-7-2","title":"Use of health systems and policy research evidence in the health policymaking in eastern Mediterranean countries: views and practices of researchers","year":2012,"lang":"en","type":"article","venue":"Implementation Science","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":127,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McMaster University; McMaster University Medical Centre","funders":"","keywords":"Health services research; Health policy; Thematic analysis; Health informatics; Government (linguistics); Medicine; Health administration; Knowledge translation; Public health; Public relations; Political science; Qualitative research; Nursing; Sociology; Social science; Knowledge management","retraction":null,"screen_n_in":null,"score":{"opus":0.976980351265113,"gpt":0.8372784395400653,"spread":0.1397019117250476,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.1213672,0.0001208581,0.0003764101,0.001803827,0.001038719,0.00009730762,0.000545679,0.00004277288,0.00002734464],"category_scores_gemma":[0.01597134,0.00009226857,0.00001161902,0.0044436,0.001291146,0.003294982,0.0003137608,0.0005053451,0.000005413436],"about_ca_system_candidate":true,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008512756,"about_ca_system_score_gemma":0.007347952,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.1547654,"about_ca_topic_score_gemma":0.01640865,"domain_scores_codex":[0.9800111,0.01216969,0.002544478,0.0004678797,0.002799964,0.002006904],"domain_scores_gemma":[0.9816723,0.01459875,0.002308617,0.000452772,0.0005628225,0.0004047184],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00003042478,0.00002228208,0.6561137,0.0009953934,0.000001372272,2.197436e-7,0.3250011,0.000002238461,0.000265858,0.006481657,0.0007619355,0.01032385],"study_design_scores_gemma":[0.0007051075,0.0003202774,0.7342193,0.000951543,0.000001192878,0.000006206323,0.2448647,0.0003176885,0.00002847526,0.00009389983,0.01840583,0.00008584875],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9169549,0.001022413,0.0001092308,0.07838944,0.0001384685,0.003245042,0.00005508105,0.000006879017,0.00007848629],"genre_scores_gemma":[0.9913681,0.00145576,0.0004997901,0.006308803,0.0001141091,0.0002262249,0.000002059027,0.000008162181,0.00001698718],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1383567,"threshold_uncertainty_score":0.9982795,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gpt","categories":["metaresearch","scholarly_communication"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"grok","categories":["metaresearch","scholarly_communication"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"opus","categories":["metaresearch","scholarly_communication"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W1975575045","doi":"10.6017/ital.v26i3.3272","title":"The Structure and Form of Folksonomy Tags: The Road to the Public Library Catalog","year":2007,"lang":"en","type":"article","venue":"Information Technology and Libraries","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":114,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Dalhousie University","funders":"","keywords":"Folksonomy; Computer science; Interactivity; Spelling; World Wide Web; Information retrieval; Control (management); Library catalog; Linguistics; Artificial intelligence","retraction":null,"screen_n_in":null,"score":{"opus":0.005393257152832807,"gpt":0.1823269148058518,"spread":0.176933657653019,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.0004179005,0.00009618884,0.00009349106,0.0002879186,0.0008300444,0.001159096,0.001427412,0.0001091342,0.000005103921],"category_scores_gemma":[0.00007483504,0.00004254846,0.00001859678,0.001218486,0.0005161135,0.02352345,0.0007201367,0.0001644974,0.00001279096],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000003002816,"about_ca_system_score_gemma":0.00009016027,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000003602368,"about_ca_topic_score_gemma":0.000003974484,"domain_scores_codex":[0.999108,0.0000174419,0.0004081371,0.00008043292,0.0001624058,0.0002236304],"domain_scores_gemma":[0.9990645,0.000155904,0.0002167799,0.0004759825,0.0000383047,0.00004850281],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000003931632,0.000001096444,0.002726611,0.000006975685,0.000006758077,9.382375e-8,0.006894307,8.283859e-7,0.000002969729,0.8045982,0.00496756,0.1807907],"study_design_scores_gemma":[0.0001211735,0.00006775498,0.009853759,0.000008315967,0.000001093568,0.00005896268,0.008070939,0.001100649,0.004611066,0.04996188,0.9260475,0.00009690336],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.4524143,0.002344859,0.1623195,0.3623256,0.001223833,0.001672961,0.00005849715,0.0008929403,0.01674755],"genre_scores_gemma":[0.9949577,0.00004696627,0.001524319,0.003302938,0.00002636158,0.00001151791,0.000008849956,0.000002033116,0.0001192792],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9210799,"threshold_uncertainty_score":0.9998778,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2610741382","doi":"10.1016/j.wneu.2017.04.157","title":"The Spectrum of Altmetrics in Neurosurgery: The Top 100 “Trending” Articles in Neurosurgical Journals","year":2017,"lang":"en","type":"review","venue":"World Neurosurgery","topic":"Social Media in Health Education","field":"Social Sciences","cited_by":98,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Toronto Western Hospital; Hospital for Sick Children; University of Toronto","funders":"","keywords":"Altmetrics; Social media; Medicine; Neurosurgery; Citation; Mainstream; Bibliometrics; Outreach; Medical education; Library science; World Wide Web; Surgery; Computer science","retraction":null,"screen_n_in":null,"score":{"opus":0.4131373920694704,"gpt":0.4947529833668562,"spread":0.08161559129738588,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","sts"],"consensus_categories":[],"category_scores_codex":[0.01112648,0.0004498558,0.001906986,0.002003773,0.001379189,0.0006141363,0.00165349,0.0002960261,0.00008678757],"category_scores_gemma":[0.04069273,0.0003017741,0.0007207623,0.006777404,0.001082182,0.0002723176,0.0001956956,0.001929884,0.00002995275],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003836448,"about_ca_system_score_gemma":0.001954194,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001701645,"about_ca_topic_score_gemma":0.006112929,"domain_scores_codex":[0.9891219,0.005077949,0.002499805,0.0006366553,0.001360876,0.001302875],"domain_scores_gemma":[0.9434631,0.05267152,0.002417709,0.001081338,0.00008517458,0.0002811122],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0000164944,0.0001560857,0.04751999,0.0005016867,0.0000109047,0.00008779835,0.0008503961,0.000001763152,1.64292e-7,0.001644994,0.002096913,0.9471128],"study_design_scores_gemma":[0.00007796227,0.000011146,0.03832578,0.001829655,0.00005106505,0.00000607369,0.00007225524,0.00000202203,7.4909e-7,0.0003904096,0.9589963,0.0002366116],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.004219275,0.9585164,2.28687e-7,0.01123419,0.02240023,0.002147876,0.00001683648,0.0000452414,0.001419708],"genre_scores_gemma":[0.01051392,0.9865433,0.000002094019,0.0002170748,0.001685351,0.0005074731,0.000003483117,0.0000734612,0.0004538512],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.9568993,"threshold_uncertainty_score":0.9999434,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2625269005","doi":"10.11613/bm.2017.030","title":"Ethical issues in publishing in predatory journals","year":2017,"lang":"en","type":"review","venue":"Biochemia Medica","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":97,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Public Health Ontario; University of Toronto","funders":"","keywords":"Publishing; Misrepresentation; Publication; Deception; Public relations; Political science; Library science; Sociology; Law; Computer science","retraction":null,"screen_n_in":null,"score":{"opus":0.8941266353903872,"gpt":0.7234011886426125,"spread":0.1707254467477747,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","research_integrity"],"category_scores_codex":[0.1373145,0.0003975683,0.002532168,0.1038878,0.0001685755,0.0122657,0.01187921,0.003114029,0.001435708],"category_scores_gemma":[0.522408,0.0002551949,0.0005101176,0.1074857,0.0004509402,0.001125718,0.002496486,0.006580627,0.00045156],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005455957,"about_ca_system_score_gemma":0.003179752,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002103249,"about_ca_topic_score_gemma":0.00006388654,"domain_scores_codex":[0.9741449,0.00129049,0.00283465,0.001552011,0.01902187,0.001156033],"domain_scores_gemma":[0.9818915,0.01127158,0.001624653,0.002547512,0.001562822,0.001101993],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00000173114,0.00005011307,0.0005362217,0.0005061114,0.00001007741,0.0001456388,0.00004652319,2.385062e-8,6.428021e-7,0.00003117385,0.1395514,0.8591204],"study_design_scores_gemma":[0.000248944,0.0000204274,0.0003436936,0.006270661,0.00001089388,0.00002270825,0.00004853031,0.00001919521,0.000005418361,0.0008772882,0.9918603,0.0002719855],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.00004736704,0.986076,0.00001017941,0.003608942,0.001024155,0.0004252271,0.00005346555,0.00001412149,0.008740528],"genre_scores_gemma":[0.0001578872,0.996204,0.0002174515,0.0001418826,0.0006053997,0.0000538875,0.00002064615,0.00002710381,0.002571774],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.8588484,"threshold_uncertainty_score":0.99999,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"evaluation","study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","research_integrity","scholarly_communication"],"domain":"evaluation","study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3006156524","doi":"10.1136/bmjopen-2019-035561","title":"Defining predatory journals and responding to the threat they pose: a modified Delphi consensus process","year":2020,"lang":"en","type":"article","venue":"BMJ Open","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":97,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa; Ottawa Hospital","funders":"Social Sciences and Humanities Research Council of Canada; Canadian Institutes of Health Research; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; Ottawa Hospital Anesthesia Alternate Funds Association; Institute of Health Services and Policy Research; University of Ottawa; Institute of Musculoskeletal Health and Arthritis; Natural Sciences and Engineering Research Council of Canada; Ottawa Hospital Research Institute; National Science Foundation","keywords":"Outreach; Snowball sampling; Delphi method; Publishing; Checklist; Public relations; Library science; Medicine; Political science; Psychology; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.7858935230383207,"gpt":0.6547070806373642,"spread":0.1311864424009566,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.03954222,0.0001405398,0.0003389383,0.01377735,0.0006030073,0.008913535,0.004422141,0.00007675587,0.0002524096],"category_scores_gemma":[0.08963586,0.00007579236,0.0000602112,0.08829046,0.00008188947,0.0004916695,0.003158334,0.0003015747,0.0003109448],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003409904,"about_ca_system_score_gemma":0.0004762844,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006838201,"about_ca_topic_score_gemma":0.00003389104,"domain_scores_codex":[0.9920571,0.000536052,0.0006957619,0.0007127378,0.005506241,0.0004921317],"domain_scores_gemma":[0.9916403,0.005009608,0.0002962572,0.0006942268,0.00162327,0.0007363859],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"observational","study_design_scores_codex":[0.004967972,0.000191508,0.1973189,0.00008048205,0.0001811282,0.0004377181,0.03539332,0.004631078,0.003111247,0.0207831,0.3689229,0.3639806],"study_design_scores_gemma":[0.008940578,0.003167517,0.4326535,0.0007237735,0.000090552,0.0007423301,0.06685396,0.1473633,0.004403548,0.1005632,0.2316962,0.002801507],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9013997,0.001192291,0.0009614991,0.05913519,0.0002580162,0.004841887,0.00009786884,0.00002530522,0.03208828],"genre_scores_gemma":[0.9951742,0.00002938729,0.0008091661,0.002458923,0.0001013108,0.0001367069,7.719067e-7,0.00001307676,0.001276507],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3611791,"threshold_uncertainty_score":0.9974006,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","scholarly_communication"],"domain":"evaluation","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","research_integrity","scholarly_communication"],"domain":"methods","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2084033360","doi":"10.1007/s10826-013-9795-1","title":"Between Black and White: Examining Grey Literature in Meta-analyses of Psychological Research","year":2013,"lang":"en","type":"article","venue":"Journal of Child and Family Studies","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":90,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa","funders":"","keywords":"Grey literature; Meta-analysis; Inclusion (mineral); Psychology; Psychological intervention; Publication bias; Systematic review; White (mutation); Applied psychology; Social psychology; Clinical psychology; MEDLINE; Medicine; Political science; Psychiatry","retraction":null,"screen_n_in":null,"score":{"opus":0.9316437456358971,"gpt":0.638685197001198,"spread":0.2929585486346992,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.07085953,0.0002132537,0.004896821,0.0008633677,0.0001142931,0.0005531159,0.0005715534,0.00008353638,0.0001945886],"category_scores_gemma":[0.02317826,0.00007542533,0.0007986194,0.001348408,0.0003340953,0.000417721,0.0001874983,0.0004878016,0.00001849277],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000007880572,"about_ca_system_score_gemma":0.000009043357,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000003357013,"about_ca_topic_score_gemma":0.000002939614,"domain_scores_codex":[0.9849581,0.005774769,0.005796618,0.0004262525,0.002841963,0.0002022656],"domain_scores_gemma":[0.9877123,0.005231768,0.002865751,0.0006650062,0.003401427,0.0001237033],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00002361691,0.0001311789,0.8564004,0.0001585144,0.009978562,0.00002844085,0.009595902,0.00002259034,0.0003531467,0.0001782133,0.104884,0.01824548],"study_design_scores_gemma":[0.0002431553,0.0002560416,0.980742,0.000163964,0.00120945,0.00002278197,0.008643724,0.00001954241,0.000007965092,0.006513298,0.002096149,0.00008197316],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9073943,0.08655063,0.00004494228,0.002523168,0.00004755854,0.0002009879,0.000004660883,5.495164e-7,0.003233201],"genre_scores_gemma":[0.9930947,0.004729332,0.001291731,0.0002545683,0.000117133,0.000003977452,2.680674e-7,0.000004031015,0.0005041969],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1243416,"threshold_uncertainty_score":0.9850499,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch"],"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":"low"}],"label_agreement":"split"},{"id":"W4386541616","doi":"10.1002/leap.1578","title":"Risks of abuse of large language models, like <scp>ChatGPT</scp>, in scientific publishing: Authorship, predatory publishing, and paper mills","year":2023,"lang":"en","type":"article","venue":"Learned Publishing","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":87,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Thompson Rivers University","keywords":"Publishing; Mill; Scientific publishing; Key (lock); Public relations; Computer science; Political science; Law; Engineering; Computer security; Mechanical engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.2084196632622204,"gpt":0.4104631433775228,"spread":0.2020434801153024,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.008828962,0.0002687126,0.0005862408,0.001634216,0.00023406,0.003680201,0.0006471984,0.000745654,0.00008204586],"category_scores_gemma":[0.01905698,0.0002656536,0.0001344357,0.003156042,0.0002616288,0.017544,0.0002742603,0.001698549,0.00001805895],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001478321,"about_ca_system_score_gemma":0.0006580333,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00893541,"about_ca_topic_score_gemma":0.002423481,"domain_scores_codex":[0.9955184,0.0003646541,0.001179526,0.0008339625,0.001141576,0.0009618078],"domain_scores_gemma":[0.9960217,0.001137769,0.0005504957,0.0009025291,0.0009051728,0.0004823552],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.000150588,0.001221345,0.2168823,0.003019982,0.0001633088,0.00007502364,0.4531141,0.001061479,0.008363557,0.01650324,0.2211933,0.07825174],"study_design_scores_gemma":[0.002885084,0.0007100066,0.1460057,0.004761049,0.000332299,0.00008002915,0.4822993,0.07179786,0.03007465,0.06000087,0.1999984,0.001054763],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9833472,0.002151461,0.00008980657,0.005415246,0.001282449,0.0005969223,0.0000952884,0.0002121178,0.006809464],"genre_scores_gemma":[0.9936768,0.0002691052,0.0004024411,0.0007351103,0.0003282057,0.00004488059,0.0005887766,0.00006487169,0.003889746],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.07719698,"threshold_uncertainty_score":0.9999796,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"methods","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","research_integrity","scholarly_communication"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2044326206","doi":"10.1007/s11192-008-0220-2","title":"A new look at evidence of scholarly citation in citation indexes and from web sources","year":2007,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":85,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Western University","funders":"","keywords":"Citation; Web of science; Library science; Citation analysis; Webometrics; Citation impact; Bibliometrics; Scholarly communication; Computer science; Sample (material); Impact factor; World Wide Web; Political science; MEDLINE; Publishing","retraction":null,"screen_n_in":null,"score":{"opus":0.5726813817650799,"gpt":0.5567807477511177,"spread":0.01590063401396213,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.06158305,0.0001795445,0.0003808583,0.2483297,0.0002090949,0.003611983,0.001883733,0.0002070862,0.0002956545],"category_scores_gemma":[0.1921769,0.0001440193,0.00009545132,0.5391781,0.0002872671,0.003999108,0.0009399208,0.0003335925,0.0001327346],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003777984,"about_ca_system_score_gemma":0.0004307377,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008616144,"about_ca_topic_score_gemma":0.0006853588,"domain_scores_codex":[0.9797834,0.0002406213,0.001294367,0.00108498,0.01684294,0.0007536688],"domain_scores_gemma":[0.9724427,0.02094738,0.0006786906,0.0007354334,0.004372299,0.0008234766],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00005946088,0.00004929365,0.8260859,0.000006706121,0.000004549861,0.000005882124,0.0008868973,0.00003081033,0.01218201,0.0002786964,0.001110754,0.1592991],"study_design_scores_gemma":[0.0007069465,0.0001980155,0.9824387,0.0000535971,0.000005681039,0.000001966001,0.0008031926,0.001849006,0.005668396,0.0061341,0.001961292,0.0001791179],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9728765,0.00912236,0.01575318,0.0004166511,0.0005281832,0.0002774931,0.00002254843,0.00001815375,0.0009849267],"genre_scores_gemma":[0.9885912,0.0006787465,0.009100274,0.0000812181,0.00006168662,0.000002291321,0.000003677136,0.000009933621,0.001471007],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2908484,"threshold_uncertainty_score":0.9974223,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2942452906","doi":"10.12927/hcpol.2019.25792","title":"Integrated Knowledge Translation with Public Health Policy Makers: A Scoping Review","year":2019,"lang":"en","type":"review","venue":"Healthcare policy","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":80,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"Izaak Walton Killam Health Centre; Dalhousie University","funders":"","keywords":"Operationalization; Knowledge translation; Knowledge management; Public health; Process (computing); Public engagement; Political science; Public relations; Medical education; Process management; Psychology; Business; Medicine; Computer science; Nursing","retraction":null,"screen_n_in":null,"score":{"opus":0.9177527360035942,"gpt":0.7631348038759823,"spread":0.1546179321276119,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","sts","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow"],"category_scores_codex":[0.009696743,0.001434553,0.00631064,0.003530257,0.002037601,0.00009541939,0.001301792,0.001009631,0.0005057713],"category_scores_gemma":[0.005012427,0.001104343,0.0005770675,0.01113304,0.0002662367,0.0006330569,0.0002704527,0.003619469,0.004654639],"about_ca_system_candidate":true,"about_ca_system_consensus":true,"about_ca_system_score_codex":0.009376298,"about_ca_system_score_gemma":0.270305,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02765785,"about_ca_topic_score_gemma":0.01727248,"domain_scores_codex":[0.9697856,0.01592114,0.006416752,0.001804161,0.00125957,0.004812778],"domain_scores_gemma":[0.9829263,0.00556445,0.005026397,0.002264728,0.001334917,0.002883181],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000003705072,0.0000217245,0.000008137725,0.4400303,0.0000302853,0.000001412247,0.003161841,2.232631e-8,5.394156e-9,0.002671424,0.004216847,0.5498543],"study_design_scores_gemma":[0.0004832947,0.00020355,0.00000427391,0.4163748,0.000058517,0.00003389223,0.0002953795,0.00000223333,5.280538e-9,0.00001823832,0.5820653,0.0004605763],"study_design_candidate":"systematic_review","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[2.430668e-7,0.7721343,0.0003138318,0.1999845,0.0005044134,0.02222573,0.0009289781,0.0004466716,0.003461378],"genre_scores_gemma":[0.000003590858,0.8766513,0.0009928799,0.1130353,0.002267613,0.004486782,0.0009153836,0.0003154665,0.001331621],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.5778484,"threshold_uncertainty_score":0.9998404,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","metaresearch"],"domain":"methods","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2219837050","doi":"10.1177/1476750315616684","title":"Transformative research as knowledge mobilization: Transmedia, bridges, and layers","year":2015,"lang":"en","type":"article","venue":"Action Research","topic":"Participatory Visual Research Methods","field":"Social Sciences","cited_by":78,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"University of Manitoba","funders":"","keywords":"Transformative learning; Participatory action research; Mainstream; Knowledge transfer; Sociology; Citizen journalism; Positivism; Action research; Mobilization; New media; Knowledge management; Public relations; Political science; Pedagogy","retraction":null,"screen_n_in":null,"score":{"opus":0.962218324450557,"gpt":0.7998834208303206,"spread":0.1623349036202364,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts"],"consensus_categories":["metaresearch","sts"],"category_scores_codex":[0.05099825,0.0001101177,0.0001841133,0.0007567924,0.001677439,0.0003155692,0.0004387271,0.0002120331,0.0003932603],"category_scores_gemma":[0.01014868,0.0001086867,0.00004173749,0.002514998,0.002800483,0.0009164697,0.00009505666,0.001258855,0.0007761148],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008245715,"about_ca_system_score_gemma":0.002936574,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006405238,"about_ca_topic_score_gemma":0.00454496,"domain_scores_codex":[0.9813911,0.01337363,0.0002578296,0.0004202554,0.003303457,0.001253721],"domain_scores_gemma":[0.9924582,0.003320733,0.00002404962,0.000268803,0.002647421,0.001280791],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0008640235,0.0008248897,0.005946948,0.0001947591,0.00008552004,0.00005246338,0.5236384,0.00003208385,0.005436298,0.1147979,0.05248011,0.2956467],"study_design_scores_gemma":[0.0008696056,0.0006330399,0.002062084,0.00004505284,0.000005015751,0.000003755787,0.1839511,0.0008418743,0.008113065,0.01212669,0.7911571,0.0001915763],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7185558,0.001234291,0.001655446,0.007883222,0.0003614354,0.001626154,0.000008725058,0.0001484254,0.2685265],"genre_scores_gemma":[0.9882405,0.001289123,0.0002646066,0.00002706759,0.0003938559,0.0002965603,0.000004698783,0.00002269745,0.009460846],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.738677,"threshold_uncertainty_score":0.9999133,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["sts"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2119100170","doi":"10.1071/he13101","title":"Closing the chasm between research and practice: evidence of and for change","year":2014,"lang":"en","type":"article","venue":"Health Promotion Journal of Australia","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":77,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Cancer Council NSW; Cancer Council South Australia; Monash University; Canadian Health Services Research Foundation","keywords":"Population health; Bridging (networking); Relevance (law); Public relations; Psychological intervention; Context (archaeology); Health promotion; Promotion (chess); Closing (real estate); Community health; Public health; Evidence-based practice; Dissemination; Medicine; Political science; Alternative medicine; Nursing; Computer science; Law; Geography; Computer security","retraction":null,"screen_n_in":null,"score":{"opus":0.9753711977544747,"gpt":0.8179584552076156,"spread":0.1574127425468591,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.05046875,0.00008530069,0.0003147728,0.0002377379,0.001160212,0.00002064425,0.0001811071,0.0001198508,0.00003387657],"category_scores_gemma":[0.0119603,0.00005879627,0.00002380285,0.000351941,0.0002666923,0.000716347,0.00006705184,0.0009822316,0.000005528983],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001437036,"about_ca_system_score_gemma":0.0006410443,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003492024,"about_ca_topic_score_gemma":0.00004698402,"domain_scores_codex":[0.9924271,0.004776276,0.001357365,0.0001637936,0.0006583432,0.0006171588],"domain_scores_gemma":[0.9838012,0.01236827,0.001761308,0.0001729523,0.001503541,0.0003927136],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"observational","study_design_scores_codex":[0.001232998,0.0003699813,0.3133735,0.02781457,0.0001197145,0.000003671683,0.3870258,0.00000314513,0.002024407,0.0350474,0.090238,0.1427469],"study_design_scores_gemma":[0.003837233,0.01653393,0.7893083,0.00868584,0.00006478326,0.0001418374,0.02116062,0.000162048,0.0002204095,0.0121137,0.1475179,0.0002533815],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5913489,0.000182646,0.002935153,0.4032719,0.0002891328,0.001932488,0.00001533852,0.000005570644,0.00001887455],"genre_scores_gemma":[0.9911797,0.0004398184,0.004615507,0.002569417,0.0009984445,0.00006625652,6.360038e-7,0.00001026017,0.0001199384],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4759348,"threshold_uncertainty_score":0.9963624,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gpt","categories":["metaresearch","scholarly_communication"],"domain":"methods","study_design":"design_other","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"grok","categories":["metaresearch"],"domain":"methods","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"opus","categories":["metaresearch"],"domain":"methods","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2796216221","doi":"10.1177/0963662518761733","title":"Identifying diffusion patterns of research articles on Twitter: A case study of online engagement with open access articles","year":2018,"lang":"en","type":"article","venue":"Public Understanding of Science","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":76,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa; Simon Fraser University","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"Open science; Social media; Science communication; Sociology; Scholarly communication; Citizen science; Ask price; World Wide Web; Altmetrics; Online community; Public engagement; Data science; Public relations; Computer science; Political science; Biology; Science education; Publishing; Business","retraction":null,"screen_n_in":null,"score":{"opus":0.50896915729909,"gpt":0.496626232536117,"spread":0.01234292476297294,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002881206,0.00009728035,0.000231788,0.0007436593,0.0005672697,0.0004536173,0.001341408,0.000009673142,0.0001286192],"category_scores_gemma":[0.00003649474,0.00007493598,0.0000300079,0.002437967,0.001170432,0.0008551654,0.001680405,0.0001352011,3.589174e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001150339,"about_ca_system_score_gemma":0.0001372697,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005020816,"about_ca_topic_score_gemma":0.001261922,"domain_scores_codex":[0.9978277,0.0001657172,0.0003713213,0.0003587165,0.0009465833,0.0003299293],"domain_scores_gemma":[0.9984764,0.0001961748,0.0002840087,0.0004961622,0.0004572796,0.00009000579],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"qualitative","study_design_scores_codex":[0.00006484748,0.00308025,0.9437094,0.00003387287,0.00008888314,0.00001453184,0.01517381,0.00002697074,0.01169206,0.02102828,0.00006047397,0.005026561],"study_design_scores_gemma":[0.002666369,0.004180921,0.08483757,0.0008741533,0.0001098891,0.00001165853,0.7495735,0.0130032,0.1070031,0.03728257,0.00001963811,0.0004374775],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.967214,0.000001843534,0.03166658,0.000146102,0.00001781885,0.0004348508,0.000005407892,0.00001310858,0.0005002314],"genre_scores_gemma":[0.9992802,7.697605e-7,0.0006444073,0.000006315426,0.00003245316,0.00001299439,0.000001083108,0.000007875936,0.00001390155],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8588719,"threshold_uncertainty_score":0.7590005,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2160349994","doi":"10.1186/1478-4505-13-10","title":"Developing and refining the methods for a ‘one-stop shop’ for research evidence about health systems","year":2015,"lang":"en","type":"article","venue":"Health Research Policy and Systems","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":75,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Ottawa Hospital; Western University","funders":"Canadian Institutes of Health Research","keywords":"Health services research; Public health; Health administration; Medicine; Health policy; Refining (metallurgy); Health informatics; Nursing","retraction":null,"screen_n_in":null,"score":{"opus":0.9912761006432657,"gpt":0.8676079469251584,"spread":0.1236681537181072,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.3264294,0.0002815917,0.001006919,0.001291473,0.01091909,0.000406406,0.0007745424,0.0002844284,0.000002451624],"category_scores_gemma":[0.1047406,0.0002089954,0.00004936869,0.002340634,0.0007221573,0.0004689608,0.0005926258,0.001790168,0.00002939402],"about_ca_system_candidate":true,"about_ca_system_consensus":true,"about_ca_system_score_codex":0.00411164,"about_ca_system_score_gemma":0.04025517,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0615779,"about_ca_topic_score_gemma":0.001716769,"domain_scores_codex":[0.9428635,0.04541701,0.002850219,0.001107527,0.002298664,0.005463082],"domain_scores_gemma":[0.8872378,0.1026434,0.0009055563,0.001018762,0.005198801,0.002995658],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0003930022,0.00002747272,0.00183289,0.03364774,0.00003902015,7.590907e-7,0.135379,0.00002037398,0.00005398919,0.6057883,0.1714796,0.0513378],"study_design_scores_gemma":[0.001580103,0.001228157,0.001124394,0.005134623,0.000002616016,0.00001931222,0.07631797,0.009024828,0.000002603361,0.003094743,0.9022591,0.0002115853],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.007809061,0.04089993,0.1416208,0.7739705,0.002044967,0.0318154,0.0004107687,0.0002052046,0.001223444],"genre_scores_gemma":[0.4775782,0.03662197,0.158694,0.1263265,0.02455236,0.1369604,0.0001080636,0.0006842675,0.03847434],"genre_candidate":"commentary","genre_consensus":null,"teacher_disagreement_score":0.7307794,"threshold_uncertainty_score":0.9997114,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"methods","study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication"],"domain":"methods","study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W3098964464","doi":"10.1186/s12961-020-00600-1","title":"Applying systems thinking to knowledge mobilisation in public health","year":2020,"lang":"en","type":"letter","venue":"Health Research Policy and Systems","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":75,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Calgary; Simon Fraser University","funders":"National Health and Medical Research Council; Tasmanian Department of Health; Medical Research Council; NSW Ministry of Health; Australian Government","keywords":"Systems thinking; Context (archaeology); Knowledge translation; Knowledge management; Action (physics); Situated; Public health; Critical systems thinking; Complex adaptive system; Management science; Health services research; Computer science; Sociology; Public relations; Political science; Medicine; Engineering; Critical thinking; Artificial intelligence","retraction":null,"screen_n_in":null,"score":{"opus":0.9346946291483899,"gpt":0.7392747966649148,"spread":0.1954198324834751,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","sts","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.07338784,0.0006335236,0.002075264,0.005104409,0.005526537,0.0005521128,0.001273054,0.001041024,0.00003356393],"category_scores_gemma":[0.01798246,0.0006067518,0.00008244524,0.005921899,0.0002828427,0.0004627375,0.000911883,0.01039156,0.001265005],"about_ca_system_candidate":true,"about_ca_system_consensus":true,"about_ca_system_score_codex":0.01194427,"about_ca_system_score_gemma":0.04719879,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.1154505,"about_ca_topic_score_gemma":0.003703585,"domain_scores_codex":[0.9209155,0.05835035,0.005680392,0.00207821,0.004366818,0.00860872],"domain_scores_gemma":[0.9754865,0.01660798,0.001598516,0.001226273,0.001454018,0.003626765],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000009654891,0.00001765142,0.000778274,0.02450293,0.00001170408,0.00001998611,0.0908573,0.00001012199,0.000001656303,0.007002151,0.8742334,0.002555193],"study_design_scores_gemma":[0.0007168852,0.0004006924,0.0004634431,0.003940989,0.000001040163,0.00001415699,0.02073199,0.001317307,4.652306e-8,0.0002568224,0.9718055,0.0003511062],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"commentary","genre_gemma":"commentary","genre_scores_codex":[0.0003599103,0.002297821,0.0005652314,0.9631823,0.002057871,0.02651053,0.0005039502,0.0002592631,0.004263142],"genre_scores_gemma":[0.02205896,0.001161924,0.0001072956,0.9320317,0.01772836,0.02274918,0.0002183524,0.0001767592,0.003767487],"genre_candidate":"commentary","genre_consensus":"commentary","teacher_disagreement_score":0.1117469,"threshold_uncertainty_score":0.9996384,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W1588909286","doi":"10.1080/08989620590957175","title":"Haunted Manuscripts: Ghost Authorship in the Medical Literature","year":2005,"lang":"en","type":"article","venue":"Accountability in Research","topic":"Pharmaceutical industry and healthcare","field":"Pharmacology, Toxicology and Pharmaceutics","cited_by":75,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto; Queen's University; Institute for Clinical Evaluative Sciences","funders":"","keywords":"Credibility; Conflict of interest; Accountability; Psychology; Neutrality; Clinical trial; Public relations; Political science; Law; Medicine","retraction":null,"screen_n_in":null,"score":{"opus":0.8261096242099695,"gpt":0.7045818707680983,"spread":0.1215277534418712,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","research_integrity","insufficient_payload"],"consensus_categories":["research_integrity"],"category_scores_codex":[0.03432653,0.0002426459,0.0003180834,0.0003882947,0.0003948528,0.000119093,0.001531168,0.001858722,0.006836003],"category_scores_gemma":[0.003957538,0.0001785863,0.00008229762,0.002200898,0.0009073133,0.0003924698,0.0002771817,0.01689769,0.0005280765],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005197475,"about_ca_system_score_gemma":0.00049232,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007046929,"about_ca_topic_score_gemma":0.005496151,"domain_scores_codex":[0.9871873,0.008462169,0.000761406,0.0006318362,0.001401137,0.001556169],"domain_scores_gemma":[0.9937984,0.004660639,0.00004547652,0.000593142,0.0003236327,0.0005786509],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"not_applicable","study_design_scores_codex":[0.003156754,0.005259734,0.8003356,0.0007300302,0.00005722278,0.001432952,0.03219696,0.0001043164,0.0007604074,0.01866697,0.05388927,0.08340971],"study_design_scores_gemma":[0.002812216,0.0001635892,0.1926813,0.0002487519,0.00001361503,0.0001415948,0.00479177,0.005919864,0.001110547,0.01010797,0.781616,0.0003927413],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8451093,0.001776189,0.000001168905,0.1365994,0.0003669391,0.0009717062,0.00005424096,0.00004199779,0.01507904],"genre_scores_gemma":[0.9813944,0.0004286747,0.00001963349,0.01652098,0.000811865,0.0002948576,0.00004629389,0.00001675305,0.0004665912],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7277268,"threshold_uncertainty_score":0.9994371,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["research_integrity","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W3166045557","doi":"10.1186/s13643-021-01733-2","title":"Dealing with predatory journal articles captured in systematic reviews","year":2021,"lang":"en","type":"letter","venue":"Systematic Reviews","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":73,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Ottawa Public Health; University of Ottawa; McGill University; Ottawa Hospital","funders":"","keywords":"Systematic review; Medicine; MEDLINE; Protocol (science); Health care; Peer review; Alternative medicine; Family medicine; Pathology","retraction":null,"screen_n_in":null,"score":{"opus":0.6956962901616436,"gpt":0.4789212342135676,"spread":0.216775055948076,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","metaepi_narrow","metaepi_broad","insufficient_payload"],"category_scores_codex":[0.4841305,0.002491905,0.05974738,0.001913323,0.000353179,0.005570516,0.006777234,0.001110448,0.007079146],"category_scores_gemma":[0.2566205,0.0009285524,0.01031699,0.003990494,0.000138609,0.0007370506,0.0003202227,0.0035758,0.01508175],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004876833,"about_ca_system_score_gemma":0.0004394086,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001554734,"about_ca_topic_score_gemma":0.0001401407,"domain_scores_codex":[0.644936,0.2415301,0.09223322,0.003344249,0.01651797,0.001438427],"domain_scores_gemma":[0.8575884,0.0240444,0.09479874,0.01923543,0.003749213,0.0005837874],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000001317445,0.00002346417,0.00008222247,0.5130781,0.0004489168,0.0005486399,0.0006755829,0.00001167409,0.000003222146,0.00005189954,0.4849653,0.0001096706],"study_design_scores_gemma":[0.0002542276,0.00004457154,0.000006473944,0.3885653,0.00292436,0.002128106,0.0008664678,0.0004058942,0.000001633261,0.0005630572,0.6033633,0.0008766351],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"commentary","genre_scores_codex":[0.0001615175,0.8112237,0.02274858,0.05507779,0.003116369,0.1008147,0.00004659493,0.00003377018,0.006776995],"genre_scores_gemma":[0.003336054,0.05326024,0.04778257,0.5993811,0.01242661,0.04872129,0.0003560355,0.001293331,0.2334428],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.7579635,"threshold_uncertainty_score":0.9995944,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","scholarly_communication"],"domain":"methods","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","research_integrity"],"domain":"methods","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2196012223","doi":"10.1186/s12961-015-0061-4","title":"Building a knowledge translation platform in Malawi to support evidence-informed health policy","year":2015,"lang":"en","type":"article","venue":"Health Research Policy and Systems","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":71,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"Department of Medicine, University of Toronto; National Commission for Science and Technology; Universidade Federal de Minas Gerais; International Development Research Centre; University of Toronto; World Health Organization","keywords":"Knowledge translation; Health services research; Health administration; Health policy; Public health; Social policy; Health informatics; Medicine; Healthcare policy; Animal ecology; Translation (biology); Political science; Nursing; Health care reform; Knowledge management; Computer science; Law; Ecology","retraction":null,"screen_n_in":null,"score":{"opus":0.9682649455052706,"gpt":0.7986569621935815,"spread":0.1696079833116891,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","sts"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.06193988,0.0002964094,0.0008698091,0.004068709,0.001963127,0.0001272887,0.0005430888,0.0002576409,0.00003156826],"category_scores_gemma":[0.02638085,0.0002721773,0.00004467246,0.00547874,0.0001931103,0.0009528382,0.000292062,0.001414726,0.0006840241],"about_ca_system_candidate":true,"about_ca_system_consensus":true,"about_ca_system_score_codex":0.009542589,"about_ca_system_score_gemma":0.08419628,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.1904004,"about_ca_topic_score_gemma":0.03781443,"domain_scores_codex":[0.9824895,0.006987938,0.003062225,0.0007416326,0.002026931,0.004691747],"domain_scores_gemma":[0.9821385,0.01021897,0.0005664226,0.0006754812,0.001040757,0.005359849],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0004054523,0.0001004778,0.01470118,0.0122371,0.00001297334,0.000008408029,0.4899458,0.00007906,0.00004486665,0.1776454,0.1985308,0.1062885],"study_design_scores_gemma":[0.002994977,0.002391996,0.01252416,0.003931655,0.000001023205,0.00003005086,0.03648499,0.001734365,0.000003886524,0.001646686,0.9379133,0.0003428898],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.2076757,0.002889771,0.002079783,0.7377219,0.001449645,0.02224901,0.0003500811,0.0003446108,0.02523956],"genre_scores_gemma":[0.957592,0.001201785,0.001231955,0.03275011,0.002544513,0.002053034,0.00001568851,0.00006025858,0.00255068],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7499163,"threshold_uncertainty_score":0.9999731,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gpt","categories":["metaresearch","scholarly_communication"],"domain":"methods","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"grok","categories":[],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"opus","categories":["metaresearch","scholarly_communication"],"domain":"reporting","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2617385918","doi":"10.1016/j.physio.2017.05.002","title":"Strategies to translate knowledge related to common musculoskeletal conditions into physiotherapy practice: a systematic review","year":2017,"lang":"en","type":"review","venue":"Physiotherapy","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":69,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Alberta; University of Ottawa","funders":"","keywords":"Physical therapy; Medicine; Physical medicine and rehabilitation; Systematic review; Musculoskeletal pain; MEDLINE","retraction":null,"screen_n_in":null,"score":{"opus":0.3720173884773054,"gpt":0.7552179448342382,"spread":0.3832005563569328,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","sts","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.004302979,0.001128797,0.006003268,0.0006486065,0.002629173,0.0001828297,0.002118901,0.0005009892,0.001158673],"category_scores_gemma":[0.001224444,0.000881193,0.0009658181,0.001749389,0.0001621055,0.001006134,0.0001509313,0.001433822,0.01348597],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009465001,"about_ca_system_score_gemma":0.00438007,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008057102,"about_ca_topic_score_gemma":0.0004738279,"domain_scores_codex":[0.984162,0.008035202,0.00418225,0.001231346,0.0009093502,0.001479904],"domain_scores_gemma":[0.9873006,0.004932904,0.003196292,0.002893504,0.000690843,0.0009858332],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00001162876,0.0001184133,1.774527e-7,0.8960375,0.0003620199,0.000004425332,0.01465024,7.974654e-7,0.00002493086,0.001650099,0.02842455,0.05871522],"study_design_scores_gemma":[0.0003606683,0.0001474853,0.000002777829,0.379091,0.0004549986,0.0000050633,0.0005936793,0.000001401458,1.83453e-7,0.0003093407,0.618497,0.0005363509],"study_design_candidate":"systematic_review","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.000008504341,0.95647,0.0002072598,0.009462573,0.0009371428,0.02722905,0.0003688546,0.0003195164,0.004997093],"genre_scores_gemma":[0.00002173659,0.9600846,0.0006472152,0.02253515,0.0003650945,0.01360242,0.0001453101,0.0002675029,0.002331027],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.5900725,"threshold_uncertainty_score":0.9997544,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gpt","categories":["metaresearch","scholarly_communication"],"domain":"methods","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"grok","categories":[],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"opus","categories":[],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2091144419","doi":"10.1016/j.lisr.2005.08.004","title":"Affective issues in library and information science systems work: A content analysis","year":2005,"lang":"en","type":"article","venue":"Library & Information Science Research","topic":"Knowledge Management and Sharing","field":"Social Sciences","cited_by":68,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Western University; University of Alberta","funders":"","keywords":"Affect (linguistics); Work (physics); Content analysis; Psychology; Information system; Search engine indexing; Information science; Computer science; Library science; Applied psychology; World Wide Web; Sociology; Political science; Social science; Engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.08864920682039217,"gpt":0.3870816490747498,"spread":0.2984324422543577,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["bibliometrics","sts","scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.006663323,0.0001066718,0.0001833061,0.006265215,0.001696057,0.005624843,0.001170723,0.00006257604,0.0001578205],"category_scores_gemma":[0.0006355058,0.00009574615,0.00004016405,0.02445631,0.002323672,0.2535383,0.0008334639,0.000284068,0.0002455568],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001439108,"about_ca_system_score_gemma":0.000736706,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002156077,"about_ca_topic_score_gemma":0.00001767963,"domain_scores_codex":[0.9960424,0.0002185798,0.0004834971,0.0002328967,0.002185384,0.0008371875],"domain_scores_gemma":[0.998845,0.0002210733,0.0001292518,0.000270399,0.0002350081,0.000299237],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00004804779,0.00004378954,0.1775164,0.00005603614,0.00001980113,0.000001008258,0.08292437,0.0007660976,0.00003092984,0.669885,0.000855224,0.06785333],"study_design_scores_gemma":[0.000558247,0.00008395691,0.3785833,0.000152705,0.00001566533,6.746104e-7,0.09047361,0.05022523,0.000714652,0.0003701104,0.4784033,0.0004185998],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.4808702,0.0002353055,0.0001203479,0.004664037,0.0001466472,0.0008236389,0.000004399007,0.0001536665,0.5129818],"genre_scores_gemma":[0.9967145,0.0009314749,0.0003905175,0.000158384,0.00009423395,0.00005514822,0.00001322859,0.000003123729,0.001639402],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.6695148,"threshold_uncertainty_score":0.9996036,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2125399358","doi":"10.1002/asi.22801","title":"Are e‐books replacing print books? tradition, serendipity, and opportunity in the adoption and use of e‐books for historical research and teaching","year":2013,"lang":"en","type":"article","venue":"Journal of the American Society for Information Science and Technology","topic":"Library Collection Development and Digital Resources","field":"Computer Science","cited_by":65,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Western University","funders":"University of Warwick","keywords":"Serendipity; Persuasion; Transparency (behavior); Process (computing); Sociology; Grounded theory; Scholarly communication; Computer science; Qualitative research; Psychology; Social science; Epistemology; Publishing; Social psychology; Political science; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.09103166340541005,"gpt":0.2980196830509568,"spread":0.2069880196455468,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001841984,0.00005522858,0.0001310693,0.0003286609,0.000553139,0.0004202591,0.0003326367,0.00003133763,1.18208e-7],"category_scores_gemma":[0.0006730176,0.00003337002,0.00003149454,0.0004844916,0.0008779413,0.00375394,0.0002125463,0.0002094323,2.826763e-8],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008042999,"about_ca_system_score_gemma":0.0001584882,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002294168,"about_ca_topic_score_gemma":0.000001700662,"domain_scores_codex":[0.9991324,0.00002473906,0.0003031462,0.00008991215,0.00031303,0.0001367744],"domain_scores_gemma":[0.9986233,0.0002751626,0.0005619713,0.0001209195,0.0003670302,0.00005160952],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0000976384,0.0001174464,0.08053091,0.0002086864,0.00005989201,6.743895e-7,0.02630575,0.0000139303,0.002266426,0.2498755,0.02449058,0.6160325],"study_design_scores_gemma":[0.002969226,0.003648475,0.4703629,0.0003735042,0.00003687302,0.001010903,0.08241104,0.04004663,0.003415145,0.1885855,0.206485,0.0006549038],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9851372,0.00003229295,0.005934283,0.008389404,0.0000327306,0.000299176,0.000001322878,0.00001013018,0.0001634454],"genre_scores_gemma":[0.9891199,0.00005391601,0.01019833,0.0005473027,0.000009039637,0.00001689465,1.409428e-7,0.000001329057,0.00005309701],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6153777,"threshold_uncertainty_score":0.4254354,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4379364963","doi":"10.1002/asi.24758","title":"A comparative study of the coverage of African journals in Web of Science, Scopus, and CrossRef","year":2023,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":63,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Council of Prairie and Pacific University Libraries; Western University","funders":"","keywords":"Scopus; Web of science; Directory; Library science; Citation; Bibliometrics; World Wide Web; Computer science; MEDLINE; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.2853328370554505,"gpt":0.536237433066305,"spread":0.2509045960108545,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.04947075,0.00005328363,0.0003189178,0.03845652,0.0003966199,0.0003736579,0.002044581,0.00005936035,0.00000212855],"category_scores_gemma":[0.0557647,0.00002675581,0.0000516654,0.1881938,0.001440606,0.002244704,0.0006814467,0.0001995124,6.576706e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001699005,"about_ca_system_score_gemma":0.001097817,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000112097,"about_ca_topic_score_gemma":0.00002097445,"domain_scores_codex":[0.9920731,0.00007773766,0.001216745,0.0001059222,0.006293956,0.0002324943],"domain_scores_gemma":[0.9843748,0.001274124,0.003239436,0.0002427871,0.01081532,0.00005349106],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00002997721,0.0001150173,0.965319,0.000007959457,0.00001500614,1.036225e-7,0.005175968,0.0002977748,0.01093299,0.003283388,0.0005685078,0.01425432],"study_design_scores_gemma":[0.00112638,0.0004335156,0.9487514,0.00003057155,0.000007590939,0.000006018097,0.02292543,0.003746074,0.01290201,0.008992556,0.001034372,0.00004403226],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9972012,0.00004301766,0.00002677777,0.0014186,0.0002310908,0.0002956442,0.00001796197,0.000001968682,0.0007637553],"genre_scores_gemma":[0.9998111,0.0000822876,0.00004244457,0.00001519143,0.00000414487,0.000002215195,3.632737e-8,7.970528e-7,0.00004176566],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1497372,"threshold_uncertainty_score":0.9787699,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["bibliometrics","metaresearch"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2583047471","doi":"10.1080/08989621.2017.1287567","title":"Best Practice to Order Authors in Multi/Interdisciplinary Health Sciences Research Publications","year":2017,"lang":"en","type":"article","venue":"Accountability in Research","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":56,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"Social Sciences and Humanities Research Council; Social Sciences and Humanities Research Council of Canada; Université de Montréal","keywords":"Order (exchange); Engineering ethics; Declaration; Best practice; Accountability; Attribution; Confusion; Computer science; Political science; Psychology; Public relations; Management science; Sociology; Business; Social psychology; Law; Engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.9220851009964572,"gpt":0.8041078746149898,"spread":0.1179772263814675,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","sts"],"category_scores_codex":[0.240632,0.000195284,0.00054918,0.002518669,0.003158268,0.0009636302,0.003281767,0.000568661,0.0003889091],"category_scores_gemma":[0.477455,0.0001738806,0.00007686738,0.005028972,0.005907601,0.001270809,0.006727421,0.01214787,0.00101392],"about_ca_system_candidate":true,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002410007,"about_ca_system_score_gemma":0.009573539,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.04582198,"about_ca_topic_score_gemma":0.1308576,"domain_scores_codex":[0.9828082,0.004035822,0.001343624,0.001906417,0.007331734,0.002574213],"domain_scores_gemma":[0.8753092,0.09518246,0.0003301474,0.008828728,0.01818358,0.002165934],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.001652888,0.01030547,0.9273467,0.0009447468,0.00002736658,0.0001012623,0.0122684,0.00004069092,0.0003332892,0.02359899,0.007547678,0.01583254],"study_design_scores_gemma":[0.002052485,0.003248475,0.8936264,0.001830177,0.0000036959,0.0000253766,0.0444391,0.001488087,0.0001041649,0.03983393,0.01305076,0.0002973254],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.4948779,0.0001682878,0.00003242673,0.4637107,0.0001443517,0.002795577,0.00001497862,0.00003121772,0.03822455],"genre_scores_gemma":[0.9812374,0.0005018513,0.01024559,0.0006807427,0.0001600285,0.0008045014,0.000008671418,0.00003658687,0.00632464],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4863594,"threshold_uncertainty_score":0.9997639,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["research_integrity","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2020943846","doi":"10.1007/s11136-011-9848-0","title":"A knowledge translation challenge: clinical use of quality of life data from cancer clinical trials","year":2011,"lang":"en","type":"article","venue":"Quality of Life Research","topic":"Cancer survivorship and care","field":"Medicine","cited_by":55,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Princess Margaret Cancer Centre; University of Toronto; Queen's University","funders":"","keywords":"Generalizability theory; Medicine; Relevance (law); Knowledge translation; Clinical trial; Quality of life (healthcare); Quality (philosophy); MEDLINE; Medical education; Psychology; Knowledge management; Nursing; Pathology; Computer science","retraction":null,"screen_n_in":null,"score":{"opus":0.9745830517939917,"gpt":0.715245748538548,"spread":0.2593373032554437,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.1030333,0.0001896569,0.002879553,0.0001975593,0.00004764119,0.00001048616,0.0007515308,0.0005527536,0.001420936],"category_scores_gemma":[0.06520387,0.0001604945,0.0006823731,0.0004790514,0.001174992,0.0002191206,0.0003417411,0.001062299,0.00002155547],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004178212,"about_ca_system_score_gemma":0.002811247,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.08307464,"about_ca_topic_score_gemma":0.03509185,"domain_scores_codex":[0.9658017,0.02445379,0.006576389,0.0008581595,0.00188088,0.0004291191],"domain_scores_gemma":[0.9474211,0.04403178,0.001782191,0.003405553,0.002442148,0.0009171746],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.03349918,0.006463597,0.7873057,0.003782631,0.001780151,0.000002355736,0.01364548,3.203699e-7,0.0009147633,0.001648197,0.007957117,0.1430006],"study_design_scores_gemma":[0.009271865,0.001390771,0.9718494,0.0007352203,0.0002661941,1.211722e-7,0.004312031,0.0002558684,0.0005908737,0.0008574569,0.01018274,0.000287486],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9814492,0.01035353,0.0004394646,0.002245977,0.0006351945,0.0009792513,0.002436436,0.00002559815,0.001435344],"genre_scores_gemma":[0.990912,0.006973034,0.0005457934,0.0001161756,0.001021481,0.00002743284,0.0002406229,0.00002893019,0.0001345578],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1845437,"threshold_uncertainty_score":0.9994919,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2766184587","doi":"10.1186/s40510-017-0196-7","title":"Effectiveness of en masse versus two-step retraction: a systematic review and meta-analysis","year":2017,"lang":"en","type":"review","venue":"Progress in Orthodontics","topic":"Orthodontics and Dentofacial Orthopedics","field":"Dentistry","cited_by":53,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Medicine; Meta-analysis; Dentistry; Randomized controlled trial; Orthodontics; Surgery; Internal medicine","retraction":null,"screen_n_in":null,"score":{"opus":0.1865449401167475,"gpt":0.4686407095626851,"spread":0.2820957694459376,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","metaepi_broad"],"consensus_categories":[],"category_scores_codex":[0.00898363,0.001013629,0.01703977,0.0007510525,0.0001414306,0.000298963,0.0009248139,0.0008251872,0.00008103392],"category_scores_gemma":[0.002495539,0.0007675448,0.004772052,0.001183769,0.0002733378,0.0002515344,0.0005040009,0.001098623,0.00003192488],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001531028,"about_ca_system_score_gemma":0.0002083872,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004509601,"about_ca_topic_score_gemma":0.00028263,"domain_scores_codex":[0.9899106,0.004641711,0.002831376,0.001059718,0.00103722,0.0005193493],"domain_scores_gemma":[0.9900385,0.002984572,0.003902062,0.00245925,0.0004080636,0.0002075382],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"systematic_review","study_design_gemma":"meta_analysis","study_design_scores_codex":[0.00002789033,0.0001173109,0.004886171,0.8754627,0.1115774,0.0003554928,0.000009958277,4.472391e-7,3.030312e-9,0.0005300533,0.000008788596,0.007023798],"study_design_scores_gemma":[0.0004719409,0.00006864161,0.0002436886,0.07067263,0.8995192,0.00008805483,0.000004795082,0.00001431119,1.764444e-7,0.000003485295,0.02825827,0.0006548142],"study_design_candidate":"meta_analysis","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.000008786069,0.9933023,0.0004452432,0.00001042968,0.0007690649,0.004751027,0.0003124353,0.00005313996,0.000347554],"genre_scores_gemma":[0.0001547628,0.995366,0.002562627,0.000007044553,0.00006480063,0.001022137,0.0001545738,0.0001199755,0.0005480474],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.80479,"threshold_uncertainty_score":0.9994776,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","metaepi_broad"],"domain":"evaluation","study_design":"meta_analysis","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","research_integrity","scholarly_communication"],"domain":"evaluation","study_design":"meta_analysis","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2043763793","doi":"10.1371/journal.pone.0097886","title":"Sharing Individual Participant Data from Clinical Trials: An Opinion Survey Regarding the Establishment of a Central Repository","year":2014,"lang":"en","type":"article","venue":"PLoS ONE","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":52,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Medical Research Council; University of Birmingham; Queen's University; Cancer Research UK","keywords":"Safeguarding; Clinical trial; Data sharing; Workload; Medicine; Information repository; Systematic review; Resource (disambiguation); MEDLINE; Alternative medicine; Computer science; Nursing; Pathology; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.9754389651248165,"gpt":0.6107774621438413,"spread":0.3646615029809752,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.6082692,0.0001918343,0.004514113,0.00009412284,0.0001526647,0.001138763,0.005037323,0.00009396602,0.0008420367],"category_scores_gemma":[0.3519663,0.00008102434,0.0006761889,0.000448914,0.0000864652,0.0004646254,0.0009303371,0.0002311298,0.0001310869],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000009410888,"about_ca_system_score_gemma":0.00006804799,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003187879,"about_ca_topic_score_gemma":0.0002415642,"domain_scores_codex":[0.7854382,0.1716076,0.02818607,0.00266045,0.0114942,0.00061346],"domain_scores_gemma":[0.8058426,0.136282,0.02410384,0.03109752,0.001860719,0.0008133398],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00006788341,0.001235264,0.9691786,0.00004135166,0.003236508,0.000001193313,0.001006524,0.00003224213,0.0003571811,0.0002442759,0.01725192,0.007347029],"study_design_scores_gemma":[0.000706616,0.0001613638,0.8253631,0.0003635198,0.004140388,9.706517e-7,0.0007806252,0.1622934,0.0003983456,0.00131045,0.004227715,0.0002535824],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.993377,0.0008387049,0.003399059,0.0002726731,0.0005726212,0.0007668837,0.000264242,0.000007552153,0.0005013076],"genre_scores_gemma":[0.9958612,0.00003652031,0.002897696,0.00008767275,0.0006299022,0.00001762677,0.0002141498,0.000009988876,0.0002452528],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.256303,"threshold_uncertainty_score":0.9998981,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1941622179","doi":"10.1111/j.1365-2648.2010.05429.x","title":"Research citation analysis of nursing academics in Canada: identifying success indicators","year":2010,"lang":"en","type":"article","venue":"Journal of Advanced Nursing","topic":"Academic Writing and Publishing","field":"Arts and Humanities","cited_by":50,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"University of Manitoba; CancerCare Manitoba","funders":"","keywords":"Scopus; Citation; Citation analysis; Nursing research; Bibliometrics; Summative assessment; Medicine; Subject (documents); Library science; Medical education; Nursing; MEDLINE; Psychology; Political science; Computer science; Pedagogy","retraction":null,"screen_n_in":null,"score":{"opus":0.08563898176354096,"gpt":0.384438687922677,"spread":0.298799706159136,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002337351,0.00009579013,0.0003653661,0.002590352,0.0002400499,0.0001272779,0.0003601243,0.00009443347,0.00008501637],"category_scores_gemma":[0.0006717645,0.00009221588,0.00009689774,0.001154934,0.0002450394,0.001231457,0.00001221409,0.002230408,2.432279e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007168347,"about_ca_system_score_gemma":0.0006902823,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.03331433,"about_ca_topic_score_gemma":0.1948358,"domain_scores_codex":[0.9977134,0.0001368331,0.0008196898,0.0001339273,0.0008606412,0.0003354683],"domain_scores_gemma":[0.9979902,0.0005455855,0.0007626659,0.000105533,0.0004843946,0.0001116509],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"qualitative","study_design_scores_codex":[0.0002807268,0.0002552246,0.06440661,0.0001179989,0.0004158328,0.00004570369,0.224884,0.01402081,0.01509439,0.06101409,0.004216431,0.6152482],"study_design_scores_gemma":[0.001816113,0.000166767,0.2109725,0.007039054,0.00106018,0.00006114073,0.7315609,0.005243616,0.007818104,0.02260489,0.01091542,0.0007412795],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9916937,0.0004240658,0.0001579309,0.001141732,0.001271362,0.00004967074,0.000007534712,0.000005066141,0.005248882],"genre_scores_gemma":[0.9985214,0.00004386604,0.000649431,0.00003197661,0.0006831235,6.411322e-7,0.000004363534,0.00001391121,0.00005125847],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6145069,"threshold_uncertainty_score":0.9731229,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","scholarly_communication"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W4406842793","doi":"10.1038/d41586-025-00212-1","title":"‘Stamp out paper mills’ — science sleuths on how to fight fake research","year":2025,"lang":"en","type":"article","venue":"Nature","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":50,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Bishop's University","funders":"","keywords":"Scientific misconduct; Spotting; Misconduct; Engineering ethics; Political science; Engineering; Computer science; Law; Medicine; Alternative medicine","retraction":null,"screen_n_in":null,"score":{"opus":0.04031818940135897,"gpt":0.4228028907455642,"spread":0.3824847013442052,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts","research_integrity"],"consensus_categories":["research_integrity"],"category_scores_codex":[0.005278902,0.00009814664,0.0001198654,0.0003513665,0.001624132,0.0002451956,0.001158502,0.002807168,0.0003059565],"category_scores_gemma":[0.007637953,0.00007976296,0.00004833618,0.001857587,0.0006374621,0.0002642359,0.0001515074,0.01040548,0.0003282108],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000307192,"about_ca_system_score_gemma":0.0008605358,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003355954,"about_ca_topic_score_gemma":0.001484175,"domain_scores_codex":[0.9969459,0.0002936342,0.00009371812,0.0004287795,0.001572181,0.0006658482],"domain_scores_gemma":[0.9983069,0.0006168683,0.00002123268,0.0002718352,0.0005919809,0.0001911358],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00002199074,0.00001997829,0.0002167521,0.000006202847,0.000004762182,0.000005265267,0.01123444,0.000001213165,0.001682321,0.303945,0.6729299,0.009932146],"study_design_scores_gemma":[0.00007106907,0.00002511527,0.0009560608,0.00008079258,0.000002392585,1.396561e-7,0.003832523,0.000004827958,0.002924868,0.00918745,0.982818,0.00009675042],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.02413682,0.0007610692,0.00003757921,0.3052607,0.002215668,0.000394245,0.00002704345,0.00007597485,0.6670908],"genre_scores_gemma":[0.8524667,0.000138528,0.0002659316,0.02258747,0.0005510114,0.00001328036,0.000002770431,0.000005334081,0.123969],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8283299,"threshold_uncertainty_score":0.9996756,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"evaluation","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["research_integrity","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W3172729171","doi":"10.1186/s12961-021-00726-w","title":"Where is critical analysis of power and positionality in knowledge translation?","year":2021,"lang":"en","type":"letter","venue":"Health Research Policy and Systems","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":50,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"University of Northern British Columbia; University of British Columbia; University of British Columbia Hospital","funders":"","keywords":"Reflexivity; Knowledge translation; Health services research; Sociology; Context (archaeology); Power (physics); Epistemology; Critical theory; Knowledge production; Engineering ethics; Social science; Political science; Knowledge management; Health care; Computer science; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.8942625729423347,"gpt":0.7697535168293935,"spread":0.1245090561129412,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["research_integrity"],"consensus_categories":[],"category_scores_codex":[0.01963942,0.000251916,0.001349744,0.003191952,0.001143178,0.00007047767,0.0002506633,0.0008576041,0.0004550705],"category_scores_gemma":[0.004563105,0.0002399922,0.0001022472,0.004457267,0.0005967239,0.0001707173,0.0001683852,0.003672042,0.00002936266],"about_ca_system_candidate":true,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009345475,"about_ca_system_score_gemma":0.01030792,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03408691,"about_ca_topic_score_gemma":0.00692544,"domain_scores_codex":[0.9767867,0.01640882,0.002405846,0.0008018083,0.001608346,0.001988449],"domain_scores_gemma":[0.9720984,0.0245929,0.0003980048,0.0005958476,0.001612673,0.0007022496],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00001898486,0.00004236725,0.01458165,0.02238607,0.0001306075,0.00002165231,0.04403347,7.220024e-7,0.000005668202,0.01211214,0.9062789,0.0003877317],"study_design_scores_gemma":[0.0006332442,0.0001465807,0.04376788,0.002210132,0.00003552651,0.00000695747,0.006896774,0.0005810474,3.690894e-7,0.0004123555,0.9450992,0.000209872],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"commentary","genre_gemma":"commentary","genre_scores_codex":[0.004495759,0.007970277,0.000064163,0.977828,0.0001976281,0.00190781,0.002340374,0.00001855658,0.005177456],"genre_scores_gemma":[0.1554835,0.004274681,0.000151813,0.8324811,0.002542152,0.000930739,0.0002955208,0.0000605449,0.003779854],"genre_candidate":"commentary","genre_consensus":"commentary","teacher_disagreement_score":0.1509878,"threshold_uncertainty_score":0.9986265,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["sts"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["sts","scholarly_communication"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W1997466337","doi":"10.1177/0963662509335410","title":"In the public interest: assessing expert and stakeholder influence in public deliberation about biobanks","year":2009,"lang":"en","type":"article","venue":"Public Understanding of Science","topic":"Risk Perception and Management","field":"Social Sciences","cited_by":49,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia; McMaster University","funders":"","keywords":"Deliberation; Stakeholder; Prima facie; Experiential knowledge; Public relations; Public participation; Biobank; Political science; Engineering ethics; Knowledge management; Epistemology; Computer science; Engineering; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.2983881626394008,"gpt":0.3883963051243476,"spread":0.09000814248494687,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.005597124,0.0001111099,0.0001356144,0.0009667375,0.0009157691,0.002613661,0.0007709435,0.00006361236,0.00005862466],"category_scores_gemma":[0.0009150154,0.00008974163,0.0000229647,0.003019589,0.00235226,0.00692013,0.00008095361,0.0001827687,0.000001978314],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007380907,"about_ca_system_score_gemma":0.0004585212,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003017966,"about_ca_topic_score_gemma":0.003581673,"domain_scores_codex":[0.9976378,0.0002332496,0.0003434357,0.0003726027,0.0008255221,0.0005874002],"domain_scores_gemma":[0.9992458,0.0001082132,0.0001392834,0.0002208103,0.0001435124,0.000142424],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","study_design_scores_codex":[0.000001943379,0.0001195076,0.0055061,0.000004956937,0.00000147396,0.00000332061,0.03548214,0.000002837297,0.001215585,0.9383353,0.0004380388,0.01888881],"study_design_scores_gemma":[0.001359728,0.0001761616,0.3958433,0.0002112779,0.000006016657,0.000009916227,0.445362,0.001896072,0.00007959508,0.1017364,0.05260095,0.0007185553],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.865985,0.0001237177,0.005172807,0.1055614,0.0001865947,0.0003418254,5.870688e-7,0.00004108977,0.02258696],"genre_scores_gemma":[0.997035,0.0003055568,0.0003501416,0.002213542,0.00003095392,0.000009727385,7.84488e-7,0.000003468769,0.00005086507],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8365989,"threshold_uncertainty_score":0.9984217,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["sts"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["sts","scholarly_communication"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2015488150","doi":"10.1186/1472-6939-9-4","title":"When research seems like clinical care: a qualitative study of the communication of individual cancer genetic research results","year":2008,"lang":"en","type":"article","venue":"BMC Medical Ethics","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":49,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":true},"ca_institutions":"McMaster University; University of Toronto","funders":"Institute of Health Services and Policy Research; Canadian Institutes of Health Research; Ontario Genomics; Ontario Genomics Institute; Genome Canada","keywords":"Obligation; Harm; Philosophy of medicine; Relevance (law); Qualitative research; Moral obligation; Empirical research; Set (abstract data type); Psychology; Clinical research; Research ethics; Public relations; Social psychology; Medicine; Alternative medicine; Political science; Sociology; Social science; Psychiatry","retraction":null,"screen_n_in":null,"score":{"opus":0.9105878742297558,"gpt":0.7520872850934804,"spread":0.1585005891362754,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts","research_integrity"],"consensus_categories":["metaresearch","research_integrity"],"category_scores_codex":[0.2353013,0.000201003,0.0009016562,0.0003851695,0.0008313052,0.00002085918,0.003271568,0.00281965,0.0001986375],"category_scores_gemma":[0.720934,0.0001359665,0.0002676245,0.001913246,0.01593659,0.00005166143,0.003394833,0.03484622,0.0000267905],"about_ca_system_candidate":true,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002349967,"about_ca_system_score_gemma":0.05458442,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.01284797,"about_ca_topic_score_gemma":0.1512597,"domain_scores_codex":[0.8820168,0.07225841,0.004169249,0.00119486,0.03921211,0.001148582],"domain_scores_gemma":[0.2729786,0.6998979,0.0006645474,0.004701107,0.02065884,0.001098991],"domain_codex":null,"domain_gemma":"methods","domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.002923922,0.004665036,0.0596434,0.003255943,0.0003516681,0.00005374713,0.9102216,0.00001110398,0.0000232547,0.00316577,0.0135014,0.002183117],"study_design_scores_gemma":[0.01724543,0.01046401,0.2465917,0.01039401,0.0002035616,0.00003256371,0.6786724,0.0001883739,0.0003270263,0.03339746,0.002115486,0.0003679888],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.967136,0.002932635,0.00004000271,0.02529691,0.0003170837,0.00221274,0.00007274284,0.00002605361,0.001965801],"genre_scores_gemma":[0.9740271,0.02065501,0.002484102,0.000618912,0.0003307197,0.0001969822,0.00002023052,0.0000470634,0.001619886],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6276395,"threshold_uncertainty_score":0.9984749,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2127448545","doi":"10.1177/0963662509102690","title":"The gap between scientists and journalists: the case of mercury science in Québec’s press","year":2009,"lang":"en","type":"article","venue":"Public Understanding of Science","topic":"Climate Change Communication and Perception","field":"Social Sciences","cited_by":46,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":true},"ca_institutions":"Université du Québec à Montréal","funders":"Université du Québec à Montréal","keywords":"Bridging (networking); Media studies; Journalism; Science communication; Sociology; Political science; Public relations; Computer science; Law; Science education","retraction":null,"screen_n_in":null,"score":{"opus":0.6052165534473128,"gpt":0.4691542361943893,"spread":0.1360623172529236,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts"],"consensus_categories":["sts"],"category_scores_codex":[0.01152,0.00005913653,0.00009867932,0.0002786806,0.004475825,0.000615802,0.001303677,0.00002501757,0.00001792225],"category_scores_gemma":[0.001402139,0.00003779781,0.00002286017,0.002412485,0.01929446,0.00103462,0.0001667422,0.0001317699,3.303531e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005910507,"about_ca_system_score_gemma":0.0009044728,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.01170338,"about_ca_topic_score_gemma":0.04534435,"domain_scores_codex":[0.9983317,0.0001532502,0.0002640718,0.0001905622,0.0006954125,0.000364976],"domain_scores_gemma":[0.99858,0.0004958221,0.0002177778,0.0003258635,0.0002447784,0.0001357541],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","study_design_scores_codex":[0.000008684699,0.00006045755,0.01358722,0.0000111461,0.000002646328,0.00000369661,0.1147308,0.000003970139,0.002601465,0.8336586,0.0001960195,0.03513526],"study_design_scores_gemma":[0.0009100214,0.0002215809,0.1406157,0.0002901152,0.00003301907,0.0001416993,0.7288572,0.002333686,0.0007293672,0.1032897,0.02204408,0.000533888],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9153531,0.0006343969,0.0003460962,0.04974548,0.0002176068,0.0002586384,0.000006973336,0.00001749054,0.03342022],"genre_scores_gemma":[0.9984014,0.001419398,0.00005379531,0.00005097573,0.00002229693,0.000001535796,2.065925e-7,0.000001609864,0.00004875291],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7303689,"threshold_uncertainty_score":0.9968202,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["sts"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["sts","scholarly_communication"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"}]}