{"meta":{"query_hash":"283450e612e9","filters":{"category":"open_science"},"cohort_total":262,"direct_labels_cover":262,"predictions_cover":262,"exported":262,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/283450e612e9","api":"https://metacan.xera.ac/api/v1/cohort?category=open_science"},"results":[{"id":"W100260889","doi":"10.5683/sp2/34sn5w","title":"Dramatic Growth of Open Access March 31, 2019","year":2019,"lang":"en","type":"dataset","venue":"Summit (Simon Fraser University)","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Computer science","score_opus":0.08996723890744256,"score_gpt":0.376945656635178,"score_spread":0.2869784177277354,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W100260889","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00012030484,0.00013664684,0.00003460896,0.00040876033,0.00012997973,0.0000054993984,0.99724823,0.00012215498,0.0017937912],"genre_scores_gemma":[0.0009377411,0.00017693706,0.00013391009,0.00028530863,0.00006160718,0.000045645025,0.9945899,0.00009355146,0.0036754333],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9965571,0.00040246246,0.00060431584,0.00062675955,0.0012581984,0.00055121694],"domain_scores_gemma":[0.9922546,0.0017861509,0.0013081256,0.0009959827,0.0026446055,0.0010104319],"candidate_categories":["bibliometrics","open_science"],"consensus_categories":[],"category_scores_codex":[0.0021009464,0.0014027375,0.0013345636,0.0055294693,0.0010569808,0.005530956,0.0019267091,0.0022671882,0.12249674],"category_scores_gemma":[0.0211523,0.00057448796,0.0014891762,0.0110019175,0.00053847756,0.0027301982,0.0028188666,0.0024971222,0.11642772],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000060023023,0.000005296668,0.00073938095,0.0002750775,0.000015740497,0.000010958583,0.000007248944,0.00007346745,0.000025555124,0.0005298037,0.9965395,0.001718015],"study_design_scores_gemma":[0.00018390012,0.000013854907,0.008951908,0.00057863985,0.00003982615,0.00005323395,0.00008981092,0.00022189932,0.00017197983,0.0014323796,0.9882257,0.000036905494],"about_ca_topic_score_codex":0.054102883,"about_ca_topic_score_gemma":0.07807013,"teacher_disagreement_score":0.9980733,"about_ca_system_score_codex":0.0025233675,"about_ca_system_score_gemma":0.0053016436,"threshold_uncertainty_score":0.40979242},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1495039464","doi":"10.7910/dvn/sikkdv","title":"Dramatic Growth of Open Access","year":2010,"lang":"en","type":"dataset","venue":"uO Research (University of Ottawa)","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science","score_opus":0.3336734854130005,"score_gpt":0.5300456706965908,"score_spread":0.1963721852835903,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1495039464","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0025829074,0.00027967946,0.00037633535,0.00096436567,0.00021586783,0.0000213295,0.98941666,0.00062411925,0.0055186115],"genre_scores_gemma":[0.0042183916,0.00019222836,0.0007538018,0.00029249556,0.000049274295,0.00005305209,0.9905956,0.00016468762,0.0036804995],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9921286,0.0009418781,0.0010141504,0.0016005492,0.003536767,0.00077800866],"domain_scores_gemma":[0.9804716,0.005044041,0.0027994341,0.0043335664,0.0056248032,0.0017266339],"candidate_categories":["bibliometrics","open_science"],"consensus_categories":[],"category_scores_codex":[0.004754566,0.0010471746,0.00082066376,0.0066212397,0.0007132218,0.0038916771,0.0019441805,0.0012716674,0.020089805],"category_scores_gemma":[0.022841308,0.0005759323,0.0008616569,0.015532903,0.00069501746,0.0034513646,0.002467482,0.0030180183,0.022551112],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011898031,0.000023878127,0.0038326252,0.00022380259,0.000022291933,0.00003472723,0.00003302641,0.0003184072,0.0001559149,0.0028199095,0.9854113,0.007005193],"study_design_scores_gemma":[0.000091741014,0.000019776742,0.022666939,0.00017759459,0.000023281817,0.00015771882,0.00014463601,0.0006700829,0.0008410295,0.0016450196,0.97351426,0.000048004968],"about_ca_topic_score_codex":0.048657157,"about_ca_topic_score_gemma":0.09260574,"teacher_disagreement_score":0.9980558,"about_ca_system_score_codex":0.003926145,"about_ca_system_score_gemma":0.0042667757,"threshold_uncertainty_score":0.096747875},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W1778783756","doi":"","title":"Research guides: Open Access & Education Research: Home","year":2011,"lang":"en","type":"libguides","venue":"","topic":"Open Education and E-Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Focus (optics); Distance education; Open education; Sociology; Pedagogy","score_opus":0.4832899441155772,"score_gpt":0.5530074639212266,"score_spread":0.06971751980564939,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1778783756","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00085813465,0.0067094234,0.04107787,0.019945523,0.0031310215,0.006962714,0.079894245,0.029930186,0.81149083],"genre_scores_gemma":[0.0031700206,0.011530383,0.059647314,0.007814785,0.0009914467,0.010246707,0.027173985,0.012717014,0.86670834],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9927938,0.002967038,0.0007471952,0.00042836153,0.0026997677,0.00036393147],"domain_scores_gemma":[0.9093863,0.053727265,0.0028561517,0.007446629,0.02026737,0.006316314],"candidate_categories":["scholarly_communication","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0144254165,0.0009732419,0.0017952509,0.0043460634,0.0025606053,0.0060596983,0.0025592453,0.0039894637,0.38802224],"category_scores_gemma":[0.04580208,0.001898665,0.0005620077,0.011177699,0.0021522308,0.00733725,0.0033897916,0.004979441,0.3412382],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002709718,0.000038110007,0.000046660592,0.00057449884,0.0000013297563,0.000025197609,0.00042560403,0.000027833788,0.00016011605,0.0033382748,0.93885326,0.056481957],"study_design_scores_gemma":[0.000023049524,0.00001780083,0.00043159828,0.00035546024,0.0000020366124,0.00004171875,0.00032778745,0.00003148391,0.00011534164,0.0016556248,0.99698395,0.000014181263],"about_ca_topic_score_codex":0.018495662,"about_ca_topic_score_gemma":0.0723986,"teacher_disagreement_score":0.99744076,"about_ca_system_score_codex":0.003045099,"about_ca_system_score_gemma":0.018282438,"threshold_uncertainty_score":0.872912},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W1932056044","doi":"10.3917/i2d.151.0063","title":"Sur la voie de l’open access et de l’open data","year":2015,"lang":"fr","type":"article","venue":"I2D - Information données & documents","topic":"Cultural Insights and Digital Impacts","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bibliothèque et Archives nationales du Québec","funders":"","keywords":"Political science; Computer science; Humanities; Philosophy","score_opus":0.5558674876823668,"score_gpt":0.467568127053682,"score_spread":0.08829936062868482,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1932056044","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006012409,0.056940984,0.117675446,0.6878296,0.020375488,0.00015521009,0.0007181144,0.0007522911,0.10954031],"genre_scores_gemma":[0.2913474,0.13522024,0.20420443,0.11822302,0.03449054,0.0014281046,0.0031565016,0.0041322927,0.20779751],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9397381,0.025987115,0.0036399513,0.004328902,0.023068028,0.0032378535],"domain_scores_gemma":[0.81722045,0.11515099,0.0054447413,0.02861398,0.023527052,0.010042749],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.06998298,0.0007391798,0.0012780038,0.00604694,0.008675363,0.035703793,0.0037062562,0.01600784,0.01642565],"category_scores_gemma":[0.13724166,0.0010977486,0.0016882624,0.009176102,0.039820723,0.07740364,0.022362055,0.023274086,0.003966194],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000030010446,0.000015590387,0.00020059515,0.00019211002,0.000010817112,0.00008295251,0.002332302,0.00016798389,0.00023289713,0.92540056,0.04143929,0.029894942],"study_design_scores_gemma":[0.000015073423,0.0000146363145,0.00034719516,0.00057129643,0.00000668647,0.00020649302,0.0019127448,0.00046912872,0.0004205313,0.26431233,0.7316742,0.000049541646],"about_ca_topic_score_codex":0.018993098,"about_ca_topic_score_gemma":0.009845676,"teacher_disagreement_score":0.9962937,"about_ca_system_score_codex":0.009651306,"about_ca_system_score_gemma":0.02227255,"threshold_uncertainty_score":0.3701098},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W1953894052","doi":"10.6017/ital.v34i3.5900","title":"Self-Archiving with Ease in an Institutional Repository: Microinteractions and the User Experience","year":2015,"lang":"en","type":"article","venue":"Information Technology and Libraries","topic":"Digital and Traditional Archives Management","field":"Arts and Humanities","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"MacEwan University; University of Alberta","funders":"","keywords":"Computer science; Usability; Process (computing); World Wide Web; Institutional repository; Human–computer interaction","score_opus":0.011984026821439769,"score_gpt":0.1807309643641611,"score_spread":0.16874693754272135,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1953894052","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9450489,0.0006076019,0.017361838,0.0017067973,0.000050639745,0.00005195585,0.000031473952,0.00056568976,0.034575034],"genre_scores_gemma":[0.9921673,0.00011300289,0.0035273503,0.00021525237,0.000026087535,0.000025510477,0.000013978374,0.000122047946,0.0037894787],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.9873552,0.009661216,0.0003758298,0.00041166204,0.0017131759,0.00048302024],"domain_scores_gemma":[0.97147155,0.021170925,0.0013799849,0.003369427,0.0012713901,0.0013368151],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.00990581,0.00040026673,0.00041330882,0.0018131662,0.004679725,0.011452131,0.0012941005,0.0018895234,0.0060097254],"category_scores_gemma":[0.025565486,0.00046099463,0.0005302691,0.0013083036,0.0064750686,0.012062925,0.008978812,0.0018014677,0.0009207539],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026340823,0.00030264657,0.026368767,0.00016346376,0.00004603888,0.0015292545,0.87107885,0.0003486299,0.0058112773,0.01918186,0.0027245155,0.07218133],"study_design_scores_gemma":[0.00008046702,0.001223682,0.045436192,0.00030020523,0.00013152433,0.0072952374,0.7701936,0.00449855,0.010218149,0.021979148,0.13824569,0.00039756965],"about_ca_topic_score_codex":0.00095165166,"about_ca_topic_score_gemma":0.0014034617,"teacher_disagreement_score":0.98854786,"about_ca_system_score_codex":0.0010867531,"about_ca_system_score_gemma":0.0008192971,"threshold_uncertainty_score":0.052387595},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"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,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_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","score_opus":0.134351923613858,"score_gpt":0.5619974727123195,"score_spread":0.4276455490984615,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2001314811","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06539757,0.11720635,0.17473441,0.107407086,0.01776757,0.000954799,0.006098953,0.003537249,0.50689614],"genre_scores_gemma":[0.62662333,0.064506814,0.08209369,0.031222992,0.011772157,0.00090508885,0.0031865966,0.00048797348,0.1792013],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990113,0.00034181934,0.00004015889,0.000114592905,0.00039166765,0.00010049604],"domain_scores_gemma":[0.9979315,0.00076002604,0.0002714889,0.00029885606,0.00033297352,0.000405142],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0019252653,0.00044863363,0.0004261963,0.0008980108,0.00034340736,0.0016435863,0.00070094224,0.0018585915,0.033190135],"category_scores_gemma":[0.0035974158,0.00024722275,0.0004069316,0.00045633334,0.0013941169,0.0017343675,0.001756539,0.0023997256,0.006804702],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011062943,0.0004890605,0.0017229316,0.00092257065,0.000089805115,0.0011516538,0.00015282731,0.00075249426,0.1789853,0.21878877,0.08551397,0.5103243],"study_design_scores_gemma":[0.0003668565,0.0013818281,0.0018919043,0.00015542305,0.000077063574,0.0025525584,0.000100568046,0.0013430458,0.093688145,0.086408235,0.8119775,0.000056934485],"about_ca_topic_score_codex":0.00020408402,"about_ca_topic_score_gemma":0.0002544231,"teacher_disagreement_score":0.99929905,"about_ca_system_score_codex":0.00056485884,"about_ca_system_score_gemma":0.0010955031,"threshold_uncertainty_score":0.11103207},"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":"W2020637674","doi":"10.1016/j.acalib.2014.07.011","title":"International Open Access Week at Small to Medium U.S. Academic Libraries: The First Five Years","year":2014,"lang":"en","type":"article","venue":"The Journal of Academic Librarianship","topic":"Web and Library Services","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Outreach; Quarter (Canadian coin); Medical education; Plan (archaeology); Library science; Psychology; Public relations; Sociology; Political science; Medicine; Computer science; History","score_opus":0.06713800596436659,"score_gpt":0.29826794654347905,"score_spread":0.23112994057911246,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2020637674","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24650776,0.03016259,0.0020902143,0.31155747,0.06681835,0.0005333623,0.013768677,0.0030060569,0.32555553],"genre_scores_gemma":[0.33346364,0.016715351,0.0013676966,0.03901696,0.016867947,0.00047586436,0.0063689263,0.0015023422,0.5842213],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9961235,0.00045467683,0.00021460583,0.00025315752,0.0018399509,0.0011140525],"domain_scores_gemma":[0.9581527,0.0032811607,0.0034673966,0.0017773941,0.0059822635,0.027339038],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.0073088156,0.0008631796,0.0007619809,0.0032941864,0.0107083,0.021130942,0.0019730455,0.0054504275,0.10083906],"category_scores_gemma":[0.019633975,0.00060818606,0.00074340706,0.003239923,0.0026482793,0.010708095,0.010197929,0.0060874936,0.040612374],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002674117,0.00073773006,0.022854134,0.00039065597,0.000025626723,0.0005237103,0.0062773293,0.00005206943,0.0005901042,0.0038154114,0.87146366,0.09300224],"study_design_scores_gemma":[0.000015126916,0.00014115786,0.06500696,0.00044423036,0.000012438269,0.00024017443,0.012655404,0.00011313293,0.00030983955,0.00080664275,0.92019874,0.00005603986],"about_ca_topic_score_codex":0.023352036,"about_ca_topic_score_gemma":0.06890227,"teacher_disagreement_score":0.99802697,"about_ca_system_score_codex":0.0045644497,"about_ca_system_score_gemma":0.0100653125,"threshold_uncertainty_score":0.3373403},"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":["scholarly_communication","open_science"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital","funders":"National Institute for Health and Care Research","keywords":"Medicine; Computer science","score_opus":0.7505475693393026,"score_gpt":0.5892674142758912,"score_spread":0.1612801550634113,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2028758832","genre_codex":"protocol","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003976455,0.017569123,0.127796,0.015817776,0.0055038966,0.50879556,0.27177852,0.0075548026,0.041207895],"genre_scores_gemma":[0.0066929758,0.008592377,0.21090186,0.0042343913,0.0010072276,0.66144776,0.09446155,0.0016502414,0.011011693],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.52650464,0.18073148,0.22907807,0.016906157,0.04158859,0.005191037],"domain_scores_gemma":[0.23026547,0.2588508,0.15424477,0.18583427,0.1553796,0.015425049],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.3913823,0.00497817,0.0148858335,0.05352815,0.006729125,0.02392831,0.012814782,0.014848854,0.09658779],"category_scores_gemma":[0.5883435,0.010959179,0.009414386,0.06585453,0.004475381,0.015590517,0.013556505,0.017724084,0.08552558],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.015044462,0.0019896457,0.0050258576,0.079762556,0.0037030121,0.0004094327,0.0015085634,0.0020657615,0.0022089165,0.03149478,0.6035673,0.25321972],"study_design_scores_gemma":[0.030600479,0.0013835783,0.00967158,0.08036275,0.0064753294,0.0006209117,0.000660982,0.0035802186,0.004047602,0.026438309,0.83532023,0.0008380534],"about_ca_topic_score_codex":0.004327005,"about_ca_topic_score_gemma":0.0059209596,"teacher_disagreement_score":0.60861766,"about_ca_system_score_codex":0.010177379,"about_ca_system_score_gemma":0.14235274,"threshold_uncertainty_score":0.750534},"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":"W2037620319","doi":"10.1525/jer.2008.3.4.19","title":"Sharing Data and Results with Study Participants: Report on a Survey of Cultural Anthropologists","year":2008,"lang":"en","type":"article","venue":"Journal of Empirical Research on Human Research Ethics","topic":"Anthropology: Ethics, History, Culture","field":"Social Sciences","cited_by":12,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Harm; Confidentiality; Anonymity; Data sharing; Public relations; Social psychology; Psychology; Survey data collection; Internet privacy; Sociology; Political science; Medicine; Law; Computer science","score_opus":0.9303605501824993,"score_gpt":0.7248475637677881,"score_spread":0.20551298641471116,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2037620319","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9880646,0.0015903101,0.0018400856,0.0031696784,0.00008585222,0.00036699738,0.00017341964,0.00003345664,0.0046755676],"genre_scores_gemma":[0.9910573,0.0019944848,0.002144808,0.0021099988,0.00006656244,0.00050098065,0.00018607361,0.00003535327,0.0019046016],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.9477129,0.039233923,0.0033861613,0.0016423482,0.0051379437,0.0028867766],"domain_scores_gemma":[0.8590726,0.08290756,0.020838534,0.00877456,0.020703923,0.007702784],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.04697492,0.00068040256,0.0007429463,0.0022261865,0.0063275876,0.005322106,0.0011895814,0.0027689636,0.002381391],"category_scores_gemma":[0.12196993,0.0010696739,0.000660795,0.001705423,0.005022198,0.00439663,0.00597687,0.0031079047,0.0007849344],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010658811,0.00016123221,0.1438807,0.0006293778,0.00004969502,0.0016094255,0.8263353,0.000041763957,0.0024001729,0.00034613337,0.0019846046,0.02245504],"study_design_scores_gemma":[0.000007301435,0.00030259957,0.036571063,0.00041547374,0.000028168892,0.0017341009,0.94542825,0.00007570966,0.0007114303,0.00023823361,0.014421901,0.00006583442],"about_ca_topic_score_codex":0.00363922,"about_ca_topic_score_gemma":0.003703968,"teacher_disagreement_score":0.9530251,"about_ca_system_score_codex":0.0012491703,"about_ca_system_score_gemma":0.0035755138,"threshold_uncertainty_score":0.24843007},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2037890806","doi":"10.1087/0953151053585028","title":"In a paperless world a new role for academic libraries: providing open access","year":2005,"lang":"en","type":"article","venue":"Learned Publishing","topic":"Research Data Management Practices","field":"Computer Science","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec","funders":"Euroscience; University of Southampton","keywords":"Mandate; Publishing; Scholarly communication; The Internet; World Wide Web; Open science; Open access publishing; Political science; Library science; Computer science; Electronic publishing; Public relations","score_opus":0.2662665729937543,"score_gpt":0.452188456835734,"score_spread":0.18592188384197972,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2037890806","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.049252596,0.016547736,0.08351867,0.49210334,0.009022537,0.00035373413,0.00032419353,0.0024382565,0.34643894],"genre_scores_gemma":[0.5731019,0.019457709,0.13091688,0.06381163,0.008525965,0.00047713664,0.0003561167,0.0016812133,0.20167156],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.968036,0.01625689,0.0027092942,0.0013632452,0.010185485,0.0014490264],"domain_scores_gemma":[0.85952294,0.052714083,0.01362135,0.034662746,0.02000172,0.019477108],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.032851845,0.00042888068,0.0007840553,0.0043506525,0.010913527,0.058776397,0.0029816325,0.0050060498,0.020628445],"category_scores_gemma":[0.06096677,0.00050436385,0.000734875,0.0074643833,0.019500153,0.04826834,0.018317552,0.005505067,0.008734235],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000110309826,0.00034535542,0.0053173243,0.00085416646,0.000034291614,0.0005970955,0.028354019,0.0002583761,0.0019010461,0.5127488,0.13372616,0.3157531],"study_design_scores_gemma":[0.000031200005,0.00010810342,0.0018381795,0.00088687195,0.000025374617,0.0007242706,0.02158528,0.00032794807,0.0011462937,0.118816584,0.85442674,0.00008316177],"about_ca_topic_score_codex":0.0018566472,"about_ca_topic_score_gemma":0.0026331877,"teacher_disagreement_score":0.9970184,"about_ca_system_score_codex":0.004655755,"about_ca_system_score_gemma":0.014949353,"threshold_uncertainty_score":0.17373925},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"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,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_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","score_opus":0.9754389651248165,"score_gpt":0.6107774621438413,"score_spread":0.3646615029809752,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2043763793","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.28562886,0.06725321,0.041482806,0.5778745,0.0049412753,0.0049107256,0.0007486649,0.00038936944,0.016770601],"genre_scores_gemma":[0.86647886,0.028724764,0.031839512,0.06524875,0.0021312975,0.0034837085,0.00036057964,0.00021420493,0.0015182939],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.5512965,0.3508019,0.049523592,0.008450148,0.03421218,0.00571568],"domain_scores_gemma":[0.15196046,0.6721758,0.061471242,0.015964424,0.07932266,0.019105338],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.44051507,0.0005341068,0.0014522073,0.0027347864,0.0051887943,0.010370989,0.0039312975,0.0088291885,0.0042433618],"category_scores_gemma":[0.675285,0.0014862557,0.0018771618,0.00285272,0.0069106715,0.018223125,0.011330178,0.0075651165,0.00087467773],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027783802,0.00035508242,0.11318387,0.029839974,0.00090087444,0.0024773732,0.34448415,0.0007269571,0.0047170836,0.01124061,0.06828233,0.42101333],"study_design_scores_gemma":[0.0008088469,0.0028513186,0.06452917,0.046276588,0.00081457716,0.0044690073,0.5058412,0.003694818,0.002206071,0.013046379,0.3546216,0.0008405156],"about_ca_topic_score_codex":0.0053286985,"about_ca_topic_score_gemma":0.0037408294,"teacher_disagreement_score":0.9960687,"about_ca_system_score_codex":0.009221412,"about_ca_system_score_gemma":0.023832588,"threshold_uncertainty_score":0.6899445},"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":"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.04629024388331381,"score_gpt":0.34519747332861633,"score_spread":0.2989072294453025,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2048204427","genre_codex":"other","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02487823,0.0026301781,0.06026515,0.025844881,0.016513288,0.00066615024,0.27724472,0.19030808,0.4016493],"genre_scores_gemma":[0.24132352,0.0037832141,0.087965816,0.009764051,0.010004425,0.0011490604,0.3777421,0.06912583,0.19914204],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9986065,0.00020028441,0.000058981466,0.00019876842,0.00066422473,0.00027125186],"domain_scores_gemma":[0.99296325,0.0012760402,0.00055198476,0.002780577,0.0011592435,0.0012690269],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0020313119,0.0010769202,0.0007259117,0.0019557655,0.00089555973,0.00317745,0.0012686013,0.0021246325,0.07315711],"category_scores_gemma":[0.013474937,0.0007412883,0.0008179917,0.001877537,0.0013893745,0.003971721,0.006618229,0.0022949772,0.077335276],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027716695,0.00004566215,0.0031953352,0.00020133004,0.000099367506,0.00013347075,0.00025768758,0.0011695743,0.0038715764,0.010097314,0.91709775,0.063553855],"study_design_scores_gemma":[0.00014534396,0.00003881482,0.0044974545,0.000102831036,0.00004248509,0.00006498994,0.00016579645,0.002610066,0.0025320668,0.018175688,0.9715703,0.000054180087],"about_ca_topic_score_codex":0.0050607114,"about_ca_topic_score_gemma":0.0060290424,"teacher_disagreement_score":0.9987314,"about_ca_system_score_codex":0.0005758785,"about_ca_system_score_gemma":0.002094778,"threshold_uncertainty_score":0.24473488},"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":"W2058753670","doi":"10.1080/03615260801970840","title":"The Public Knowledge Project and the Simon Fraser University Library: A Partnership in Open Source and Open Access","year":2008,"lang":"en","type":"article","venue":"The Serials Librarian","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Suite; General partnership; Publishing; Open source software; Open source; World Wide Web; Software; Computer science; Open science; Library science; Political science","score_opus":0.16831412608541518,"score_gpt":0.28825825780087777,"score_spread":0.11994413171546259,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2058753670","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.030857023,0.01389173,0.03428817,0.20134555,0.0064885817,0.00083685317,0.002436627,0.008447438,0.7014081],"genre_scores_gemma":[0.22556339,0.012835241,0.07775545,0.008705531,0.002303946,0.001323977,0.0058467304,0.004451556,0.6612142],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9724194,0.0090270005,0.0008450793,0.0016230947,0.013312384,0.0027729755],"domain_scores_gemma":[0.9472923,0.014765933,0.002480688,0.006194946,0.0067116404,0.022554517],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.029221952,0.0008491158,0.0008702231,0.0059731035,0.014286203,0.025144173,0.0030621937,0.0062124426,0.040403135],"category_scores_gemma":[0.036338273,0.00072552654,0.0005638013,0.010826953,0.016640162,0.019119376,0.015972763,0.0044749817,0.010032823],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024704213,0.00048200777,0.0016037296,0.00029749837,0.00002790308,0.0006151724,0.0054171635,0.00065920234,0.000591337,0.280818,0.42329952,0.28594142],"study_design_scores_gemma":[0.00007348284,0.00007814841,0.0014175981,0.00018960432,0.0000085665315,0.00019899757,0.0033453463,0.00047806377,0.0011288563,0.020891748,0.9721371,0.000052402207],"about_ca_topic_score_codex":0.038451776,"about_ca_topic_score_gemma":0.033495575,"teacher_disagreement_score":0.9969378,"about_ca_system_score_codex":0.012211393,"about_ca_system_score_gemma":0.06758636,"threshold_uncertainty_score":0.15454233},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2093745477","doi":"10.1038/sdata.2014.49","title":"An open science resource for establishing reliability and reproducibility in functional connectomics","year":2014,"lang":"en","type":"article","venue":"Scientific Data","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":497,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Douglas Mental Health University Institute; Université de Montréal; Institut Universitaire de Gériatrie de Montréal","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; National Institute on Drug Abuse; National Institute of Biomedical Imaging and Bioengineering; National Institute of Neurological Disorders and Stroke; National Institute of Mental Health; National Key Research and Development Program of China","keywords":"Connectomics; Reliability (semiconductor); Reproducibility; Computer science; Neuroimaging; Connectome; Open science; Artificial intelligence; Functional connectivity; Psychology; Neuroscience; Statistics; Mathematics","score_opus":0.14234638286896062,"score_gpt":0.3469814610188182,"score_spread":0.2046350781498576,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2093745477","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0068717157,0.0019194211,0.66883,0.008834189,0.0031275223,0.007889228,0.13096169,0.043989077,0.12757716],"genre_scores_gemma":[0.039016005,0.0022142269,0.7109549,0.002778745,0.0025581853,0.025411323,0.1668875,0.016321393,0.03385768],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9334393,0.020816386,0.011617647,0.005015661,0.027103765,0.002007339],"domain_scores_gemma":[0.43359545,0.17116196,0.02091087,0.23737553,0.12366255,0.013293593],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.10747018,0.0018853596,0.004302735,0.020782102,0.0043686703,0.00779435,0.00918653,0.005400045,0.13306266],"category_scores_gemma":[0.3429892,0.002544304,0.0027049913,0.02326208,0.0032302514,0.007877639,0.0134454295,0.0051832623,0.0850973],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011278404,0.0006635409,0.0037558936,0.0025066908,0.00031212653,0.0005521677,0.00074182166,0.002864054,0.009119995,0.04755579,0.46922266,0.46157733],"study_design_scores_gemma":[0.0008428746,0.00045894418,0.013275452,0.002117909,0.00029993505,0.00084043166,0.00029034476,0.011726297,0.012406238,0.10864294,0.84874785,0.00035069208],"about_ca_topic_score_codex":0.0035047624,"about_ca_topic_score_gemma":0.004478469,"teacher_disagreement_score":0.9908135,"about_ca_system_score_codex":0.0029430406,"about_ca_system_score_gemma":0.022810351,"threshold_uncertainty_score":0.5683634},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2112262712","doi":"","title":"AUPress: A Comparison of an Open Access University Press with Traditional Presses","year":2011,"lang":"en","type":"article","venue":"AUSpace (Athabasca University)","topic":"Open Education and E-Learning","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Amazon rainforest; Ranking (information retrieval); Statistical analysis; Computer science; Library science; World Wide Web; Statistics; Advertising; Mathematics; Information retrieval; Business","score_opus":0.1162601051428903,"score_gpt":0.2957868818465557,"score_spread":0.1795267767036654,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2112262712","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9241243,0.00065122586,0.00014809406,0.00027202474,0.00007041356,0.0001137768,0.0012159685,0.00005343383,0.073350705],"genre_scores_gemma":[0.9762496,0.0005187386,0.00032923982,0.00012365087,0.00005899669,0.00010481081,0.0015388321,0.0000465924,0.021029476],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99522555,0.0007102675,0.00023744607,0.0003694376,0.0029946773,0.0004625587],"domain_scores_gemma":[0.9836715,0.0042173536,0.0020125145,0.0008774355,0.0055671483,0.0036540155],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.0020129278,0.00032285546,0.00035406873,0.0045271316,0.0022069318,0.00555053,0.00074829924,0.00047410958,0.019141179],"category_scores_gemma":[0.01304449,0.0001814648,0.00025445537,0.008444325,0.0017706916,0.0034894163,0.002009989,0.00071009964,0.0017664983],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.007295421,0.0029687267,0.54687834,0.0018480215,0.00050549256,0.0035211095,0.09129576,0.0007400642,0.015223524,0.023920743,0.037740894,0.268062],"study_design_scores_gemma":[0.00007360483,0.00065221434,0.92181623,0.000060058468,0.0000515711,0.00030612253,0.026911026,0.00025163547,0.0011929407,0.00022587586,0.048413467,0.000045231904],"about_ca_topic_score_codex":0.25728387,"about_ca_topic_score_gemma":0.43047184,"teacher_disagreement_score":0.9992517,"about_ca_system_score_codex":0.006952412,"about_ca_system_score_gemma":0.0036713611,"threshold_uncertainty_score":0.5115726},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"qualitative","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":"empirical","about_ca_system":true,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2114332566","doi":"10.1177/1556264614564970","title":"Parent Perspectives on Privacy and Governance for a Pediatric Repository of Non-Biological, Research Data","year":2014,"lang":"en","type":"article","venue":"Journal of Empirical Research on Human Research Ethics","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Alberta Health Services; PolicyWise for Children & Families; University of Calgary","funders":"","keywords":"Focus group; Nonprobability sampling; Thematic analysis; Corporate governance; Qualitative research; Data collection; Qualitative property; Psychology; Focus (optics); Altruism (biology); Internet privacy; Public relations; Business; Sociology; Political science; Computer science; Social psychology; Marketing; Social science","score_opus":0.9457734969064294,"score_gpt":0.7653637445849779,"score_spread":0.1804097523214515,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2114332566","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.91960484,0.0036747423,0.018129257,0.047533777,0.00017645491,0.00010678247,0.00012281073,0.000043433083,0.010607751],"genre_scores_gemma":[0.9916618,0.0014843537,0.0041350676,0.0017167514,0.000036645848,0.00007003075,0.000026475333,0.000012417522,0.00085641164],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9461286,0.047606505,0.0014462242,0.0011526962,0.0024073378,0.0012585617],"domain_scores_gemma":[0.8908992,0.08153653,0.012955281,0.0035232196,0.006190331,0.0048955167],"candidate_categories":["metaresearch","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.05333753,0.00021059954,0.00043026064,0.00070109684,0.0055248304,0.005121571,0.00089563953,0.0018154286,0.001923051],"category_scores_gemma":[0.076444365,0.0005888439,0.0003731103,0.00095082104,0.008389647,0.0057076607,0.0042253295,0.0032247351,0.00020254943],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007112608,0.0000321439,0.019276662,0.00016314899,0.000014922067,0.0028726521,0.9462029,0.00010419114,0.00087736023,0.016931493,0.0017040244,0.011749462],"study_design_scores_gemma":[0.000021551738,0.0003171499,0.011641731,0.00095542736,0.00006202406,0.0076648067,0.89557755,0.00050991634,0.0020710337,0.0077258693,0.07338942,0.00006354678],"about_ca_topic_score_codex":0.0047626323,"about_ca_topic_score_gemma":0.0049251346,"teacher_disagreement_score":0.99818456,"about_ca_system_score_codex":0.004102806,"about_ca_system_score_gemma":0.0077911234,"threshold_uncertainty_score":0.28207916},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"methods","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2124995107","doi":"10.2471/blt.09.064659","title":"Open access: a giant leap towards bridging health inequities","year":2009,"lang":"en","type":"article","venue":"Bulletin of the World Health Organization","topic":"Health Sciences Research and Education","field":"Health Professions","cited_by":40,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"University of Warwick; University of Southampton","keywords":"Bridging (networking); Public health; Health equity; Public relations; Political science; Medicine; Computer science; Computer security; Nursing","score_opus":0.154399671463718,"score_gpt":0.5057955029556225,"score_spread":0.35139583149190445,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2124995107","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014077935,0.01709197,0.032119498,0.7933201,0.0063276347,0.00009013598,0.00033287067,0.0008327293,0.13580722],"genre_scores_gemma":[0.6552067,0.031036416,0.05500699,0.14942826,0.026028214,0.00038693612,0.0008516651,0.0011498394,0.080905],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9787224,0.009083905,0.00096234964,0.0012041032,0.0073561845,0.0026710362],"domain_scores_gemma":[0.87665784,0.06763808,0.005650212,0.018658917,0.011754628,0.019640395],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.03721351,0.0006070518,0.0007979509,0.004525103,0.0056848503,0.017060166,0.0030687312,0.008089226,0.051943254],"category_scores_gemma":[0.06541655,0.0004410101,0.0008448633,0.0032965622,0.018957168,0.037033483,0.028616907,0.007585741,0.0040328437],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007849985,0.00016594044,0.0029636996,0.00076922786,0.000044321805,0.00030498035,0.005600336,0.00026642988,0.00074438535,0.6174425,0.11302169,0.25859815],"study_design_scores_gemma":[0.000024134188,0.000086118496,0.0025904211,0.0011233983,0.000016472266,0.00034845507,0.0058110724,0.00033973792,0.00036648504,0.42795977,0.561284,0.000049907627],"about_ca_topic_score_codex":0.0033173584,"about_ca_topic_score_gemma":0.0034924354,"teacher_disagreement_score":0.99693125,"about_ca_system_score_codex":0.0035741737,"about_ca_system_score_gemma":0.015596324,"threshold_uncertainty_score":0.19680625},"labels":[{"model":"gemma","categories":["open_science"],"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":"low"}],"label_agreement":"split"},{"id":"W2136741769","doi":"10.1080/02681102.2015.1081868","title":"Viscous Open Data: The Roles of Intermediaries in an Open Data Ecosystem","year":2015,"lang":"en","type":"article","venue":"Information Technology for Development","topic":"E-Government and Public Services","field":"Social Sciences","cited_by":88,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"International Development Research Centre","keywords":"Intermediary; Open data; Business; Ecosystem; Environmental resource management; Knowledge management; Computer science; World Wide Web; Marketing; Environmental science; Ecology","score_opus":0.1319685321627065,"score_gpt":0.38197853332723863,"score_spread":0.25001000116453215,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2136741769","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18705185,0.008318261,0.3412868,0.14318979,0.0008751187,0.0005227995,0.0006683119,0.00075101067,0.31733602],"genre_scores_gemma":[0.93551004,0.0034121163,0.04717626,0.0025743996,0.00036938098,0.00021006558,0.00017996413,0.0001629608,0.010404884],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.97973186,0.012413574,0.0011514538,0.0016854603,0.0035205516,0.001497115],"domain_scores_gemma":[0.942494,0.037575606,0.0055719833,0.007082655,0.0035139108,0.003761977],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.022709597,0.0005964064,0.0006883971,0.0057875374,0.010175982,0.03689556,0.0019149364,0.0047687343,0.0056754136],"category_scores_gemma":[0.032774422,0.0010027966,0.00090569624,0.0085397875,0.028031362,0.05970064,0.02380626,0.006629572,0.0010434142],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003503804,0.000033778862,0.0060111242,0.000111789144,0.000020189023,0.00048769737,0.013116625,0.0006114841,0.00037444476,0.95901537,0.0017359006,0.018446581],"study_design_scores_gemma":[0.00002374476,0.000054765427,0.0027679757,0.0008881828,0.000049778395,0.0006088315,0.029073497,0.0057012187,0.0009883088,0.79379547,0.16596782,0.00008039961],"about_ca_topic_score_codex":0.0044808146,"about_ca_topic_score_gemma":0.0039712833,"teacher_disagreement_score":0.9980851,"about_ca_system_score_codex":0.0065348367,"about_ca_system_score_gemma":0.01287056,"threshold_uncertainty_score":0.12010127},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2144889117","doi":"10.1186/gm262","title":"Towards a data sharing Code of Conduct for international genomic research","year":2011,"lang":"en","type":"article","venue":"Genome Medicine","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":112,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Wellcome Trust","keywords":"Biobank; Data sharing; Context (archaeology); Genomics; Data science; Engineering ethics; Knowledge management; Computer science; Medicine; Bioinformatics; Biology; Engineering; Genetics; Genome","score_opus":0.958708070608172,"score_gpt":0.7003965024361307,"score_spread":0.25831156817204126,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2144889117","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007893616,0.001582916,0.6658031,0.22863151,0.0042068567,0.0042921803,0.00026232743,0.0009257119,0.086401775],"genre_scores_gemma":[0.12000096,0.0009904945,0.8163772,0.04090736,0.0016074919,0.007538904,0.0005805702,0.0006384202,0.011358534],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.44676253,0.36954826,0.060568657,0.019247303,0.086884096,0.016989073],"domain_scores_gemma":[0.37169826,0.35630834,0.023620326,0.09623809,0.12327875,0.02885632],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.48361647,0.001907053,0.003798811,0.006672906,0.01975622,0.051197436,0.0109129455,0.038228244,0.0039723436],"category_scores_gemma":[0.41193873,0.0030751913,0.0043133833,0.0072934856,0.060961466,0.03228576,0.03153252,0.057332978,0.003108119],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003140145,0.00008885261,0.00064506417,0.00013524073,0.000019600928,0.00015045881,0.00796953,0.00056938257,0.00024620633,0.9663018,0.012346328,0.011496239],"study_design_scores_gemma":[0.00013277809,0.0001530053,0.0012217636,0.0023694986,0.000054526692,0.00063508505,0.0078023807,0.00441184,0.0010272441,0.7908803,0.19102071,0.00029089794],"about_ca_topic_score_codex":0.01046165,"about_ca_topic_score_gemma":0.0068627708,"teacher_disagreement_score":0.98908705,"about_ca_system_score_codex":0.021404006,"about_ca_system_score_gemma":0.1092634,"threshold_uncertainty_score":0.6367929},"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"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2164878690","doi":"10.1525/jer.2007.2.1.3","title":"Sharing Data and Results in Ethnographic Research: Why This Should Not Be an Ethical Imperative","year":2007,"lang":"en","type":"article","venue":"Journal of Empirical Research on Human Research Ethics","topic":"Qualitative Research Methods and Ethics","field":"Social Sciences","cited_by":30,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"McMaster University; U.S. Department of Homeland Security","keywords":"Ethnography; Variety (cybernetics); Sociology; Engineering ethics; Research ethics; Relation (database); Politics; Field (mathematics); Epistemology; Public relations; Political science; Law; Computer science; Anthropology","score_opus":0.9682090312510818,"score_gpt":0.7973267805208706,"score_spread":0.17088225073021124,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2164878690","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":"methods","domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018589718,0.016136825,0.23880948,0.69059706,0.009323833,0.0024703222,0.00009421416,0.00036177097,0.023616837],"genre_scores_gemma":[0.44129702,0.0077174017,0.31595463,0.20733649,0.008134648,0.014148139,0.00010766354,0.0005257082,0.0047782995],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.14038448,0.7810988,0.02124252,0.01433287,0.038804185,0.004137149],"domain_scores_gemma":[0.09897628,0.7669228,0.02812664,0.06494318,0.032466546,0.008564637],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.7221875,0.0023390003,0.004927411,0.004134824,0.025393981,0.02582071,0.008575341,0.034461427,0.0031686346],"category_scores_gemma":[0.6821493,0.0037537355,0.0022190823,0.0051825023,0.19481096,0.0674697,0.02988466,0.03832062,0.002406904],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036096503,0.00021747849,0.0038870398,0.0039402586,0.00022622135,0.0009770195,0.28206342,0.0008123347,0.001371973,0.61865664,0.03426753,0.05321905],"study_design_scores_gemma":[0.00029539526,0.00034349196,0.0015881876,0.008193579,0.00009563758,0.001805518,0.06071355,0.0017570802,0.0014425876,0.7923148,0.13110034,0.00034986576],"about_ca_topic_score_codex":0.0041263816,"about_ca_topic_score_gemma":0.0049456153,"teacher_disagreement_score":0.27781248,"about_ca_system_score_codex":0.012165728,"about_ca_system_score_gemma":0.032060064,"threshold_uncertainty_score":0.3425923},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science","scholarly_communication"],"domain":"reproducibility","study_design":"qualitative","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2176602576","doi":"10.15200/winn.144842.20295","title":"Teaching Open Science","year":2015,"lang":"en","type":"dataset","venue":"The Winnower","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Syllabus; Open science; Computer science; Process (computing); Graduate students; Mathematics education; Psychology; Library science; Pedagogy; Mathematics","score_opus":0.1546858394041231,"score_gpt":0.4386236280975025,"score_spread":0.2839377886933794,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2176602576","genre_codex":"other","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017820468,0.026483644,0.07153908,0.27850655,0.02227481,0.0003522641,0.00080638664,0.0024786224,0.57973814],"genre_scores_gemma":[0.21334438,0.029997282,0.05406952,0.06270484,0.009720403,0.00067570584,0.0013051728,0.002691239,0.62549144],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99271166,0.0024726454,0.00027096382,0.0012856317,0.0023229087,0.0009362112],"domain_scores_gemma":[0.984382,0.005564853,0.0007039183,0.0011213799,0.0016505263,0.0065772943],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.009219905,0.00093965646,0.0008065638,0.002052663,0.007967534,0.010946653,0.0016282131,0.0034593558,0.078885406],"category_scores_gemma":[0.017651755,0.00070922956,0.0009693164,0.0014634173,0.0066402764,0.011809697,0.010024515,0.009679651,0.027469749],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000055604563,0.0002560519,0.0014991157,0.00053143856,0.00001613085,0.00034012247,0.027321713,0.00033461585,0.0014463719,0.22334333,0.5392815,0.205574],"study_design_scores_gemma":[0.000008127527,0.0000375501,0.00035951624,0.00039957312,0.0000021107844,0.00016350689,0.0038398847,0.00013678537,0.00035018154,0.056826595,0.9378628,0.000013277083],"about_ca_topic_score_codex":0.0021387585,"about_ca_topic_score_gemma":0.009498203,"teacher_disagreement_score":0.9983718,"about_ca_system_score_codex":0.007085172,"about_ca_system_score_gemma":0.008658569,"threshold_uncertainty_score":0.26389796},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2261976862","doi":"10.3138/jsp.47.2.147","title":"Publishing Undergraduate Research: Linking Teaching and Research through a Dedicated Peer-Reviewed Open Access Journal","year":2016,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"Higher Education Practises and Engagement","field":"Social Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Publishing; Work (physics); Discipline; Process (computing); Peer review; Quality (philosophy); Public relations; Engineering ethics; Sociology; Library science; Political science; Computer science; Engineering; Social science","score_opus":0.5479534976218382,"score_gpt":0.5658860954472149,"score_spread":0.017932597825376706,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2261976862","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06245448,0.088648416,0.19728085,0.25797212,0.12105765,0.009456882,0.0010920684,0.00733211,0.2547055],"genre_scores_gemma":[0.41238585,0.090732664,0.30675927,0.024035694,0.051648293,0.0049793604,0.0017392301,0.0049961586,0.102723524],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.85009134,0.06880609,0.019381916,0.0055096103,0.052356657,0.0038543325],"domain_scores_gemma":[0.39664978,0.22173181,0.047980595,0.09035577,0.13655363,0.10672852],"candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.16416618,0.0008763562,0.0019459326,0.015422801,0.0077509587,0.042769358,0.0035856771,0.005343182,0.022814132],"category_scores_gemma":[0.411885,0.0012251886,0.0008357514,0.014479821,0.014413137,0.02286996,0.024384353,0.0072484068,0.013343504],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000112698115,0.00053925044,0.0047704508,0.0043085786,0.00009742742,0.00066568004,0.014769165,0.00019015514,0.0038552498,0.0332095,0.09458781,0.8428941],"study_design_scores_gemma":[0.00009519782,0.00088053127,0.012102487,0.005816852,0.00008178985,0.0017857595,0.022695117,0.0005125671,0.002699714,0.049554385,0.9035906,0.00018508142],"about_ca_topic_score_codex":0.00057348434,"about_ca_topic_score_gemma":0.0019486535,"teacher_disagreement_score":0.9964143,"about_ca_system_score_codex":0.0047486555,"about_ca_system_score_gemma":0.060800936,"threshold_uncertainty_score":0.8682041},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"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":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2286638398","doi":"10.15353/joci.v12i2.3221","title":"User Centred Methods for Measuring the Value of Open Data","year":2016,"lang":"en","type":"article","venue":"The Journal of Community Informatics","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"International Development Research Centre","keywords":"Open data; Computer science; Key (lock); Measure (data warehouse); Data science; Data quality; Quality (philosophy); Open source; Knowledge management; World Wide Web; Data mining; Engineering; Software; Operations management; Computer security; Metric (unit)","score_opus":0.6407874768984645,"score_gpt":0.5418677551865244,"score_spread":0.09891972171194008,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2286638398","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039313465,0.00083041796,0.9321584,0.0014272687,0.00021791339,0.0050069103,0.0013643325,0.0010879596,0.018593285],"genre_scores_gemma":[0.28113097,0.00028668786,0.7075211,0.0004286971,0.00012122446,0.0075378628,0.00091857073,0.00027988307,0.0017749423],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.5351165,0.3583815,0.020669589,0.014137904,0.06950147,0.0021930397],"domain_scores_gemma":[0.29772294,0.51893395,0.03627321,0.07685195,0.065830566,0.0043873223],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.19837213,0.0019717445,0.0014587542,0.021254541,0.0027125708,0.011065094,0.004913965,0.00391737,0.0035614667],"category_scores_gemma":[0.39556494,0.0012734364,0.0017978828,0.017933067,0.005966787,0.01295467,0.010373471,0.004107169,0.0012083288],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012691485,0.0017260896,0.07768032,0.0035498217,0.0009597534,0.00031859853,0.04693558,0.009654727,0.010187529,0.2301989,0.0092087975,0.6083107],"study_design_scores_gemma":[0.0007644551,0.0047010104,0.11649581,0.0038085815,0.0007583477,0.001548819,0.029942403,0.15892518,0.047539286,0.41123253,0.22219482,0.002088852],"about_ca_topic_score_codex":0.0026333781,"about_ca_topic_score_gemma":0.003068114,"teacher_disagreement_score":0.995086,"about_ca_system_score_codex":0.0051757535,"about_ca_system_score_gemma":0.0060511865,"threshold_uncertainty_score":0.9885499},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2335974841","doi":"10.5334/jeps.401","title":"Registered Reports for Student Research","year":2016,"lang":"en","type":"article","venue":"Journal of European Psychology Students","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Medical education; Class (philosophy); Plan (archaeology); Field (mathematics); Psychology; Public relations; Computer science; Political science; Medicine; Artificial intelligence; History","score_opus":0.9486016462783351,"score_gpt":0.7303958046683219,"score_spread":0.21820584161001322,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2335974841","genre_codex":"editorial","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0014693162,0.008205758,0.019475475,0.3316524,0.39530748,0.014469654,0.015075675,0.011836308,0.20250793],"genre_scores_gemma":[0.017802145,0.012198575,0.055655085,0.19660024,0.15144864,0.035131563,0.017332012,0.016545704,0.49728605],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.80917776,0.08653403,0.045679666,0.008278306,0.044120494,0.0062097623],"domain_scores_gemma":[0.25716963,0.22766903,0.07897755,0.13547929,0.25608942,0.0446151],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.1488837,0.0024359655,0.004431057,0.008167745,0.005342278,0.025491083,0.009189889,0.027818833,0.41927946],"category_scores_gemma":[0.56062365,0.0026385165,0.0045565744,0.011330018,0.0044947835,0.014251128,0.011423874,0.023711002,0.39981762],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012656146,0.0000551226,0.00012602343,0.0015892268,0.000019889303,0.00009401661,0.00015788546,0.000023836137,0.00013422804,0.0036432217,0.95896,0.03507006],"study_design_scores_gemma":[0.00009237472,0.00007606542,0.00031171358,0.0013239409,0.000011679837,0.00011678607,0.00016438546,0.000074161566,0.00016517183,0.0024604052,0.99516946,0.00003382494],"about_ca_topic_score_codex":0.00054403173,"about_ca_topic_score_gemma":0.0011084741,"teacher_disagreement_score":0.8511163,"about_ca_system_score_codex":0.0049240496,"about_ca_system_score_gemma":0.028908718,"threshold_uncertainty_score":0.82832736},"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":["metaresearch","open_science"],"domain":"reporting","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2341380340","doi":"10.5281/zenodo.49700","title":"La Science Ouverte: Un Espace De Visibilité Et De Promotion De La Science","year":2015,"lang":"fr","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Cultural Insights and Digital Impacts","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université Laval","funders":"","keywords":"Political science","score_opus":0.15131101489138107,"score_gpt":0.3338857560635405,"score_spread":0.1825747411721594,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2341380340","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04063361,0.115342595,0.018155698,0.119693294,0.010171194,0.000029781559,0.00077279354,0.00023974026,0.69496137],"genre_scores_gemma":[0.68317574,0.06908394,0.009011243,0.006040739,0.01104407,0.00014948705,0.00050197274,0.0005398051,0.220453],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99576306,0.0018824134,0.00016826374,0.00050886953,0.0012096505,0.00046766325],"domain_scores_gemma":[0.9939575,0.003938769,0.00043390368,0.0004797325,0.00073640863,0.00045367974],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.004313332,0.0012007905,0.0008439308,0.0038688232,0.0060392413,0.019138167,0.000910436,0.005087304,0.018961838],"category_scores_gemma":[0.012261148,0.00042459514,0.00073951937,0.005128279,0.015386544,0.013906936,0.00585079,0.009098729,0.002356671],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007251622,0.000014817697,0.0004073991,0.00011712907,0.0000062803138,0.0000741145,0.006012316,0.00021164339,0.00025811378,0.95900327,0.010101511,0.023720806],"study_design_scores_gemma":[0.000020012714,0.000023254974,0.002256569,0.0003350234,0.000008391981,0.0002884109,0.0045476486,0.0003943185,0.0005156373,0.15869094,0.83288574,0.000033968445],"about_ca_topic_score_codex":0.05235288,"about_ca_topic_score_gemma":0.025163034,"teacher_disagreement_score":0.99908954,"about_ca_system_score_codex":0.011366333,"about_ca_system_score_gemma":0.00651042,"threshold_uncertainty_score":0.10409629},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2375639168","doi":"","title":"Principles for International Registration of Protocol Information and Results from Human Trials of Health-related Interventions-The Ottawa Statement,Part One","year":2006,"lang":"en","type":"article","venue":"Journal of Evidence-Based Medicine","topic":"Pharmaceutical studies and practices","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Statement (logic); Protocol (science); Psychological intervention; Trial registration; Computer science; Clinical trial; Medicine; Alternative medicine; Political science; Law; Pathology; Nursing","score_opus":0.5417398428296815,"score_gpt":0.5303620964796135,"score_spread":0.011377746350067941,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2375639168","genre_codex":"methods","genre_gemma":"other","domain_codex":"methods","domain_gemma":"reporting","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":"reporting","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003929837,0.02323085,0.50143385,0.063971184,0.024519388,0.292123,0.017971395,0.0060104607,0.066810094],"genre_scores_gemma":[0.0144366035,0.006979391,0.62374073,0.020631582,0.0026453466,0.31076202,0.007012089,0.0012032844,0.012588953],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.25275534,0.42653775,0.22923453,0.015678132,0.062119447,0.013674822],"domain_scores_gemma":[0.16317223,0.4062825,0.08125337,0.15839419,0.18047199,0.01042564],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.64706874,0.005531512,0.018701008,0.02455129,0.017742457,0.031920854,0.029204229,0.05752732,0.012844706],"category_scores_gemma":[0.66222936,0.018211206,0.026847422,0.031890877,0.029396372,0.015181091,0.011334592,0.07725605,0.02657185],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.009580787,0.0018776937,0.003621982,0.046579603,0.0041507166,0.0012745092,0.02249246,0.0027066318,0.0040617795,0.2789238,0.45191854,0.17281152],"study_design_scores_gemma":[0.01309203,0.0018534639,0.016915318,0.057176583,0.0058790776,0.0021478792,0.0032379036,0.004222302,0.008142403,0.09832635,0.78675115,0.0022556053],"about_ca_topic_score_codex":0.03259796,"about_ca_topic_score_gemma":0.0487968,"teacher_disagreement_score":0.35293126,"about_ca_system_score_codex":0.027322466,"about_ca_system_score_gemma":0.17600673,"threshold_uncertainty_score":0.43522716},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reporting","study_design":"not_applicable","genre":"methods","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"reporting","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2382972747","doi":"10.34105/j.kmel.2016.08.001","title":"Editorial: Models, technologies and approaches toward widening the open access to learning and education","year":2016,"lang":"en","type":"editorial","venue":"Knowledge Management & E-Learning An International Journal","topic":"Online Learning and Analytics","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lakehead University","funders":"","keywords":"Open learning; Knowledge management; Informal learning; Open education; Globe; Productivity; Business model; Computer science; Business; Marketing; Economic growth; Teaching method; Sociology; World Wide Web; Cooperative learning; Psychology; Economics; Pedagogy","score_opus":0.04908212569322858,"score_gpt":0.37564597554272766,"score_spread":0.3265638498494991,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2382972747","genre_codex":"editorial","genre_gemma":"editorial","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000039585724,0.0025909864,0.0001666413,0.013606952,0.9819339,0.00001337129,0.000027514845,0.000039550967,0.001581486],"genre_scores_gemma":[0.00033590395,0.0030968864,0.00008879622,0.004910034,0.9855806,0.000013855073,0.000019951229,0.000030361414,0.0059235664],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99384993,0.0006584695,0.00066543254,0.0007958357,0.0036428007,0.0003874902],"domain_scores_gemma":[0.9822618,0.0072923885,0.0011472898,0.00049202226,0.0059669577,0.0028395576],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0051263133,0.003940955,0.0033821356,0.0035409855,0.0037768465,0.009886229,0.0036352596,0.010374573,0.0139504885],"category_scores_gemma":[0.017187716,0.0010225241,0.002420147,0.0018596674,0.003080926,0.0067991,0.0018155894,0.014728856,0.011367029],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019810583,0.00001926101,0.000024199004,0.00015963743,0.000008887521,0.00006651531,0.000021459788,0.000032498046,0.00006495225,0.0005689015,0.99410653,0.004907447],"study_design_scores_gemma":[0.00003370274,0.000040601004,0.0002333418,0.00042820655,0.000027370466,0.00021885883,0.000088878944,0.00020751492,0.00012495533,0.0017444323,0.9968335,0.000018694129],"about_ca_topic_score_codex":0.001023199,"about_ca_topic_score_gemma":0.0028256194,"teacher_disagreement_score":0.9963647,"about_ca_system_score_codex":0.0020870008,"about_ca_system_score_gemma":0.003426644,"threshold_uncertainty_score":0.046669066},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2399833365","doi":"","title":"Open Data in Vancouver: The Inspiration and the Vision.","year":2011,"lang":"en","type":"article","venue":"IASSIST Conference","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Computer science; Computer graphics (images); Computer vision","score_opus":0.603460426186831,"score_gpt":0.4682804701029587,"score_spread":0.13517995608387234,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2399833365","genre_codex":"commentary","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02139573,0.06230604,0.007902379,0.8079834,0.00831012,0.00015115138,0.0064480156,0.0003153773,0.085187845],"genre_scores_gemma":[0.44296113,0.104705304,0.045257904,0.07390509,0.0050878376,0.00030306552,0.013235471,0.0009905595,0.31355375],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98809296,0.0026789156,0.0006077593,0.0007754612,0.006220844,0.0016240265],"domain_scores_gemma":[0.9291067,0.018191691,0.0018914742,0.00514285,0.027765034,0.01790219],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.018571083,0.000460454,0.00053157296,0.0031576678,0.0073599215,0.02064526,0.0023178747,0.0048952466,0.010733386],"category_scores_gemma":[0.046774916,0.00060341513,0.00035136676,0.011215091,0.009108307,0.0073148212,0.0071369475,0.008179782,0.001973329],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016432858,0.00007829132,0.013266832,0.00044894338,0.000032843785,0.0003472199,0.0022903744,0.00090187864,0.00044384587,0.09403537,0.6293016,0.2586885],"study_design_scores_gemma":[0.000012286687,0.000018039846,0.010730329,0.000687786,0.000009596343,0.000074686985,0.0044331765,0.0005524134,0.00042141063,0.02531041,0.9577023,0.000047512935],"about_ca_topic_score_codex":0.7962735,"about_ca_topic_score_gemma":0.8730136,"teacher_disagreement_score":0.99768215,"about_ca_system_score_codex":0.03136478,"about_ca_system_score_gemma":0.114898786,"threshold_uncertainty_score":0.4098525},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W2470439906","doi":"10.1038/ng.3499","title":"Analysis of five years of controlled access and data sharing compliance at the International Cancer Genome Consortium","year":2016,"lang":"en","type":"letter","venue":"Nature Genetics","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":16,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Institute for Cancer Research; McGill University","funders":"","keywords":"Biology; Genome; Cancer; Compliance (psychology); Genetics; Computational biology; Gene","score_opus":0.38139663352041053,"score_gpt":0.5693888942973265,"score_spread":0.187992260776916,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2470439906","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9061851,0.0006528499,0.0037657234,0.07702329,0.00035253947,0.000513806,0.0041586254,0.00011155942,0.007236518],"genre_scores_gemma":[0.9798105,0.0001241612,0.0019991542,0.013714228,0.00013409236,0.00083384366,0.001456747,0.00006671385,0.0018605674],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.76122856,0.16029584,0.014651298,0.01030341,0.03876429,0.014756578],"domain_scores_gemma":[0.3216025,0.38637266,0.10550522,0.08594224,0.079162836,0.021414591],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.18205918,0.00023362138,0.0007061569,0.0019701049,0.006565556,0.006119216,0.003423781,0.005575918,0.002574387],"category_scores_gemma":[0.4096178,0.00063309347,0.0008483717,0.006421211,0.0048156125,0.0035042095,0.006514498,0.006425435,0.00045644815],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0026847045,0.0010853783,0.8270976,0.00023672536,0.000760795,0.0014638886,0.03555708,0.0009189604,0.0005530881,0.013651339,0.058268994,0.05772139],"study_design_scores_gemma":[0.00037465684,0.0012391486,0.9020314,0.00059859594,0.00041231,0.00090624206,0.02076352,0.0033037644,0.0017202095,0.0064480146,0.061871465,0.00033070883],"about_ca_topic_score_codex":0.07225588,"about_ca_topic_score_gemma":0.05269331,"teacher_disagreement_score":0.9965762,"about_ca_system_score_codex":0.013740409,"about_ca_system_score_gemma":0.03563632,"threshold_uncertainty_score":0.9628325},"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","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W2530827738","doi":"10.11613/bm.2016.035","title":"IMPACT Observatory: tracking the evolution of clinical trial data sharing and research integrity","year":2016,"lang":"en","type":"review","venue":"Biochemia Medica","topic":"Research Data Management Practices","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Computer Research Institute of Montréal","funders":"FP7 People: Marie-Curie Actions; European Commission","keywords":"Data sharing; Clinical trial; Business; Process (computing); Psychological intervention; Open data; Data integrity; Medicine; Public relations; Political science; Knowledge management; Computer science; Computer security; Alternative medicine; Nursing; World Wide Web","score_opus":0.7990351710026269,"score_gpt":0.6297849765274307,"score_spread":0.16925019447519618,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2530827738","genre_codex":"empirical","genre_gemma":"review","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.30913863,0.015666284,0.3081348,0.08966754,0.002123291,0.056631546,0.065080926,0.019493105,0.1340639],"genre_scores_gemma":[0.61073655,0.003914204,0.3065442,0.0034401729,0.0005184175,0.044404563,0.022673614,0.001776719,0.0059915567],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.572156,0.28764126,0.053688165,0.024584873,0.054586064,0.0073436787],"domain_scores_gemma":[0.18060899,0.45662567,0.13580002,0.11134243,0.09764636,0.017976528],"candidate_categories":["metaresearch","research_integrity"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.4792106,0.0014016757,0.001689774,0.022512222,0.00631971,0.019416463,0.0053493013,0.0030048988,0.0073754825],"category_scores_gemma":[0.6393362,0.002413603,0.0019284775,0.029823093,0.007344704,0.021608941,0.02831391,0.0052305358,0.0025342482],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0027935952,0.000647248,0.36456287,0.01093146,0.0009222694,0.0003723305,0.06814769,0.003949818,0.0021095874,0.04134142,0.074138775,0.430083],"study_design_scores_gemma":[0.0014965103,0.0030106995,0.37989008,0.016043479,0.0012655624,0.00063583365,0.042589452,0.020534659,0.009199006,0.06609812,0.45821744,0.001019088],"about_ca_topic_score_codex":0.01225093,"about_ca_topic_score_gemma":0.011635275,"teacher_disagreement_score":0.9969951,"about_ca_system_score_codex":0.01785802,"about_ca_system_score_gemma":0.06430642,"threshold_uncertainty_score":0.6422261},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"reproducibility","study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science","research_integrity"],"domain":"reproducibility","study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2548535514","doi":"","title":"Sustaining Open Access","year":2016,"lang":"en","type":"article","venue":"University of North Texas Digital Library (University of North Texas)","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Publishing; Electronic publishing; Session (web analytics); Open access publishing; Computer science; World Wide Web; Library science; Political science; The Internet; Law","score_opus":0.02350295721281676,"score_gpt":0.20639585649628636,"score_spread":0.1828928992834696,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2548535514","genre_codex":"other","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010405359,0.0017830753,0.012671716,0.20230988,0.0131142875,0.00034535845,0.00056852406,0.0013137063,0.7574881],"genre_scores_gemma":[0.23665471,0.0023091708,0.008601546,0.10020234,0.011048787,0.0006790077,0.0013644301,0.0011140212,0.63802606],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.986484,0.0024303051,0.0004226422,0.0009835253,0.007701516,0.0019780374],"domain_scores_gemma":[0.95453346,0.010384197,0.0017714901,0.0070348717,0.011446526,0.014829535],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.016991623,0.0003928506,0.00032671692,0.0013282569,0.007006101,0.013731126,0.0020908485,0.0065374714,0.112879686],"category_scores_gemma":[0.062265757,0.0003686912,0.0006525105,0.0010679965,0.0035073396,0.014749997,0.015843004,0.0071331444,0.025465243],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000043842698,0.00010489089,0.00064560946,0.0001287077,0.00000992261,0.00023937745,0.0030148702,0.00010293307,0.0015953156,0.15217254,0.7349207,0.10702134],"study_design_scores_gemma":[0.000010860154,0.00004043342,0.0009922513,0.00008843068,0.0000042331662,0.00007004969,0.0016107816,0.00010172088,0.00057604053,0.02135769,0.97513026,0.000017277323],"about_ca_topic_score_codex":0.0029089623,"about_ca_topic_score_gemma":0.0066397446,"teacher_disagreement_score":0.9979091,"about_ca_system_score_codex":0.0022443305,"about_ca_system_score_gemma":0.011415256,"threshold_uncertainty_score":0.37762022},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2588666165","doi":"10.1186/s13063-017-1816-8","title":"Challenges to complete and useful data sharing","year":2017,"lang":"en","type":"letter","venue":"Trials","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":24,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Hamilton Health Sciences; Population Health Research Institute; McMaster University; St. Joseph’s Healthcare Hamilton; Impact","funders":"","keywords":"Data sharing; Data quality; The Internet; Quality (philosophy); Clinical trial; Data collection; Computer science; Medicine; Data access; Collaboratory; Internet privacy; Data management; Modalities; Data science; World Wide Web; Data mining; Alternative medicine; Business; Database","score_opus":0.9778865347180541,"score_gpt":0.7107495091616779,"score_spread":0.2671370255563762,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2588666165","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":"methods","domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011811263,0.003067089,0.022413788,0.96241695,0.0043147667,0.00050536264,0.00008827212,0.000115177194,0.005897424],"genre_scores_gemma":[0.056724682,0.0068891337,0.0874874,0.81523496,0.02154405,0.0066558667,0.00043007382,0.00040539372,0.0046285344],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.22302644,0.62851286,0.053910565,0.014538713,0.0738541,0.00615732],"domain_scores_gemma":[0.11189113,0.7635477,0.020862967,0.057041097,0.036382213,0.010274871],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.53927433,0.0011940241,0.0041850987,0.0035472517,0.009555996,0.018700391,0.014611654,0.034748185,0.007631347],"category_scores_gemma":[0.74222535,0.0027654476,0.0055133747,0.004438653,0.03343457,0.036724605,0.026822317,0.05359113,0.0054590493],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035263013,0.00019377223,0.0030940245,0.0027383266,0.00040303898,0.003571884,0.020570548,0.0023480298,0.00064211583,0.32446787,0.41581842,0.22579929],"study_design_scores_gemma":[0.00022658605,0.0002078293,0.0012535122,0.006834957,0.00007359982,0.0036320176,0.007570919,0.0033658815,0.00039395958,0.39742517,0.5787494,0.00026618483],"about_ca_topic_score_codex":0.0036287436,"about_ca_topic_score_gemma":0.002795795,"teacher_disagreement_score":0.46072567,"about_ca_system_score_codex":0.013146303,"about_ca_system_score_gemma":0.048711948,"threshold_uncertainty_score":0.56815684},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"reproducibility","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2602562802","doi":"10.1145/3051088","title":"Guilt-free data reuse","year":2017,"lang":"en","type":"article","venue":"Communications of the ACM","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"Stanford Bio-X; Simons Institute for the Theory of Computing, University of California Berkeley; University of Toronto; University of Pennsylvania; Microsoft Research; Alfred P. Sloan Foundation; National Science Foundation","keywords":"Computer science; Reuse; Inference; Process (computing); Set (abstract data type); Data mining; Data set; Statistical inference; Simple (philosophy); Machine learning; Data validation; Artificial intelligence; Database","score_opus":0.23705528447005014,"score_gpt":0.38373756465797776,"score_spread":0.14668228018792762,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2602562802","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007494125,0.0005282303,0.98134065,0.0033537892,0.00014842422,0.00048083396,0.00022548187,0.0012302172,0.005198284],"genre_scores_gemma":[0.36803856,0.00073329336,0.6151727,0.0043201186,0.0007274918,0.0019562948,0.0009849042,0.0011303738,0.0069362787],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.7983206,0.10927827,0.01314596,0.028945116,0.045281537,0.005028552],"domain_scores_gemma":[0.34586903,0.20553726,0.016433386,0.40986684,0.018394927,0.0038985177],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.13672173,0.0023289665,0.004596208,0.0041283476,0.005463113,0.010410225,0.017148253,0.0073964815,0.007156057],"category_scores_gemma":[0.45412353,0.002848975,0.006348056,0.004551127,0.021954242,0.030220015,0.032848194,0.013762041,0.0035028427],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009871159,0.00032301174,0.00678565,0.00078503275,0.0007556559,0.00095236284,0.0048386795,0.020319859,0.003689176,0.70415443,0.010373012,0.24603598],"study_design_scores_gemma":[0.00015249455,0.00018649966,0.0006779141,0.00030598705,0.00017170949,0.0007333475,0.00038075622,0.055273302,0.006253976,0.9183808,0.017384224,0.00009898747],"about_ca_topic_score_codex":0.0019222741,"about_ca_topic_score_gemma":0.0015420171,"teacher_disagreement_score":0.98285174,"about_ca_system_score_codex":0.0036547938,"about_ca_system_score_gemma":0.010771222,"threshold_uncertainty_score":0.7230623},"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":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2607649737","doi":"10.1108/tg-12-2015-0053","title":"Conceptualizing citizen participation in open data use at the city level","year":2017,"lang":"en","type":"article","venue":"Transforming Government People Process and Policy","topic":"E-Government and Public Services","field":"Social Sciences","cited_by":53,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"HEC Montréal","funders":"","keywords":"Open data; Open government; Conceptualization; e-participation; Public relations; Government (linguistics); Construct (python library); Knowledge management; Citizen science; Business; Political science; Computer science","score_opus":0.21344261645953178,"score_gpt":0.44312073901661064,"score_spread":0.22967812255707887,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2607649737","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7755965,0.0009434039,0.067645885,0.02065464,0.00010462281,0.00072037475,0.00018010674,0.00011279254,0.13404158],"genre_scores_gemma":[0.99499726,0.00013369549,0.003322013,0.00023761833,0.000010614738,0.00018650715,0.000036444155,0.000012553944,0.0010633514],"study_design_codex":"qualitative","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.95223755,0.038232695,0.001026885,0.0017322969,0.0032055152,0.0035651005],"domain_scores_gemma":[0.96738106,0.02031274,0.0036376072,0.0025113006,0.0030985766,0.0030588112],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.028470607,0.0007088188,0.0004587806,0.0042973734,0.010343063,0.012889755,0.0022900258,0.0030289304,0.003567716],"category_scores_gemma":[0.02267882,0.00050130545,0.00083614653,0.00495573,0.039071158,0.014707345,0.018451782,0.0037034543,0.0002946002],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000527814,0.00014020166,0.041702867,0.00022771631,0.000032307144,0.0005072326,0.64213383,0.0010381659,0.0008589589,0.2947722,0.0009018628,0.017631775],"study_design_scores_gemma":[0.000027672546,0.00013020741,0.018018998,0.0006849641,0.000046168676,0.0002996706,0.8002516,0.0033042328,0.0006805206,0.109367415,0.06711321,0.000075289885],"about_ca_topic_score_codex":0.025916513,"about_ca_topic_score_gemma":0.01958553,"teacher_disagreement_score":0.028470607,"about_ca_system_score_codex":0.011759054,"about_ca_system_score_gemma":0.015519193,"threshold_uncertainty_score":0.15056872},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2653473579","doi":"10.29173/cais211","title":"Le portail institutionnel Persée à l’épreuve des usages : croiser les approches méthodologiques en Sciences Humaines pour améliorer le partage de connaissances scientifiques en libre accès","year":2013,"lang":"fr","type":"article","venue":"Proceedings of the Annual Conference of CAIS / Actes du congrès annuel de l ACSI","topic":"Cultural Insights and Digital Impacts","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Humanities; Semiotics; Sociology; Art; Philosophy; Linguistics","score_opus":0.21623131729380998,"score_gpt":0.3240195976157355,"score_spread":0.10778828032192553,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2653473579","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.14074023,0.003695597,0.7301641,0.0085131405,0.0002863761,0.0011531946,0.00047076104,0.0012184444,0.11375821],"genre_scores_gemma":[0.59348994,0.002618424,0.3779004,0.00068823475,0.000095052186,0.0016855282,0.00038053634,0.0004670243,0.022674875],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.94758487,0.04162966,0.0019310542,0.002768757,0.005309541,0.0007761029],"domain_scores_gemma":[0.8732215,0.0930997,0.0043408307,0.0147894835,0.01318489,0.0013635746],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.037029415,0.0007774016,0.00078090455,0.009935546,0.0036288744,0.0171173,0.0024148135,0.0025281385,0.015795195],"category_scores_gemma":[0.07913461,0.0007478091,0.001499515,0.00941038,0.011087187,0.016241493,0.0062656254,0.002453852,0.0025386817],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004350587,0.00030751512,0.023125624,0.0024407376,0.0001347918,0.00040885966,0.20102422,0.0015521478,0.005643786,0.29880354,0.0036306062,0.46249312],"study_design_scores_gemma":[0.00021340571,0.0005590086,0.031601954,0.0054540737,0.0003317245,0.0016097676,0.29187408,0.02345944,0.015930815,0.28171748,0.3468881,0.00036018653],"about_ca_topic_score_codex":0.009751333,"about_ca_topic_score_gemma":0.011576365,"teacher_disagreement_score":0.9975852,"about_ca_system_score_codex":0.004921581,"about_ca_system_score_gemma":0.010506259,"threshold_uncertainty_score":0.19583261},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2747139175","doi":"10.3138/jsp.49.1.26","title":"Open Publication, Digital Abundance, and Scarce Labour","year":2017,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"Digital Humanities and Scholarship","field":"Arts and Humanities","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Scholarship; Publishing; Redistribution (election); Process (computing); Set (abstract data type); Sociology; World Wide Web; Data science; Economics; Computer science; Political science; Economic growth; Law","score_opus":0.07946204133224566,"score_gpt":0.27352050488146984,"score_spread":0.19405846354922418,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2747139175","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.17080078,0.01310441,0.045047853,0.14280248,0.00085902464,0.00011247725,0.0002851065,0.00020303641,0.6267848],"genre_scores_gemma":[0.95623744,0.0045268163,0.006108041,0.0046945754,0.0013036422,0.00016401209,0.00010853665,0.00012219607,0.026734719],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9844816,0.007782544,0.00087205286,0.001503422,0.0040701963,0.0012902422],"domain_scores_gemma":[0.94349676,0.036293503,0.0054883454,0.007886195,0.0033286498,0.0035065215],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.014536379,0.00025202017,0.0007774982,0.00291924,0.010588856,0.017975196,0.0020330835,0.0035591,0.021938715],"category_scores_gemma":[0.037047084,0.00043704125,0.00039937507,0.0049092243,0.0383738,0.026918674,0.02066283,0.0037722404,0.0031365573],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007167233,0.000043474203,0.0020407932,0.00021778996,0.000006502254,0.000213378,0.00991339,0.00028773563,0.00028859443,0.94776005,0.003210471,0.0359461],"study_design_scores_gemma":[0.000021273667,0.000049172424,0.00163731,0.0005877022,0.000008053103,0.00030879074,0.011481129,0.00037960958,0.0003009506,0.8614679,0.12373562,0.000022459337],"about_ca_topic_score_codex":0.001516215,"about_ca_topic_score_gemma":0.0020211681,"teacher_disagreement_score":0.9820248,"about_ca_system_score_codex":0.004101006,"about_ca_system_score_gemma":0.0051061683,"threshold_uncertainty_score":0.07687664},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","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":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2748041157","doi":"10.1177/1556264617723137","title":"Transparency of Biobank Access in Canada: An Assessment of Industry Access and the Availability of Information on Access Policies and Resulting Research","year":2017,"lang":"en","type":"article","venue":"Journal of Empirical Research on Human Research Ethics","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":14,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Windsor; University of Toronto","funders":"University of Toronto","keywords":"Biobank; Transparency (behavior); Business; Commercialization; Accountability; Data access; Public access; Internet privacy; Public relations; Computer security; Political science; Computer science; Marketing; Law","score_opus":0.9494966952269831,"score_gpt":0.8060133584040988,"score_spread":0.14348333682288428,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2748041157","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9046152,0.004770338,0.002068908,0.035033666,0.00007259711,0.00058593036,0.0024370162,0.000090934634,0.050325457],"genre_scores_gemma":[0.9943012,0.0013535556,0.0011745063,0.0011252812,0.0000130603885,0.00009202457,0.0003315421,0.00001774518,0.0015911373],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9183228,0.014891633,0.0058323587,0.0029809277,0.046673335,0.01129895],"domain_scores_gemma":[0.7022223,0.12285791,0.05010008,0.008774982,0.09212005,0.023924591],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.053187016,0.00031778315,0.0005910734,0.009305049,0.017789902,0.01141419,0.0029070321,0.0015944389,0.002787322],"category_scores_gemma":[0.153746,0.00078415324,0.0007271192,0.016692247,0.0092619285,0.0034021195,0.008140898,0.0027100993,0.000118514625],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.0006035166,0.00022389906,0.59636,0.0016062318,0.00026109137,0.0020141173,0.23391953,0.0010835403,0.0019553301,0.040706914,0.01711954,0.10414622],"study_design_scores_gemma":[0.000033005843,0.00007793637,0.8626648,0.00094408524,0.000094134,0.0003094251,0.08115474,0.0016403606,0.0007036598,0.0026962205,0.04947012,0.0002114211],"about_ca_topic_score_codex":0.9902035,"about_ca_topic_score_gemma":0.9920358,"teacher_disagreement_score":0.99709296,"about_ca_system_score_codex":0.18902844,"about_ca_system_score_gemma":0.33716723,"threshold_uncertainty_score":0.9406121},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W2749249904","doi":"10.1108/ajim-01-2017-0024","title":"Incorporating data sharing to the reward system of science","year":2017,"lang":"en","type":"article","venue":"Aslib Journal of Information Management","topic":"Research Data Management Practices","field":"Computer Science","cited_by":27,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"National Institutes of Health","keywords":"Computer science; Similarity (geometry); Set (abstract data type); Data science; Originality; Data sharing; Information retrieval; Value (mathematics); Data set; Scientometrics; Scale (ratio); World Wide Web; Psychology; Artificial intelligence","score_opus":0.13976620657204658,"score_gpt":0.3809386428785568,"score_spread":0.2411724363065102,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2749249904","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"incentives","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"incentives","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18965541,0.0039444277,0.6398966,0.08845523,0.0019196784,0.003157568,0.00187549,0.0016615697,0.06943406],"genre_scores_gemma":[0.78582853,0.00072255713,0.20806035,0.0016173286,0.0005269198,0.0011688536,0.00053658854,0.00015030235,0.0013884896],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.5420693,0.37104312,0.028427394,0.017723065,0.035663377,0.005073868],"domain_scores_gemma":[0.25455934,0.50148684,0.058329895,0.13335922,0.044210702,0.008054021],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.30078575,0.00081551436,0.0021989704,0.014451682,0.007942603,0.022262307,0.004987215,0.0031603603,0.00418784],"category_scores_gemma":[0.5976182,0.0011411406,0.0017701221,0.023735432,0.017472731,0.037046514,0.030349698,0.004085178,0.000961457],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008226594,0.00050547323,0.06486609,0.0019901877,0.000551144,0.00030130614,0.015999021,0.01296776,0.00090652896,0.5192867,0.0074195312,0.3743837],"study_design_scores_gemma":[0.000225822,0.0006789789,0.019544302,0.0025750129,0.0003694462,0.0004119706,0.008521958,0.04906228,0.0028873682,0.8420732,0.07330253,0.00034715762],"about_ca_topic_score_codex":0.0059921783,"about_ca_topic_score_gemma":0.0043990645,"teacher_disagreement_score":0.99501276,"about_ca_system_score_codex":0.015325778,"about_ca_system_score_gemma":0.02821679,"threshold_uncertainty_score":0.8622557},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics","open_science"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2763626063","doi":"","title":"Open Data and Big Data Programs in Local Government Policy Analysis","year":2014,"lang":"en","type":"article","venue":"Scholarship@Western (Western University)","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Open data; Big data; Government (linguistics); Local government; Computer science; Open government; Data science; Business; Political science; Public administration; World Wide Web; Data mining","score_opus":0.2551241424000368,"score_gpt":0.3563462926011085,"score_spread":0.10122215020107173,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2763626063","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5389368,0.014267725,0.03613243,0.15337531,0.0002894401,0.00071881665,0.00063442305,0.00010991184,0.2555352],"genre_scores_gemma":[0.9895816,0.0021249999,0.0030649023,0.0013780029,0.00004888026,0.00019618917,0.000069598,0.000017792067,0.0035180673],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.9682188,0.022682672,0.00079097634,0.00088048715,0.004336442,0.0030906398],"domain_scores_gemma":[0.8892967,0.08996137,0.0074608196,0.0024592744,0.0066270856,0.0041947425],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.041912187,0.00017459011,0.0003835036,0.0040674284,0.006517551,0.01174024,0.0012146265,0.0016833713,0.0037598084],"category_scores_gemma":[0.06561894,0.00041318,0.00032001472,0.008749813,0.0152692925,0.008687473,0.0060614278,0.0025452285,0.00016502704],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012248213,0.00014552785,0.038327247,0.0009665641,0.00005092578,0.00032628278,0.05500593,0.004268581,0.00030440232,0.8086155,0.005716167,0.086150326],"study_design_scores_gemma":[0.00007396549,0.0001700237,0.06835468,0.0044527333,0.000094532144,0.00014888015,0.24175213,0.0062386957,0.0020674556,0.317748,0.35876563,0.00013326191],"about_ca_topic_score_codex":0.15489648,"about_ca_topic_score_gemma":0.26240903,"teacher_disagreement_score":0.9987854,"about_ca_system_score_codex":0.051826764,"about_ca_system_score_gemma":0.073217094,"threshold_uncertainty_score":0.37603122},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2766363682","doi":"10.3138/jsp.49.1.66","title":"Evaluating and Promoting Open Data Practices in Open Access Journals","year":2017,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"Research Data Management Practices","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Directory; Scholarly communication; Citation; Open data; Data sharing; Incentive; Open science; Computer science; Data access; Data science; World Wide Web; Business; Political science; Publishing; Medicine; Database","score_opus":0.8265358079998114,"score_gpt":0.6426917435008372,"score_spread":0.18384406449897417,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2766363682","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95159817,0.0019451154,0.013384272,0.011973369,0.00014322117,0.00065777777,0.0002847579,0.00012229309,0.019891063],"genre_scores_gemma":[0.98424757,0.0010863293,0.012652475,0.00061522855,0.00009937186,0.00023699841,0.00015401174,0.000044021956,0.0008639833],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.69953626,0.15881518,0.041540235,0.010828089,0.08319383,0.006086431],"domain_scores_gemma":[0.14535432,0.5369654,0.15670711,0.051932175,0.09237747,0.016663626],"candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.23636778,0.00033374492,0.0008546298,0.01653444,0.0067546186,0.02174595,0.0029230646,0.0021700768,0.0015896851],"category_scores_gemma":[0.62481177,0.0007902996,0.0004712532,0.018553065,0.011353549,0.017486734,0.011723987,0.0033944,0.00058678887],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030482482,0.00050835154,0.5791142,0.0014112087,0.000132622,0.00025885348,0.16555387,0.00077584316,0.0037515303,0.027318511,0.0023206722,0.2185496],"study_design_scores_gemma":[0.00012725449,0.0009363255,0.53632665,0.0038557902,0.00020397342,0.00065623206,0.29997298,0.005586197,0.012224429,0.05828528,0.0814372,0.00038766424],"about_ca_topic_score_codex":0.0056906915,"about_ca_topic_score_gemma":0.0067794793,"teacher_disagreement_score":0.9970769,"about_ca_system_score_codex":0.009818789,"about_ca_system_score_gemma":0.024809018,"threshold_uncertainty_score":0.9416945},"labels":[{"model":"gemma","categories":["metaresearch","scholarly_communication","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"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":"agree"},{"id":"W2766932290","doi":"10.3138/jsp.49.1.5","title":"Traditional versus Open Access Scholarly Journal Publishing: An Economic Perspective","year":2017,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Publishing; Misinformation; Electronic publishing; Scholarly communication; Perspective (graphical); Open access publishing; Public relations; Business; Computer science; Political science; World Wide Web; The Internet; Law","score_opus":0.8658095751057954,"score_gpt":0.6303082985146546,"score_spread":0.23550127659114073,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2766932290","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.17805542,0.0499257,0.14090084,0.10954782,0.0017828757,0.0005607911,0.0010133907,0.00012071779,0.51809245],"genre_scores_gemma":[0.9427169,0.023027997,0.012218623,0.0018235799,0.0021254816,0.00022385217,0.0001560923,0.000041067837,0.017666511],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99298084,0.0029011192,0.00021735216,0.00033210655,0.003002299,0.0005663164],"domain_scores_gemma":[0.963295,0.030546093,0.002644482,0.000807015,0.0016531536,0.001054184],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.007957962,0.00063869817,0.0013078238,0.0038638262,0.0012267645,0.01090852,0.0018388701,0.004692772,0.010886471],"category_scores_gemma":[0.023464762,0.00053463964,0.0011430111,0.004346468,0.006685282,0.011854176,0.002474442,0.0033319346,0.0011499884],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011323043,0.00014152253,0.0013322689,0.00020296598,0.0000578057,0.00010892692,0.00012195514,0.013349137,0.00015963802,0.9674526,0.0020798298,0.014880292],"study_design_scores_gemma":[0.00009457048,0.00024359026,0.0030452663,0.00039234615,0.000082562656,0.00019289232,0.0011388846,0.059858065,0.0003847641,0.9135857,0.020907138,0.00007426767],"about_ca_topic_score_codex":0.0019206675,"about_ca_topic_score_gemma":0.00205615,"teacher_disagreement_score":0.99816114,"about_ca_system_score_codex":0.005372273,"about_ca_system_score_gemma":0.0025976507,"threshold_uncertainty_score":0.042086244},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2770958413","doi":"10.1159/000481682","title":"From the Data on Many, Precision Medicine for “One”: The Case for Widespread Genomic Data Sharing","year":2017,"lang":"en","type":"review","venue":"Biomedicine Hub","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":13,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Genomics","funders":"","keywords":"Interoperability; Data science; Context (archaeology); Precision medicine; Data sharing; Globe; Domain (mathematical analysis); Genomic medicine; Genomics; Health care; Computer science; Scale (ratio); Genome; Computational biology; Geography; World Wide Web; Medicine; Biology; Political science; Genetics; Cartography; Alternative medicine","score_opus":0.33019822203882127,"score_gpt":0.44965099735852004,"score_spread":0.11945277531969878,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2770958413","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000107946726,0.9802846,0.0008499887,0.015415376,0.0010070872,0.000006964624,0.000032623535,0.000018610823,0.002276717],"genre_scores_gemma":[0.001392571,0.9910499,0.0008728755,0.005153725,0.00092326093,0.00001436645,0.000041829513,0.000009907157,0.000541522],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99788046,0.00082476955,0.00019658398,0.00021757082,0.00074741093,0.00013314355],"domain_scores_gemma":[0.9919714,0.005676682,0.00039435696,0.0004329241,0.0010249107,0.00049976696],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.005393961,0.0008074359,0.0014742434,0.002904645,0.0008999244,0.0037871539,0.0020834545,0.0039577372,0.0049386155],"category_scores_gemma":[0.008306016,0.0003933003,0.0009036095,0.003988513,0.004168005,0.010081165,0.00322935,0.007693303,0.0026692168],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006523839,0.000035486417,0.00030605606,0.0146529805,0.00017256055,0.00045463265,0.0005009986,0.00043942197,0.00042111255,0.08487401,0.09439691,0.80368066],"study_design_scores_gemma":[0.0000049870905,0.000014535942,0.00015888552,0.0050014197,0.00003051937,0.000758401,0.00021501146,0.000044510405,0.00008227204,0.018830072,0.9748438,0.000015717445],"about_ca_topic_score_codex":0.0018342896,"about_ca_topic_score_gemma":0.0025791102,"teacher_disagreement_score":0.9979165,"about_ca_system_score_codex":0.0025948663,"about_ca_system_score_gemma":0.0067425463,"threshold_uncertainty_score":0.028526366},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"review","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":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2772565825","doi":"10.1038/nbt.4029","title":"The international MAQC Society launches to enhance reproducibility of high-throughput technologies","year":2017,"lang":"en","type":"article","venue":"Nature Biotechnology","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":38,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Princess Margaret Cancer Centre; University of Toronto; Ontario Institute for Cancer Research","funders":"Biotechnology and Biological Sciences Research Council","keywords":"Reproducibility; Throughput; Emerging technologies; Computer science; Telecommunications; Chemistry; Artificial intelligence; Chromatography","score_opus":0.052898998107023545,"score_gpt":0.40772936774633867,"score_spread":0.35483036963931514,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2772565825","genre_codex":"methods","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.044717215,0.0036353902,0.67638093,0.11190995,0.014392612,0.004718973,0.019186208,0.046619006,0.078439765],"genre_scores_gemma":[0.15648593,0.0016489761,0.7216789,0.026055519,0.0047047995,0.0045782663,0.046136662,0.009756011,0.028954873],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9213642,0.022044105,0.003333588,0.0048651895,0.041715585,0.0066773472],"domain_scores_gemma":[0.6340673,0.060213335,0.019131077,0.08127206,0.16627803,0.03903807],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.1953359,0.0022426515,0.0021195314,0.0054203505,0.0038225981,0.012454309,0.006940656,0.0038749326,0.009871212],"category_scores_gemma":[0.1672353,0.0013168481,0.0032790364,0.0076730857,0.0032829829,0.006416385,0.010547464,0.009463915,0.005234974],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014875026,0.0011893156,0.031774435,0.0012186882,0.0012269112,0.00024143711,0.0020125862,0.0094268145,0.059793342,0.07802026,0.4773993,0.3362095],"study_design_scores_gemma":[0.00047507123,0.0007367432,0.045457404,0.00035999928,0.00026226672,0.00021923069,0.00041225302,0.022892907,0.05023664,0.036898095,0.84175813,0.0002913423],"about_ca_topic_score_codex":0.015220464,"about_ca_topic_score_gemma":0.01716727,"teacher_disagreement_score":0.99305934,"about_ca_system_score_codex":0.010000848,"about_ca_system_score_gemma":0.045368522,"threshold_uncertainty_score":0.9922941},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch"],"domain":"reproducibility","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2791957709","doi":"10.3389/fnins.2018.00047","title":"Open Science Meets Stem Cells: A New Drug Discovery Approach for Neurodegenerative Disorders","year":2018,"lang":"en","type":"review","venue":"Frontiers in Neuroscience","topic":"Pluripotent Stem Cells Research","field":"Biochemistry, Genetics and Molecular Biology","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Montreal Neurological Institute and Hospital; McGill University","funders":"Québec Consortium for Drug Discovery; Fondation Brain Canada","keywords":"Drug discovery; Induced pluripotent stem cell; Open science; Neuroscience; Computer science; Translational research; Stem cell; Computational biology; Data science; Medicine; Bioinformatics; Biology; Embryonic stem cell; Pathology","score_opus":0.04700102043857651,"score_gpt":0.33360719776178527,"score_spread":0.28660617732320876,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2791957709","genre_codex":"commentary","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015446795,0.22539178,0.2584078,0.35186347,0.025970565,0.000800198,0.0018065231,0.0030926384,0.117220215],"genre_scores_gemma":[0.1952019,0.33800858,0.2338922,0.12444263,0.023432357,0.0014239711,0.0023574429,0.00079435157,0.08044661],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.996962,0.0009206643,0.00012767618,0.00031031185,0.0014560837,0.00022325868],"domain_scores_gemma":[0.9969404,0.0010970961,0.0002225273,0.00035586188,0.00058325654,0.00080087036],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.006623906,0.0010072504,0.0011736031,0.0018815268,0.0013740205,0.0048548137,0.0017371493,0.0032106172,0.010654753],"category_scores_gemma":[0.004309987,0.0004691965,0.0012953627,0.0010488279,0.005485145,0.0065691737,0.005049219,0.0069451435,0.004088565],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032689815,0.0003061623,0.0005651144,0.0011336264,0.00013637972,0.00054829963,0.00069442403,0.0014490455,0.021448268,0.42296043,0.13984728,0.4105841],"study_design_scores_gemma":[0.00019287303,0.0004297114,0.00028267014,0.00039616818,0.00007660357,0.00052552624,0.00022434152,0.0011041318,0.0073948135,0.2066027,0.782721,0.000049542537],"about_ca_topic_score_codex":0.001125437,"about_ca_topic_score_gemma":0.0023495404,"teacher_disagreement_score":0.9982628,"about_ca_system_score_codex":0.0027079196,"about_ca_system_score_gemma":0.0045760595,"threshold_uncertainty_score":0.035643697},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2792505333","doi":"10.47622/978-1-928331-56-8","title":"The Social Dynamics of Open Data","year":2017,"lang":"en","type":"book","venue":"African Minds eBooks","topic":"Data Quality and Management","field":"Decision Sciences","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Comisión Nacional de Investigación Científica y Tecnológica; International Development Research Centre","keywords":"Open data; Data collection; Data science; Dynamics (music); Computer science; Open science; Open research; Open source; Management science; World Wide Web; Sociology; Social science; Engineering; Statistics; Mathematics","score_opus":0.412274993420936,"score_gpt":0.4866937168620067,"score_spread":0.0744187234410707,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2792505333","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.018992102,0.037769757,0.071359426,0.26990563,0.004633219,0.00032762237,0.000646609,0.00037658642,0.59598905],"genre_scores_gemma":[0.60195786,0.092731856,0.06402482,0.026173051,0.010014652,0.0019023411,0.0012037271,0.0018428718,0.20014887],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.96313906,0.021016717,0.0011399275,0.0021118107,0.01147941,0.0011130476],"domain_scores_gemma":[0.9102168,0.07608012,0.00246062,0.0054304684,0.0038178975,0.0019941276],"candidate_categories":["sts","open_science"],"consensus_categories":[],"category_scores_codex":[0.029302951,0.00043277937,0.0005072934,0.0049382662,0.008546771,0.02700382,0.0016624476,0.0035646947,0.009153101],"category_scores_gemma":[0.050625943,0.00072137505,0.00065996486,0.00750654,0.03471075,0.034540202,0.01597724,0.007995419,0.0020983077],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000075005396,0.000010067623,0.00027314477,0.00013524729,0.000008534557,0.000073140596,0.016931292,0.00022691839,0.00010711958,0.9233433,0.02795492,0.030928979],"study_design_scores_gemma":[0.000005137843,0.00000850852,0.0003834277,0.00039200284,0.000003737586,0.000109616994,0.006553587,0.00039021924,0.00016884388,0.3418851,0.65008086,0.000018941571],"about_ca_topic_score_codex":0.0029869173,"about_ca_topic_score_gemma":0.0027844275,"teacher_disagreement_score":0.99833757,"about_ca_system_score_codex":0.0087575745,"about_ca_system_score_gemma":0.008634376,"threshold_uncertainty_score":0.1549707},"labels":[{"model":"gemma","categories":["open_science","sts"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2796216221","doi":"10.1177/0963662518761733","title":"Identifying diffusion patterns of research articles on Twitter: A case study of online engagement with open access articles","year":2018,"lang":"en","type":"article","venue":"Public Understanding of Science","topic":"Complex Network Analysis Techniques","field":"Physics and Astronomy","cited_by":76,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; Simon Fraser University","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"Open science; Social media; Science communication; Sociology; Scholarly communication; Citizen science; Ask price; World Wide Web; Altmetrics; Online community; Public engagement; Data science; Public relations; Computer science; Political science; Biology; Science education; Publishing; Business","score_opus":0.50896915729909,"score_gpt":0.49662623253611704,"score_spread":0.01234292476297294,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2796216221","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99491537,0.0000772599,0.0022402601,0.00046578402,0.000007622414,0.00007718722,0.0002654909,0.000024791674,0.0019263396],"genre_scores_gemma":[0.99030906,0.0001856615,0.0072455276,0.00009755554,0.00002524966,0.00012915225,0.00040708468,0.000036543963,0.0015640991],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9970682,0.0014800363,0.0002018005,0.0004018073,0.00054295483,0.0003051607],"domain_scores_gemma":[0.96605366,0.025547387,0.0035917468,0.0016058567,0.0020494687,0.0011520078],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.004231453,0.00038424705,0.00045126225,0.003794928,0.002615267,0.0021988216,0.0009094948,0.0017134113,0.0015625487],"category_scores_gemma":[0.01971113,0.00033158576,0.0004886667,0.004522181,0.001045995,0.003731254,0.0019330345,0.0010229674,0.0005525774],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008167667,0.0018617547,0.61788327,0.0007391648,0.00025352588,0.013043894,0.22274835,0.0032520527,0.02422444,0.008782652,0.0039191456,0.1024749],"study_design_scores_gemma":[0.00014442478,0.00092075975,0.53968847,0.00033461393,0.00029439287,0.007772573,0.31734806,0.063271314,0.017774958,0.011104495,0.04103617,0.00030968903],"about_ca_topic_score_codex":0.00703724,"about_ca_topic_score_gemma":0.012797692,"teacher_disagreement_score":0.9978012,"about_ca_system_score_codex":0.0010949492,"about_ca_system_score_gemma":0.00069878256,"threshold_uncertainty_score":0.022378325},"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":"W2797167981","doi":"10.1016/j.cjca.2018.04.007","title":"Sodium, Chloride, and Cardiovascular Mortality—Open Access Data Is Worth Its Salt","year":2018,"lang":"en","type":"letter","venue":"Canadian Journal of Cardiology","topic":"Electrolyte and hormonal disorders","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"McGill University Health Centre","funders":"","keywords":"Medicine; Sodium; Salt (chemistry); Metallurgy","score_opus":0.12320330815539914,"score_gpt":0.35149768490316496,"score_spread":0.22829437674776581,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2797167981","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006517278,0.0035157907,0.000057604135,0.9728987,0.018028814,0.0000048563343,0.00015427108,0.000019740766,0.0046685017],"genre_scores_gemma":[0.016836451,0.009013027,0.00043575632,0.62981623,0.33271155,0.000031488507,0.0003955332,0.000039182512,0.010720789],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987558,0.000356479,0.00018208209,0.00013205696,0.0004466356,0.00012704918],"domain_scores_gemma":[0.99179924,0.0039018334,0.0005012512,0.00036892103,0.0019274065,0.0015012689],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.001863213,0.00045711154,0.001253446,0.0007198543,0.0022330966,0.0020676572,0.0012534967,0.012856647,0.014642711],"category_scores_gemma":[0.01592856,0.00026065944,0.00076292903,0.0006992012,0.0015875659,0.0025450203,0.00080178404,0.015712563,0.006723365],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013882502,0.000088490146,0.003402367,0.00007054162,0.000028392602,0.001973127,0.000049043516,0.000044019886,0.00009194519,0.0020588343,0.9695859,0.022468567],"study_design_scores_gemma":[0.00043853847,0.00012439686,0.008562016,0.0010543654,0.0000992874,0.0041428143,0.0006715832,0.0006499187,0.00018952704,0.02731934,0.95666444,0.00008371561],"about_ca_topic_score_codex":0.0094172675,"about_ca_topic_score_gemma":0.017527727,"teacher_disagreement_score":0.9987465,"about_ca_system_score_codex":0.0017796186,"about_ca_system_score_gemma":0.0023027675,"threshold_uncertainty_score":0.048984766},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2808252916","doi":"10.1038/s41436-018-0017-5","title":"Development of a consent resource for genomic data sharing in the clinical setting","year":2018,"lang":"en","type":"article","venue":"Genetics in Medicine","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":25,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"National Institute of Child Health and Human Development; National Cancer Institute; National Human Genome Research Institute","keywords":"Data sharing; Informed consent; Resource (disambiguation); Key (lock); Genomic information; Computer science; Medicine; Data science; Internet privacy; Genome; Alternative medicine; Biology; Pathology; Computer security","score_opus":0.1593591139675099,"score_gpt":0.419254181272979,"score_spread":0.2598950673054691,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2808252916","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022651851,0.0007205953,0.74693525,0.034892652,0.0019352629,0.040186416,0.019371156,0.024552127,0.1087547],"genre_scores_gemma":[0.09992935,0.0008226653,0.79493076,0.00572584,0.0009623107,0.027588524,0.024463214,0.0058275512,0.039749812],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.93173784,0.04577492,0.0099149,0.0022651637,0.0076932716,0.0026138595],"domain_scores_gemma":[0.7138975,0.11311555,0.010498192,0.090726346,0.047825582,0.023936776],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.12039241,0.0009540701,0.0012465204,0.006354935,0.0029337478,0.0062325844,0.005045917,0.0035796885,0.082345605],"category_scores_gemma":[0.24501838,0.0013977934,0.0012577417,0.0034021065,0.0019486572,0.009781816,0.014182063,0.00431504,0.03735049],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017369875,0.00094164925,0.010615447,0.0011443064,0.0001147534,0.003707268,0.0056440877,0.0026659917,0.006747781,0.07744888,0.39109161,0.49814132],"study_design_scores_gemma":[0.0009099688,0.00048521903,0.0067329444,0.0025055045,0.00009110785,0.0033541794,0.004003181,0.016197002,0.0116789555,0.07446558,0.87921596,0.00036048362],"about_ca_topic_score_codex":0.004360471,"about_ca_topic_score_gemma":0.0039901854,"teacher_disagreement_score":0.12039241,"about_ca_system_score_codex":0.002920524,"about_ca_system_score_gemma":0.03010617,"threshold_uncertainty_score":0.63670355},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2884538269","doi":"10.2196/10939","title":"User Perspectives of a Web-Based Data-Sharing Platform (Open Humans) on Ethical Oversight in Participant-Led Research: Protocol for a Quantitative Study","year":2018,"lang":"en","type":"article","venue":"JMIR Research Protocols","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Protocol (science); Research ethics; Computer science; Data collection; Data sharing; World Wide Web; Psychology; Data science; Knowledge management; Internet privacy; Medicine; Alternative medicine; Sociology","score_opus":0.9669381264645935,"score_gpt":0.8054270301916884,"score_spread":0.16151109627290516,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2884538269","genre_codex":"protocol","genre_gemma":"protocol","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"protocol","genre_consensus":"protocol","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027712563,0.00005445908,0.0030990872,0.00025895186,0.00012712933,0.99267125,0.0004863467,0.000038700775,0.00049292884],"genre_scores_gemma":[0.000761757,0.000018132723,0.002069031,0.00007591893,0.0000077951745,0.996974,0.000033256947,0.000002292905,0.000057759215],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.8823141,0.09013789,0.012354124,0.004240951,0.005700182,0.0052527343],"domain_scores_gemma":[0.8593613,0.07699949,0.012537937,0.017626565,0.027639354,0.0058354023],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.15722206,0.0035827137,0.0034954704,0.0040685977,0.0076237097,0.004491921,0.0035937186,0.0079807695,0.0289706],"category_scores_gemma":[0.13487785,0.003096292,0.0033278055,0.004512559,0.0059618033,0.00427908,0.00484475,0.0083609205,0.0073089125],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.11546818,0.05745503,0.018740334,0.09282231,0.0011340593,0.0052595837,0.1531803,0.011122447,0.02622035,0.045206923,0.10378713,0.36960337],"study_design_scores_gemma":[0.12560587,0.070867956,0.04625188,0.035415795,0.000842489,0.001276801,0.09045743,0.01261225,0.01873511,0.040899094,0.5553375,0.0016978105],"about_ca_topic_score_codex":0.0023394956,"about_ca_topic_score_gemma":0.0044266167,"teacher_disagreement_score":0.84277797,"about_ca_system_score_codex":0.008331006,"about_ca_system_score_gemma":0.024124049,"threshold_uncertainty_score":0.83147967},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","research_integrity"],"domain":"methods","study_design":"observational","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2886427871","doi":"10.1007/s00439-018-1922-z","title":"Correction to: Canada: will privacy rules continue to favour open science?","year":2018,"lang":"en","type":"erratum","venue":"Human Genetics","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University; McGill University Health Centre","funders":"","keywords":"Biology; Human genetics; Open science; Library science; Genetics; Computer science; Mathematics; Gene","score_opus":0.08976443223311317,"score_gpt":0.40854406229482215,"score_spread":0.31877963006170895,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2886427871","genre_codex":"commentary","genre_gemma":"editorial","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"editorial","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00042094814,0.0012961666,0.00096463115,0.56877136,0.3989413,0.000076631484,0.0034160432,0.00075191056,0.025361001],"genre_scores_gemma":[0.017042806,0.005192066,0.0052261627,0.49055338,0.064374916,0.0001771093,0.0022790928,0.0022126664,0.4129418],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9836977,0.0010725859,0.0016269017,0.0015297792,0.010073883,0.001999147],"domain_scores_gemma":[0.9134662,0.014707691,0.0018959496,0.0031877076,0.06341898,0.0033234358],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.008279758,0.0013194807,0.0017102691,0.0027416693,0.012496288,0.008910769,0.0052246344,0.018894536,0.045633104],"category_scores_gemma":[0.110968396,0.0011299417,0.0013283867,0.004187532,0.0065482175,0.003490493,0.0025607233,0.028952658,0.026380645],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004826736,0.0000013469521,0.00003071068,0.000018000375,0.0000019361237,0.000052273168,0.000063279855,0.000009980544,0.000011304036,0.0018204424,0.99710566,0.0008802938],"study_design_scores_gemma":[0.0000132201085,0.0000035259654,0.0003400052,0.00017353219,0.000013796357,0.000098853976,0.00023446012,0.000112996786,0.00015501895,0.0015871774,0.9972314,0.000036075395],"about_ca_topic_score_codex":0.75515324,"about_ca_topic_score_gemma":0.7588872,"teacher_disagreement_score":0.99477535,"about_ca_system_score_codex":0.035701703,"about_ca_system_score_gemma":0.100090414,"threshold_uncertainty_score":0.49257737},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W2887247001","doi":"","title":"The Open Library at AU: Supporting Open Access and Open Educational Resources","year":2015,"lang":"en","type":"article","venue":"AUSpace (Athabasca University)","topic":"Open Education and E-Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Open educational resources; Open education; Business; World Wide Web; Computer science","score_opus":0.058266006245947854,"score_gpt":0.3265462655149584,"score_spread":0.26828025926901056,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2887247001","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.032891188,0.0022046978,0.069525,0.06631856,0.0025526767,0.00075055106,0.0009052497,0.02623891,0.79861325],"genre_scores_gemma":[0.33828592,0.0025707479,0.13681717,0.015090728,0.0016723135,0.0005651208,0.0018816516,0.0053233006,0.49779302],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99384886,0.001488938,0.00029650048,0.00052889925,0.0026958226,0.0011409859],"domain_scores_gemma":[0.975941,0.003946133,0.0010186378,0.0037406909,0.005230814,0.010122749],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.007030948,0.0003727384,0.00040991607,0.0023247765,0.008149231,0.016079139,0.0027279607,0.0037269825,0.05809358],"category_scores_gemma":[0.022958318,0.0006611654,0.00054206303,0.0020646055,0.0024681073,0.021714566,0.021893073,0.0040852656,0.025080077],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021372124,0.00040892282,0.005210254,0.00049823313,0.000025195863,0.0010250659,0.009372221,0.00058091746,0.0032419863,0.14405812,0.46124128,0.3741242],"study_design_scores_gemma":[0.000025552436,0.000051131465,0.0015557348,0.00014006154,0.000013446401,0.00031297337,0.0017373062,0.0017527383,0.0014420992,0.0166916,0.9762146,0.00006279802],"about_ca_topic_score_codex":0.03458409,"about_ca_topic_score_gemma":0.04799104,"teacher_disagreement_score":0.997272,"about_ca_system_score_codex":0.005047547,"about_ca_system_score_gemma":0.01572762,"threshold_uncertainty_score":0.19434237},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W2887866346","doi":"10.1136/bmjebm-2018-111040","title":"More than 2 billion pairs of eyeballs: Why aren’t you sharing medical knowledge on Wikipedia?","year":2018,"lang":"en","type":"article","venue":"BMJ evidence-based medicine","topic":"Wikis in Education and Collaboration","field":"Social Sciences","cited_by":22,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"","keywords":"Reading (process); Knowledge translation; Internet privacy; World Wide Web; Public relations; Knowledge sharing; Encyclopedia; Resource (disambiguation); Information Dissemination; Computer science; Medical education; Political science; Library science; Knowledge management; Medicine; Law","score_opus":0.11614870997935471,"score_gpt":0.45045097746982166,"score_spread":0.33430226749046693,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2887866346","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09620333,0.03273565,0.007737419,0.7757362,0.0089267045,0.00009599153,0.0022293502,0.0008511896,0.07548412],"genre_scores_gemma":[0.7074915,0.031508945,0.008191612,0.18023525,0.009839619,0.00021090753,0.0027031328,0.0011404739,0.058678586],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9942002,0.0030224645,0.0003011314,0.0006593342,0.0013841295,0.0004326397],"domain_scores_gemma":[0.97650105,0.011557912,0.002390532,0.0014558363,0.0032148773,0.004879835],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.006098276,0.0004545269,0.000451853,0.0020646576,0.007071405,0.00762608,0.0010710934,0.004033742,0.023021696],"category_scores_gemma":[0.036286265,0.0005023506,0.00055910175,0.0020737674,0.00427073,0.016181724,0.005278031,0.0037760914,0.005868815],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030513207,0.00009196724,0.022331217,0.00078472873,0.00016738876,0.000996945,0.022168346,0.00018119763,0.00057520164,0.023735506,0.7142459,0.2144164],"study_design_scores_gemma":[0.000040155865,0.000099486286,0.013827849,0.0015993562,0.00009735989,0.0016425903,0.03511271,0.000689816,0.00054563116,0.03589777,0.9103121,0.00013517714],"about_ca_topic_score_codex":0.010310659,"about_ca_topic_score_gemma":0.014107617,"teacher_disagreement_score":0.9989289,"about_ca_system_score_codex":0.002186321,"about_ca_system_score_gemma":0.0017832698,"threshold_uncertainty_score":0.07701528},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2889522933","doi":"10.5539/res.v10n4p33","title":"A Universal Digital Library Based on Unimpeded Access (and Some Proposals Based on Information Ethics)","year":2018,"lang":"en","type":"article","venue":"Review of European Studies","topic":"Privacy, Security, and Data Protection","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Utopia; Commodity; Set (abstract data type); Universal design; Computer science; Ideal (ethics); Digital library; The Imaginary; Internet privacy; World Wide Web; Knowledge management; Computer security; Data science; Business; Law; Political science; Psychology","score_opus":0.07543377302822572,"score_gpt":0.36295292740925117,"score_spread":0.28751915438102543,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2889522933","genre_codex":"other","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012081037,0.17458068,0.07974211,0.22552374,0.0031362523,0.00021123531,0.00017202355,0.00024756475,0.5043054],"genre_scores_gemma":[0.6388766,0.13905706,0.040080175,0.10049032,0.0047594006,0.0009015581,0.00020797426,0.00023834046,0.075388625],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9783432,0.011340317,0.0011922363,0.0027129974,0.0051645646,0.0012466043],"domain_scores_gemma":[0.9843575,0.009054813,0.0013187083,0.0030783,0.0015805999,0.0006100626],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.014900735,0.00045025314,0.00075607863,0.0038125666,0.0034977414,0.015504811,0.0021179153,0.012449448,0.0059798993],"category_scores_gemma":[0.021173613,0.0005199569,0.00129108,0.0043336907,0.05423028,0.035672653,0.008591954,0.007836716,0.0019086655],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000004048674,0.000006042019,0.000054551085,0.00013589353,0.000002896964,0.000016448988,0.00028972482,0.000051380823,0.00002052926,0.99007696,0.0014695702,0.007872057],"study_design_scores_gemma":[0.000015066806,0.000050028466,0.0005759463,0.0013304292,0.000019883824,0.00038910203,0.0010611453,0.00024529314,0.00035701343,0.65103734,0.34487578,0.000042977987],"about_ca_topic_score_codex":0.003175305,"about_ca_topic_score_gemma":0.002062883,"teacher_disagreement_score":0.99788207,"about_ca_system_score_codex":0.009651052,"about_ca_system_score_gemma":0.011782067,"threshold_uncertainty_score":0.07880354},"labels":[{"model":"gemma","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2895962866","doi":"10.23914/ap.v8i2.152","title":"Open Data as Public Archaeology: The Monumental Archive Project","year":2018,"lang":"en","type":"article","venue":"AP Online Journal in Public Archaeology","topic":"Archaeological Research and Protection","field":"Earth and Planetary Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"College of Engineering, Michigan State University; Michigan State University","keywords":"Value (mathematics); Literacy; Open data; Data sharing; Archaeology; Digital curation; Sociology; History; World Wide Web; Political science; Computer science; Law","score_opus":0.16927500712508403,"score_gpt":0.3809947452148179,"score_spread":0.21171973808973388,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2895962866","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7605539,0.009917567,0.031489268,0.028666507,0.0008656981,0.0012600975,0.009708456,0.00078832253,0.15675029],"genre_scores_gemma":[0.9330934,0.004486675,0.025208263,0.0007060834,0.00025694826,0.000836972,0.0045180563,0.00054505724,0.030348435],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.98764324,0.006899918,0.00045188513,0.00086196273,0.0030052979,0.0011376768],"domain_scores_gemma":[0.9620886,0.019037528,0.0040133703,0.008956667,0.0027575272,0.0031464256],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.017596604,0.0003585653,0.0005309785,0.0032702428,0.009381763,0.014710517,0.0021138478,0.0023476991,0.00493409],"category_scores_gemma":[0.027891835,0.0005585167,0.00028918704,0.009757777,0.011257626,0.0073948856,0.013345482,0.002106101,0.00093976874],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007018335,0.00054440513,0.034649078,0.0014517317,0.00009507003,0.0015138584,0.43106616,0.0013260338,0.0021636372,0.17927568,0.040112913,0.30709952],"study_design_scores_gemma":[0.000091620175,0.00018320758,0.068630405,0.0018002343,0.00009863422,0.0006945733,0.328884,0.001010285,0.002378299,0.019267881,0.5768618,0.000099016295],"about_ca_topic_score_codex":0.07996803,"about_ca_topic_score_gemma":0.14303887,"teacher_disagreement_score":0.9978862,"about_ca_system_score_codex":0.00763017,"about_ca_system_score_gemma":0.016372435,"threshold_uncertainty_score":0.15900517},"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":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2900628939","doi":"10.1038/s41436-018-0370-4","title":"Lessons learned from two decades of BRCA1 and BRCA2 genetic testing: the evolution of data sharing and variant classification","year":2018,"lang":"en","type":"review","venue":"Genetics in Medicine","topic":"BRCA gene mutations in cancer","field":"Biochemistry, Genetics and Molecular Biology","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"National Human Genome Research Institute; National Institutes of Health","keywords":"Breast cancer; Ovarian cancer; Gene; Genetic testing; Identification (biology); Biology; Cancer; Genetic variation; Genetics","score_opus":0.3085729764609272,"score_gpt":0.4407115127165762,"score_spread":0.13213853625564903,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2900628939","genre_codex":"commentary","genre_gemma":"review","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009083605,0.016362041,0.004183006,0.97312874,0.0042345603,0.000008607806,0.00004263749,0.000038287926,0.0010938105],"genre_scores_gemma":[0.083520666,0.06442166,0.024223756,0.7888967,0.030731244,0.00022527043,0.00023463232,0.0005370641,0.0072089937],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.90422,0.06487907,0.0049709785,0.008004384,0.015213421,0.002712093],"domain_scores_gemma":[0.6212569,0.3029656,0.006302216,0.018172657,0.036923654,0.014378994],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.16892067,0.00079530844,0.0014513673,0.0031591903,0.008027715,0.015754446,0.005444654,0.015579435,0.0033225038],"category_scores_gemma":[0.19836582,0.0009320164,0.0013443584,0.003365832,0.04750838,0.030534206,0.011907333,0.03827399,0.0011063836],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019792792,0.00007220999,0.0027171394,0.00078791846,0.00017290824,0.0009159896,0.030873597,0.00084317505,0.00038748453,0.24908254,0.47068402,0.24326503],"study_design_scores_gemma":[0.00004507727,0.00009895333,0.0011438174,0.0024056428,0.00004752444,0.0009693058,0.012524642,0.00065439224,0.00046217046,0.2450765,0.73638344,0.00018849596],"about_ca_topic_score_codex":0.023163404,"about_ca_topic_score_gemma":0.031079788,"teacher_disagreement_score":0.83107936,"about_ca_system_score_codex":0.012105357,"about_ca_system_score_gemma":0.022238757,"threshold_uncertainty_score":0.8933486},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"reproducibility","study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication","open_science"],"domain":"reproducibility","study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2909101430","doi":"10.1111/ddi.12885","title":"Open access solutions for biodiversity journals: Do not replace one problem with another","year":2019,"lang":"en","type":"article","venue":"Diversity and Distributions","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":29,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Memorial University of Newfoundland; McGill University","funders":"","keywords":"Publishing; Biodiversity; Diversity (politics); Process (computing); Library science; Sociology; Political science; Ecology; Computer science; Biology; Law","score_opus":0.1362599909243479,"score_gpt":0.29944774820363573,"score_spread":0.16318775727928783,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2909101430","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001684807,0.01420109,0.03566092,0.8444741,0.04656604,0.00013916781,0.0002662901,0.00252777,0.05447983],"genre_scores_gemma":[0.06511437,0.03173262,0.085061185,0.6006684,0.10302911,0.0011473558,0.0009968011,0.00727888,0.104971275],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.89859575,0.03571782,0.008257857,0.008393035,0.043741543,0.0052939584],"domain_scores_gemma":[0.6699416,0.13384864,0.02257812,0.08097502,0.053350326,0.039306253],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.08414495,0.0017441531,0.0022433654,0.0064954013,0.011892789,0.043753374,0.007330474,0.03481151,0.049582437],"category_scores_gemma":[0.26691726,0.0013038167,0.0019638715,0.0069497097,0.035384923,0.13696277,0.03825483,0.047141064,0.050961055],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012439901,0.00013308128,0.0009585626,0.0012262489,0.00008180576,0.0003632322,0.0051982077,0.00015280604,0.00062836986,0.3754416,0.4628171,0.15287459],"study_design_scores_gemma":[0.000019656609,0.000043462947,0.00020803625,0.0010983445,0.00001817568,0.0005072118,0.0015748917,0.00014553689,0.00016903567,0.122047655,0.87411433,0.000053701577],"about_ca_topic_score_codex":0.0011645567,"about_ca_topic_score_gemma":0.0013037472,"teacher_disagreement_score":0.9926695,"about_ca_system_score_codex":0.0042076567,"about_ca_system_score_gemma":0.016057285,"threshold_uncertainty_score":0.44500637},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"design_other","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2914963886","doi":"10.1002/asi.24180","title":"Identifying the business model dimensions of data sharing: A value‐based approach","year":2019,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"Big Data and Business Intelligence","field":"Business, Management and Accounting","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Business value; Knowledge management; Business model; Computer science; Value (mathematics); Data sharing; Value capture; Value network; Knowledge sharing; Dimension (graph theory); Process (computing); Key (lock); Data science; Value creation; Business; Marketing","score_opus":0.11522505554396606,"score_gpt":0.32393098991079966,"score_spread":0.2087059343668336,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2914963886","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.41845936,0.000747931,0.44203314,0.011624749,0.00008034263,0.0013563271,0.000554991,0.00014872648,0.124994464],"genre_scores_gemma":[0.88662744,0.0002748172,0.110729,0.00015692458,0.000017226814,0.00045935082,0.0002452847,0.000026196307,0.0014637697],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9909695,0.005934478,0.000318477,0.00059432164,0.0015565645,0.0006267624],"domain_scores_gemma":[0.97996014,0.013504122,0.0016776031,0.0015775457,0.0023851097,0.000895343],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008504783,0.0012365739,0.00060219684,0.012330825,0.004015088,0.017089859,0.0025490592,0.0022900966,0.0036300162],"category_scores_gemma":[0.020740183,0.00079919165,0.0015468826,0.009593622,0.009664154,0.019454464,0.007854198,0.0031309268,0.0003062784],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007979259,0.00038620498,0.025169527,0.00032563772,0.000103217724,0.0005287737,0.03219603,0.0049983286,0.0014371222,0.87870526,0.0011725709,0.05489755],"study_design_scores_gemma":[0.00004661916,0.0001222272,0.013494975,0.0008643339,0.000095318,0.0008816626,0.16126835,0.09143414,0.0020608206,0.69455534,0.035068426,0.00010783191],"about_ca_topic_score_codex":0.00500077,"about_ca_topic_score_gemma":0.0060181026,"teacher_disagreement_score":0.017089859,"about_ca_system_score_codex":0.009415702,"about_ca_system_score_gemma":0.006529802,"threshold_uncertainty_score":0.06831598},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication","open_science"],"domain":"reproducibility","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2923998330","doi":"10.1016/j.jhepr.2019.03.001","title":"JHEP Reports: A new EASL open access journal","year":2019,"lang":"en","type":"editorial","venue":"JHEP Reports","topic":"Liver Disease Diagnosis and Treatment","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Hepatology; Publication; Outreach; Impact factor; Audience measurement; Medicine; Internal medicine; Publishing; Political science; Public relations; Medical education; Family medicine; Law","score_opus":0.03957957871228053,"score_gpt":0.40626750418538915,"score_spread":0.36668792547310863,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2923998330","genre_codex":"editorial","genre_gemma":"editorial","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016456847,0.032427017,0.007247812,0.1361424,0.732711,0.00069648767,0.008231195,0.008075099,0.07282328],"genre_scores_gemma":[0.010010649,0.038671788,0.009173187,0.078258716,0.7645352,0.0008060962,0.00875897,0.0053251586,0.08446023],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98056215,0.0025215936,0.0038854205,0.0018167328,0.009721624,0.0014925606],"domain_scores_gemma":[0.80489004,0.057792366,0.021113375,0.013512268,0.07005162,0.03264035],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.021904608,0.0017614261,0.0029392415,0.013238523,0.0028728135,0.024740482,0.003672582,0.008950745,0.117923185],"category_scores_gemma":[0.11667976,0.0016969645,0.0023019724,0.0052892286,0.0035612145,0.009087892,0.0058482857,0.012848582,0.09703373],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000079252146,0.00007640453,0.000602094,0.0010573444,0.00004015269,0.00046684692,0.00007398477,0.000060010996,0.00023051255,0.0029506392,0.92882,0.06554283],"study_design_scores_gemma":[0.000030290468,0.00003766443,0.00066249014,0.0007404422,0.000021780277,0.0006348225,0.00006382524,0.000059713628,0.00015796295,0.0012081532,0.99634933,0.000033547938],"about_ca_topic_score_codex":0.00045362354,"about_ca_topic_score_gemma":0.00052502955,"teacher_disagreement_score":0.9963274,"about_ca_system_score_codex":0.0022302894,"about_ca_system_score_gemma":0.008936084,"threshold_uncertainty_score":0.39449233},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2946706779","doi":"10.31235/osf.io/gns9a","title":"Evaluating Civic Open Data Standards","year":2018,"lang":"en","type":"preprint","venue":"","topic":"E-Government and Public Services","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Interoperability; Standardization; Computer science; Open data; Open government; Documentation; Data science; Open standard; Data quality; World Wide Web; Knowledge management; Business","score_opus":0.402563644893216,"score_gpt":0.5567453614558108,"score_spread":0.15418171656259477,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2946706779","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.75318843,0.007901164,0.12785171,0.011407766,0.0005551621,0.007828174,0.006089513,0.0014332237,0.08374489],"genre_scores_gemma":[0.8451275,0.0016548313,0.14162382,0.00063918,0.0000795444,0.003964224,0.005220409,0.00027679387,0.0014135934],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.62668675,0.18565443,0.055183545,0.010543647,0.118012615,0.003919017],"domain_scores_gemma":[0.33740234,0.3547983,0.060014788,0.04894096,0.19066477,0.008178821],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.28752917,0.0012077125,0.0015095781,0.025557535,0.0035630239,0.010598687,0.0029174834,0.002613018,0.0017566319],"category_scores_gemma":[0.5054933,0.00067449256,0.0015182225,0.029105144,0.0062189903,0.01847127,0.009539528,0.0027064262,0.00043814583],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008675351,0.0019179786,0.20436697,0.011514225,0.0012750857,0.00026938284,0.030618405,0.014513391,0.004756668,0.16171962,0.01766902,0.5505118],"study_design_scores_gemma":[0.0006210072,0.007674369,0.2952558,0.021074966,0.0016836937,0.000868406,0.102354236,0.07600148,0.044591974,0.17215343,0.2765897,0.001130963],"about_ca_topic_score_codex":0.0059713735,"about_ca_topic_score_gemma":0.006907442,"teacher_disagreement_score":0.28752917,"about_ca_system_score_codex":0.018096382,"about_ca_system_score_gemma":0.016898198,"threshold_uncertainty_score":0.8786034},"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":[],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2955083888","doi":"10.5334/dsj-2019-027","title":"Developing a Model Guidelines Addressing Legal Impediments to Open Access to Publicly Funded Research Data in Malaysia","year":2019,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Data access; Computer science; Business; Knowledge management; Database","score_opus":0.8726138537959646,"score_gpt":0.6560074045530535,"score_spread":0.21660644924291106,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2955083888","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12226425,0.0024426836,0.55601853,0.07100996,0.0005218898,0.009929707,0.0035889496,0.0011747254,0.23304935],"genre_scores_gemma":[0.3761988,0.0024551034,0.59543157,0.002745833,0.000073404335,0.0060992385,0.0020886667,0.00012363515,0.014783679],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9555452,0.019769104,0.0060824393,0.0025940654,0.013284896,0.0027243008],"domain_scores_gemma":[0.9080643,0.04199493,0.0106866695,0.0058242385,0.030667767,0.0027621726],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.045470092,0.00078736426,0.00082003407,0.006250395,0.0030731163,0.012542127,0.003387281,0.004101526,0.0036621492],"category_scores_gemma":[0.081729814,0.0010253057,0.0015187517,0.0044178455,0.0034861944,0.01091832,0.0059532165,0.0042735576,0.0010234636],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008924114,0.0004767223,0.015494873,0.001191053,0.000062020255,0.0009010655,0.007178247,0.036765147,0.0011798664,0.8386798,0.017846864,0.080135114],"study_design_scores_gemma":[0.00024942736,0.0005073796,0.013274772,0.0056460034,0.00023172685,0.00097051705,0.02014584,0.11905024,0.0061087282,0.5163679,0.31701678,0.00043073477],"about_ca_topic_score_codex":0.037212554,"about_ca_topic_score_gemma":0.049561106,"teacher_disagreement_score":0.9966127,"about_ca_system_score_codex":0.014231072,"about_ca_system_score_gemma":0.0602402,"threshold_uncertainty_score":0.24047166},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W2984852203","doi":"10.22230/src.2019v10n2a309","title":"Challenges and Achievements in Founding an Open Access Journal on Educational Leadership and Research Use","year":2019,"lang":"en","type":"article","venue":"Scholarly and Research Communication","topic":"Educational Assessment and Improvement","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"General partnership; Publishing; George (robot); Political science; Library science; Public relations; Professional development; Educational leadership; Management; Sociology; Pedagogy; Computer science","score_opus":0.8834125910531597,"score_gpt":0.6643350685157996,"score_spread":0.21907752253736013,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2984852203","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.017178072,0.03747861,0.009060023,0.84656113,0.017276049,0.00027105145,0.00029086438,0.00047843013,0.071405776],"genre_scores_gemma":[0.48680335,0.1455894,0.07757178,0.12805948,0.07315331,0.0013212225,0.0019748774,0.0029774492,0.082549155],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.77767086,0.07775884,0.028266262,0.008174303,0.08995767,0.018172031],"domain_scores_gemma":[0.17936395,0.44008237,0.03526361,0.038432144,0.20379029,0.10306768],"candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.36547133,0.0009911356,0.0024960167,0.017975803,0.015479787,0.0963933,0.007306809,0.017055506,0.02077446],"category_scores_gemma":[0.5312337,0.0014023468,0.0014532254,0.024519153,0.02755031,0.07818957,0.035324704,0.018867144,0.008722287],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022228222,0.0008845424,0.0098549,0.0022252349,0.000108115266,0.000649331,0.019529033,0.00036703335,0.00064591493,0.25748405,0.2009945,0.5070351],"study_design_scores_gemma":[0.00009096319,0.0004567263,0.013300237,0.005264988,0.000062749255,0.00081942335,0.04332713,0.0011297726,0.0010999952,0.110357046,0.82375836,0.00033258868],"about_ca_topic_score_codex":0.0030155785,"about_ca_topic_score_gemma":0.0042832065,"teacher_disagreement_score":0.9926932,"about_ca_system_score_codex":0.008214058,"about_ca_system_score_gemma":0.0551215,"threshold_uncertainty_score":0.7824868},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"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":"design_other","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2989484692","doi":"10.1007/s11192-019-03301-x","title":"Do articles in open access journals have more frequent altmetric activity than articles in subscription-based journals? An investigation of the research output of Finnish universities","year":2019,"lang":"en","type":"article","venue":"Scientometrics","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":105,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Turun Yliopisto","keywords":"Altmetrics; Discipline; Scientometrics; Bibliometrics; Social media; Library science; Computer science; Sociology; Social science; World Wide Web","score_opus":0.838152280754456,"score_gpt":0.6455667087672443,"score_spread":0.1925855719872117,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2989484692","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99326444,0.0010676997,0.00030005226,0.00025024443,0.00003563472,0.000026342084,0.0007693393,0.000010322519,0.0042759487],"genre_scores_gemma":[0.9983316,0.00040191837,0.00018815685,0.00002853516,0.00007264311,0.000019400679,0.0004634517,0.0000071426603,0.00048720487],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99160695,0.0019151017,0.001827177,0.0011128639,0.0027279148,0.0008100394],"domain_scores_gemma":[0.9074108,0.05556415,0.025274768,0.0022608207,0.0065483116,0.0029411982],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.008427206,0.00034334336,0.000693315,0.01350868,0.0011354105,0.005366983,0.0006843494,0.00089534046,0.0052178707],"category_scores_gemma":[0.063900284,0.0002105929,0.0008833063,0.018014345,0.0011731354,0.003730272,0.0029875662,0.00041769666,0.0009842133],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035258528,0.00010411146,0.9596991,0.0006300591,0.00024864508,0.000508191,0.015253242,0.00013283094,0.0011366385,0.00079670944,0.0004198752,0.020718029],"study_design_scores_gemma":[0.000009326968,0.00011982936,0.9823498,0.00009192167,0.00014745127,0.00026849326,0.013581686,0.00023175824,0.00047911066,0.00041786124,0.002282898,0.000019938047],"about_ca_topic_score_codex":0.004726922,"about_ca_topic_score_gemma":0.0032220185,"teacher_disagreement_score":0.9915728,"about_ca_system_score_codex":0.0012262786,"about_ca_system_score_gemma":0.001218159,"threshold_uncertainty_score":0.044567823},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2996762233","doi":"10.1016/j.softx.2021.100658","title":"Open data from the first and second observing runs of Advanced LIGO and Advanced Virgo","year":2021,"lang":"en","type":"article","venue":"SoftwareX","topic":"Pulsars and Gravitational Waves Research","field":"Physics and Astronomy","cited_by":494,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Division of Human Resource Development; Dirección General de Cultura y Patrimonio, Generalitat Valenciana; Conselleria d'Educació, Investigació, Cultura i Esport; Australian Research Council; Science and Technology Facilities Council; Centre Eau Terre Environnement, Institut National de la Recherche Scientifique; Istituto Nazionale di Fisica Nucleare; Universitat de les Illes Balears; Ministry of Education, India; National Research Foundation of Korea; Hungarian Scientific Research Fund; Generalitat Valenciana; Industry Canada; Kavli Foundation; National Natural Science Foundation of China; National Research, Development and Innovation Office; Agencia Estatal de Investigación; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; Govern de les Illes Balears; Nederlandse Organisatie voor Wetenschappelijk Onderzoek; Agence Nationale de la Recherche; Centre National de la Recherche Scientifique; Council of Scientific and Industrial Research, India; National Research Foundation; Nemzeti Kutatási Fejlesztési és Innovációs Hivatal; Abdus Salam International Centre for Theoretical Physics; Instituto Nazionale di Fisica Nucleare; European Commission; Russian Foundation for Basic Research; ICTP South American Institute for Fundamental Research; Canadian Institute for Advanced Research; Ministero dello Sviluppo Economico; Institut des Origines de Lyon; Russian Science Foundation; Ministerio de Ciencia y Tecnología; Leverhulme Trust; Scottish Funding Council; European Regional Development Fund; Scottish Universities Physics Alliance; Conseil Régional, Île-de-France; Ontario Ministry of Economic Development and Innovation; Science and Engineering Research Board; National Science Foundation; Royal Society","keywords":"LIGO; Sky; Sensitivity (control systems); Open source; Gravitational wave; Series (stratigraphy)","score_opus":0.035869667950014665,"score_gpt":0.3476143321169346,"score_spread":0.3117446641669199,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2996762233","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.031604487,0.00023579273,0.004442972,0.00026498147,0.00023963615,0.0001727683,0.9437699,0.007164647,0.012104821],"genre_scores_gemma":[0.02192924,0.00011394079,0.005611961,0.000096637974,0.00007723742,0.00023414723,0.96751976,0.001659091,0.0027579947],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978702,0.00014889192,0.00019511799,0.0003689877,0.0011248479,0.00029200065],"domain_scores_gemma":[0.9940842,0.0009798489,0.000994904,0.0015479999,0.001765521,0.000627563],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0015360344,0.0007907562,0.0007234494,0.004333135,0.00085009943,0.0015135774,0.0009353342,0.0011042615,0.017947366],"category_scores_gemma":[0.0074209613,0.0005059097,0.0007580117,0.0061278758,0.0005337821,0.0014192375,0.0024507188,0.0016810907,0.0143933715],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019729724,0.0006004369,0.07924129,0.0020918706,0.00036688568,0.00041702887,0.001532981,0.0032630137,0.016194211,0.006463892,0.77781904,0.11003632],"study_design_scores_gemma":[0.0002224951,0.00011608203,0.23196213,0.0002340864,0.000107241875,0.0001887368,0.0003960342,0.0019786416,0.009574508,0.0048234416,0.75022864,0.00016799905],"about_ca_topic_score_codex":0.0097358,"about_ca_topic_score_gemma":0.013047647,"teacher_disagreement_score":0.9990647,"about_ca_system_score_codex":0.0005490902,"about_ca_system_score_gemma":0.0014108999,"threshold_uncertainty_score":0.060039937},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W2998766662","doi":"10.7554/elife.52157","title":"Open exploration","year":2020,"lang":"en","type":"article","venue":"eLife","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":23,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Institute of Mental Health; National Institute on Drug Abuse; Social Sciences and Humanities Research Council of Canada; Knut och Alice Wallenbergs Stiftelse; National Science Foundation","keywords":"Open science; Open data; Focus (optics); Open research; Data science; Computer science; Engineering ethics; Political science; Sociology; World Wide Web; Engineering; Mathematics","score_opus":0.9687226010785637,"score_gpt":0.6272201453176606,"score_spread":0.3415024557609031,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2998766662","genre_codex":"methods","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006411664,0.026865585,0.62826294,0.09673294,0.011859476,0.009826278,0.0038770284,0.0030686425,0.21309552],"genre_scores_gemma":[0.2083095,0.017546928,0.6437314,0.043359526,0.0055644596,0.04089014,0.00347137,0.0021338891,0.034992866],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.64518654,0.29176396,0.02159431,0.014775857,0.023572296,0.0031070074],"domain_scores_gemma":[0.28479403,0.57571614,0.016808176,0.09614366,0.022428628,0.004109367],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.2912602,0.002234066,0.004183237,0.008785592,0.005492136,0.01828771,0.0062401826,0.008388541,0.07323016],"category_scores_gemma":[0.55151224,0.0021135372,0.006593472,0.008061446,0.016517716,0.024544308,0.028907344,0.0087434035,0.010101524],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008151419,0.000094166506,0.0012170171,0.016226878,0.0008464704,0.0003460148,0.013425902,0.0010706513,0.00052560697,0.6931985,0.062635906,0.20959772],"study_design_scores_gemma":[0.00037070012,0.00011070106,0.00042827654,0.0127784135,0.00028364398,0.0002941125,0.0032237268,0.0011450917,0.00064533897,0.59366125,0.38698596,0.00007282912],"about_ca_topic_score_codex":0.0010076603,"about_ca_topic_score_gemma":0.0014262825,"teacher_disagreement_score":0.9937598,"about_ca_system_score_codex":0.0045105335,"about_ca_system_score_gemma":0.024681624,"threshold_uncertainty_score":0.87400234},"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":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3022519489","doi":"10.1016/j.buildenv.2020.106848","title":"Current practices and infrastructure for open data based research on occupant-centric design and operation of buildings","year":2020,"lang":"en","type":"article","venue":"Building and Environment","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Australian Research Council; Engineering and Physical Sciences Research Council; Natural Sciences and Engineering Research Council of Canada; Energistyrelsen; National Science Foundation","keywords":"Transparency (behavior); Open data; Metadata; Data sharing; Work (physics); Status quo; Computer science; Open science; Open research; Built environment; Big data; Best practice; Open standard; Data science; Architectural engineering; Engineering; Interoperability; Computer security; World Wide Web; Civil engineering","score_opus":0.31811095319812477,"score_gpt":0.4080395077723598,"score_spread":0.08992855457423504,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3022519489","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0291351,0.017362513,0.84254193,0.03234287,0.0006830976,0.0004205253,0.0009915244,0.0016010216,0.07492147],"genre_scores_gemma":[0.47321194,0.029480344,0.4824641,0.002686157,0.00080581475,0.0009875195,0.0021554718,0.0007360842,0.0074725347],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9722713,0.011537338,0.0023315153,0.0052975924,0.007127881,0.0014344427],"domain_scores_gemma":[0.8551828,0.062247884,0.0073664044,0.04707417,0.02332078,0.004807983],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.031621907,0.0010457778,0.0013271031,0.005131416,0.003978017,0.026975669,0.009413101,0.004256543,0.009143324],"category_scores_gemma":[0.053224467,0.0013973255,0.0014163138,0.009031996,0.016526882,0.03200112,0.010122091,0.0064908336,0.0029918957],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013117475,0.00038115476,0.008984944,0.001629316,0.0000651944,0.000082706814,0.007180664,0.007221092,0.0024906662,0.7357824,0.005693569,0.23035707],"study_design_scores_gemma":[0.00002625825,0.00022216697,0.0045707747,0.003686056,0.00008887839,0.00028243705,0.018380897,0.028102007,0.0073768776,0.6019788,0.3351387,0.00014610104],"about_ca_topic_score_codex":0.008733206,"about_ca_topic_score_gemma":0.006146254,"teacher_disagreement_score":0.9905869,"about_ca_system_score_codex":0.00810098,"about_ca_system_score_gemma":0.012215524,"threshold_uncertainty_score":0.1672346},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"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,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_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","score_opus":0.2872569636919766,"score_gpt":0.3757555810503705,"score_spread":0.08849861735839387,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3024722698","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014035875,0.004986748,0.73493654,0.05464263,0.000987072,0.0005095543,0.00053216994,0.00045864805,0.1889108],"genre_scores_gemma":[0.67685527,0.00660453,0.24168862,0.0075926455,0.002554784,0.00174643,0.00062455755,0.00047589268,0.061857287],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.96825856,0.012740256,0.0031459986,0.0034274762,0.010405793,0.00202186],"domain_scores_gemma":[0.9001599,0.049039,0.006541979,0.025877008,0.014824716,0.00355735],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.028046958,0.00093637337,0.0015784438,0.004178528,0.007923909,0.017969154,0.0039207893,0.009916371,0.008444413],"category_scores_gemma":[0.10084422,0.0015941104,0.0020431702,0.0050748847,0.021686148,0.042062435,0.012147124,0.009426426,0.0035309184],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011587728,0.000007573176,0.00011217675,0.000044743723,0.000012463287,0.00008289494,0.00055837986,0.0005705766,0.00004801716,0.99206793,0.0023059866,0.0041777277],"study_design_scores_gemma":[0.000026045023,0.000011499163,0.000075587275,0.000073034775,0.000016063605,0.00017814356,0.00016065604,0.0019779706,0.00011600847,0.9675925,0.029752064,0.000020525844],"about_ca_topic_score_codex":0.008645134,"about_ca_topic_score_gemma":0.0036322968,"teacher_disagreement_score":0.9960792,"about_ca_system_score_codex":0.0066887187,"about_ca_system_score_gemma":0.008105069,"threshold_uncertainty_score":0.14832824},"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":"W3036127210","doi":"10.17605/osf.io/ht9qn","title":"UBC Library OSF Workshop","year":2020,"lang":"en","type":"article","venue":"OSF Preprints (OSF Preprints)","topic":"Distributed and Parallel Computing Systems","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Computer science; Library science","score_opus":0.021487220061600077,"score_gpt":0.23880727596980253,"score_spread":0.21732005590820244,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3036127210","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004265977,0.001384886,0.0039652092,0.00596944,0.010384228,0.00046139274,0.008145423,0.0047780643,0.9606454],"genre_scores_gemma":[0.008514849,0.0006912568,0.003590094,0.0010657415,0.0009297832,0.0002480215,0.007506184,0.0011897394,0.9762645],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976332,0.00033642078,0.00007251189,0.0002517025,0.0012030739,0.0005030256],"domain_scores_gemma":[0.99643147,0.00022146253,0.00008415803,0.000460022,0.0012324619,0.0015703753],"candidate_categories":["open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0038606504,0.00086658064,0.0005905249,0.0028908425,0.0056236084,0.010108998,0.002336167,0.002470802,0.40756762],"category_scores_gemma":[0.0043672663,0.00045564087,0.00086358027,0.0020056923,0.00066015864,0.0032099718,0.0046918876,0.0022408883,0.2781192],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000041686955,0.000071050126,0.00006683089,0.000040153926,0.000001111282,0.000052610027,0.000094105475,0.000051017894,0.00023103092,0.0029833722,0.96147525,0.034891807],"study_design_scores_gemma":[0.000007059089,0.000015523932,0.0003369939,0.000038225942,9.71115e-7,0.000027928161,0.00018372337,0.000057876885,0.00024266107,0.00080878893,0.99827075,0.000009412625],"about_ca_topic_score_codex":0.012840693,"about_ca_topic_score_gemma":0.02850611,"teacher_disagreement_score":0.99766386,"about_ca_system_score_codex":0.0043941136,"about_ca_system_score_gemma":0.0072583198,"threshold_uncertainty_score":0.8450329},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3037441421","doi":"10.7554/elife.54967","title":"A community-maintained standard library of population genetic models","year":2020,"lang":"en","type":"article","venue":"eLife","topic":"Evolution and Genetic Dynamics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":256,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"National Institute of General Medical Sciences; Dirección General de Asuntos del Personal Académico, Universidad Nacional Autónoma de México; Simons Center for Quantitative Biology, Cold Spring Harbor Laboratory; National Institutes of Health; University of California Institute for Mexico and the United States; Consejo Nacional de Ciencia y Tecnología; Robertson Foundation; National Human Genome Research Institute; Villum Fonden; National Evolutionary Synthesis Center","keywords":"Python (programming language); Computer science; Data science; Population; Resource (disambiguation); Inference; Artificial intelligence; Programming language","score_opus":0.014266963803551146,"score_gpt":0.23201849023861798,"score_spread":0.21775152643506684,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3037441421","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0026182756,0.00063978706,0.616427,0.00057112036,0.0003454142,0.00026759208,0.11525263,0.2347475,0.02913067],"genre_scores_gemma":[0.029638927,0.0022936452,0.54551196,0.001168791,0.00021624252,0.0025002568,0.29183015,0.09680907,0.030031009],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99874985,0.00024721274,0.00014278268,0.0002571547,0.00046072862,0.00014219755],"domain_scores_gemma":[0.9961843,0.001500056,0.00025324515,0.0010172528,0.00079951,0.0002455871],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0026910114,0.0021114724,0.002031507,0.003515246,0.0014379012,0.003537042,0.008034902,0.0026147272,0.08275625],"category_scores_gemma":[0.013331081,0.0023825408,0.0036718743,0.0054195696,0.00074072083,0.0037565269,0.0042696754,0.0044002947,0.07924429],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032766792,0.00027580452,0.005154004,0.0021845894,0.00048695432,0.0007524983,0.0005128496,0.090499304,0.005086982,0.0825556,0.65851295,0.15365082],"study_design_scores_gemma":[0.00028843942,0.00004620381,0.001205652,0.0003987042,0.000113484224,0.0006317216,0.00006394004,0.16867302,0.004270319,0.10001734,0.72409016,0.00020092637],"about_ca_topic_score_codex":0.008765559,"about_ca_topic_score_gemma":0.012809986,"teacher_disagreement_score":0.08275625,"about_ca_system_score_codex":0.0014114361,"about_ca_system_score_gemma":0.0055470467,"threshold_uncertainty_score":0.27684724},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3091824907","doi":"10.1016/s2666-5247(20)30122-1","title":"A multinational listeriosis outbreak and the importance of sharing genomic data","year":2020,"lang":"en","type":"article","venue":"The Lancet Microbe","topic":"Listeria monocytogenes in Food Safety","field":"Biochemistry, Genetics and Molecular Biology","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Canadian Food Inspection Agency","funders":"Centers for Disease Control and Prevention; National Institutes of Health","keywords":"Outbreak; Multinational corporation; Subtyping; Genome; Data sharing; Dissemination; Whole genome sequencing; Data science; Food safety; Food microbiology; Biotechnology; Business; Biology; Genetics; Computer science; Medicine; Virology; Food science; Gene; Telecommunications","score_opus":0.12693532863311782,"score_gpt":0.32590081970032536,"score_spread":0.19896549106720754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3091824907","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06345123,0.13581124,0.01162039,0.7154103,0.007689613,0.00040269445,0.022641288,0.0009939015,0.041979335],"genre_scores_gemma":[0.49979326,0.2339431,0.08237658,0.11329921,0.0071233767,0.0006745818,0.05291371,0.0011461156,0.008730133],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9875947,0.0066675735,0.0015770793,0.0010690492,0.0025529992,0.00053857494],"domain_scores_gemma":[0.9175109,0.04478374,0.0076429774,0.012366737,0.011572205,0.0061233803],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.022468721,0.0005778152,0.0010181449,0.008170518,0.0017593572,0.0072950628,0.0020717327,0.003179966,0.00941026],"category_scores_gemma":[0.08039479,0.00038198227,0.0010201256,0.012677614,0.002199364,0.010034846,0.0076006954,0.004916934,0.0014883513],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005310289,0.00012188078,0.03746769,0.009951173,0.0009923457,0.0057728654,0.009385844,0.00080376747,0.0017738665,0.01389147,0.3141893,0.60511875],"study_design_scores_gemma":[0.0000885599,0.00017707597,0.06818115,0.032027256,0.00061455753,0.005266929,0.01900423,0.00095036865,0.00095874467,0.04020808,0.83224976,0.0002732185],"about_ca_topic_score_codex":0.017954854,"about_ca_topic_score_gemma":0.027046524,"teacher_disagreement_score":0.99792826,"about_ca_system_score_codex":0.0022675453,"about_ca_system_score_gemma":0.008896248,"threshold_uncertainty_score":0.11882734},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W3093456527","doi":"10.12688/f1000research.24887.1","title":"The ELIXIR Human Copy Number Variations Community: building bioinformatics infrastructure for research","year":2020,"lang":"en","type":"preprint","venue":"F1000Research","topic":"Cancer Genomics and Diagnostics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Hospital for Sick Children; Agence Nationale de la Recherche","keywords":"Elixir (programming language); Context (archaeology); Genomics; Copy-number variation; Population; Biology; Computational biology; Bioinformatics; Medicine; Computer science; Genetics; Genome; Gene; Environmental health","score_opus":0.10132440633974632,"score_gpt":0.4412595132739694,"score_spread":0.3399351069342231,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3093456527","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008175395,0.0076388386,0.64445275,0.05454586,0.0032485256,0.0019506112,0.06486294,0.16674331,0.04838182],"genre_scores_gemma":[0.038113896,0.0041779527,0.7448628,0.013440334,0.0016147869,0.0034522833,0.15263467,0.024645243,0.01705807],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98622245,0.0062675164,0.00072736084,0.0026654424,0.0029354042,0.0011818212],"domain_scores_gemma":[0.9605425,0.013430839,0.0016475106,0.008173452,0.008295948,0.007909767],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.046459477,0.0020194242,0.0023323637,0.0048041637,0.0021525985,0.007669456,0.006005218,0.0032184357,0.028077431],"category_scores_gemma":[0.056696117,0.0013780075,0.0017065607,0.0053034546,0.0013069658,0.00942893,0.015738543,0.0055743665,0.048013072],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014072105,0.0001425814,0.0047861678,0.0011730546,0.00022248684,0.0003656674,0.0016946167,0.0031620536,0.010121282,0.04236195,0.7295554,0.20500742],"study_design_scores_gemma":[0.00039818417,0.00016200397,0.0047310647,0.0006372858,0.000106251784,0.00033044076,0.00056926743,0.021963067,0.007440366,0.08065034,0.8828398,0.00017195454],"about_ca_topic_score_codex":0.0041594245,"about_ca_topic_score_gemma":0.004717171,"teacher_disagreement_score":0.9939948,"about_ca_system_score_codex":0.002664575,"about_ca_system_score_gemma":0.012160927,"threshold_uncertainty_score":0.24570417},"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":[],"domain":null,"study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3105389930","doi":"10.12688/wellcomeopenres.16378.2","title":"Fostering global data sharing: highlighting the recommendations of the Research Data Alliance COVID-19 working group","year":2021,"lang":"en","type":"preprint","venue":"Wellcome Open Research","topic":"COVID-19 diagnosis using AI","field":"Medicine","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; Laurentian University; Environment and Climate Change Canada","funders":"Biotechnology and Biological Sciences Research Council; Université de Bordeaux; Horizon 2020; Rural Development Administration; European Commission; Agence Universitaire de la Francophonie; Wellcome Trust; Fundação de Amparo à Pesquisa do Estado de São Paulo; National Institute on Drug Abuse; University of Michigan; National Science Foundation","keywords":"Data sharing; Preparedness; Interoperability; Medical research; Metadata; Knowledge management; Political science; Public relations; Data science; Medicine; Computer science; World Wide Web; Alternative medicine","score_opus":0.8031307428435279,"score_gpt":0.5907681340689679,"score_spread":0.21236260877456004,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3105389930","genre_codex":"commentary","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006090938,0.0046404502,0.02916879,0.94623,0.011230389,0.0007930467,0.00022099336,0.0004899984,0.006617234],"genre_scores_gemma":[0.013987918,0.017041055,0.32543314,0.60950124,0.009686422,0.0040207035,0.0016937532,0.0022311355,0.016404752],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.46766317,0.3474498,0.06254283,0.022611951,0.08220974,0.017522404],"domain_scores_gemma":[0.30834496,0.42674932,0.01978408,0.050945237,0.1561543,0.038022112],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.5368228,0.0034990981,0.003216232,0.010069251,0.015018408,0.06460757,0.016273612,0.044760797,0.007317626],"category_scores_gemma":[0.49117553,0.0038306871,0.006158464,0.0105916355,0.04378767,0.083325356,0.056400258,0.075162195,0.007871823],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008165976,0.00020011328,0.0014283726,0.0027741876,0.00009246621,0.00088026683,0.03173889,0.00087059726,0.00085578824,0.17394616,0.6977135,0.089417934],"study_design_scores_gemma":[0.000054107626,0.000045265733,0.0003491621,0.005443684,0.00003955976,0.000303833,0.01246981,0.00055597984,0.00044117583,0.06915187,0.9109612,0.00018434448],"about_ca_topic_score_codex":0.035537,"about_ca_topic_score_gemma":0.042338397,"teacher_disagreement_score":0.9837264,"about_ca_system_score_codex":0.022396503,"about_ca_system_score_gemma":0.17198612,"threshold_uncertainty_score":0.57118},"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":["metaresearch","open_science","scholarly_communication"],"domain":"reproducibility","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W3114048375","doi":"10.5703/1288284317200","title":"The Time Has Come…For Next-Generation Open Access Models","year":2020,"lang":"en","type":"article","venue":"","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Purdue Pharma (Canada)","funders":"","keywords":"Publication; Variety (cybernetics); Computer science; Set (abstract data type); World Wide Web; Data science; Public relations; Business; Political science; Advertising","score_opus":0.39223384233690894,"score_gpt":0.331854251938978,"score_spread":0.06037959039793095,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3114048375","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01091226,0.02696649,0.0820758,0.74286044,0.01749326,0.00018402073,0.00059848477,0.0010911068,0.117818154],"genre_scores_gemma":[0.33750382,0.057123918,0.20893085,0.26246664,0.024267871,0.0013298291,0.0021965466,0.0017566291,0.10442395],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9899512,0.0035374458,0.0003811347,0.0012413684,0.0035207893,0.0013680091],"domain_scores_gemma":[0.93952143,0.028550457,0.0017354238,0.008026531,0.009163781,0.013002405],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.02876144,0.0010686127,0.0010945168,0.0011026075,0.004302406,0.01578617,0.0043051913,0.0114281075,0.045943692],"category_scores_gemma":[0.04658533,0.0006582544,0.0017418573,0.0014277393,0.010206217,0.060456827,0.009625771,0.018502368,0.011344726],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014698604,0.00013042327,0.00051367923,0.0003658376,0.000034093002,0.00014264105,0.00064956874,0.0013700404,0.00048050488,0.84713125,0.10425437,0.044780694],"study_design_scores_gemma":[0.000040302166,0.00012631278,0.00031570919,0.00047184827,0.000019814079,0.00016542255,0.0011505708,0.0025323667,0.00028078337,0.40096915,0.5938647,0.00006312419],"about_ca_topic_score_codex":0.0036671215,"about_ca_topic_score_gemma":0.0032089285,"teacher_disagreement_score":0.9956948,"about_ca_system_score_codex":0.0055196923,"about_ca_system_score_gemma":0.009662597,"threshold_uncertainty_score":0.15369701},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3122450121","doi":"10.22024/unikent/03/fal.7","title":"What's Feminist About Open Access? A Relational Approach to Copyright in the Academy","year":2011,"lang":"en","type":"article","venue":"University of Kent","topic":"Copyright and Intellectual Property","field":"Business, Management and Accounting","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"York University","funders":"","keywords":"Intellectual property; Feminist legal theory; Sociology; Scholarship; Context (archaeology); Criticism; Political science; Feminism; Law and economics; Law","score_opus":0.1695429822840149,"score_gpt":0.2723435056772202,"score_spread":0.1028005233932053,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3122450121","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.039714873,0.022938756,0.012155186,0.43825534,0.002018357,0.00003097145,0.00006397287,0.000062997875,0.4847595],"genre_scores_gemma":[0.9354678,0.009511787,0.00173098,0.010309397,0.0019541802,0.00004703302,0.00001355842,0.000061248094,0.04090401],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.994206,0.003613925,0.00010211301,0.00045385657,0.0011557493,0.0004682497],"domain_scores_gemma":[0.99293184,0.005498736,0.0003910846,0.00040106836,0.0003706018,0.00040666928],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.005380667,0.00037918933,0.00038523367,0.0013881061,0.011524619,0.014587218,0.0011453644,0.0042161946,0.007959253],"category_scores_gemma":[0.009641198,0.0002075769,0.00031589737,0.0014715939,0.061768055,0.017971773,0.004466584,0.0051293345,0.0005505998],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000065918925,0.000008440386,0.00018378055,0.000032440927,0.0000019063243,0.0000859165,0.032848153,0.000073355026,0.00006502249,0.95589745,0.0041291653,0.0066678314],"study_design_scores_gemma":[0.000006206794,0.000025972447,0.0007508289,0.00036194935,0.0000066503635,0.00027638074,0.05211055,0.0003477508,0.0002986514,0.49030566,0.45548767,0.000021733722],"about_ca_topic_score_codex":0.008669147,"about_ca_topic_score_gemma":0.007512192,"teacher_disagreement_score":0.99885464,"about_ca_system_score_codex":0.01040421,"about_ca_system_score_gemma":0.0038655458,"threshold_uncertainty_score":0.07548821},"labels":[{"model":"gemma","categories":["sts","scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3122462534","doi":"10.2139/ssrn.3157968","title":"A Community Effort to Protect Genomic Data Sharing, Collaboration and Outsourcing","year":2017,"lang":"en","type":"article","venue":"SSRN Electronic Journal","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Vanderbilt University","keywords":"Outsourcing; Business; Data sharing; Knowledge management; Computer science; Marketing; Medicine","score_opus":0.05435773140033389,"score_gpt":0.32169304510683455,"score_spread":0.26733531370650065,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3122462534","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.105937295,0.0012269743,0.70218366,0.08566258,0.0010760255,0.0022089677,0.0002985864,0.0019815632,0.09942432],"genre_scores_gemma":[0.74008715,0.00078838854,0.21369484,0.01284172,0.0006013893,0.0009973648,0.00043079542,0.00029552143,0.030262763],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.957529,0.023634331,0.0009978383,0.003811419,0.011304041,0.0027233388],"domain_scores_gemma":[0.8913805,0.028022762,0.004540658,0.0532558,0.014572598,0.008227698],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.056750767,0.00070977287,0.0008270442,0.0034659985,0.010690454,0.008934686,0.0045941407,0.009723603,0.0060793716],"category_scores_gemma":[0.07995138,0.0007671349,0.0016992658,0.0051813694,0.008791395,0.014382159,0.027623503,0.008313258,0.0013520805],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003044077,0.0010500194,0.0074943556,0.0003359142,0.0001781469,0.00071813684,0.0069674775,0.006851624,0.0057384525,0.7202917,0.032699298,0.21737045],"study_design_scores_gemma":[0.00044030295,0.0009103877,0.004302504,0.0005273351,0.00021647784,0.0019705065,0.008956556,0.056038037,0.017722344,0.59529805,0.31337237,0.00024518344],"about_ca_topic_score_codex":0.0058580614,"about_ca_topic_score_gemma":0.006056916,"teacher_disagreement_score":0.99540585,"about_ca_system_score_codex":0.0039226376,"about_ca_system_score_gemma":0.029794104,"threshold_uncertainty_score":0.3001303},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W3123529604","doi":"10.2139/ssrn.3654298","title":"The Open Science Ecosystem: A Systematic Framework Anchored in Values, Ethics, and FAIRER Data","year":2020,"lang":"en","type":"article","venue":"SSRN Electronic Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Environment and Climate Change Canada","funders":"","keywords":"Ecosystem; Environmental resource management; Open science; Political science; Environmental ethics; Engineering ethics; Environmental science; Ecology; Engineering; Biology; Mathematics; Philosophy; Statistics","score_opus":0.13756659628440568,"score_gpt":0.4012703544478779,"score_spread":0.26370375816347225,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3123529604","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010959045,0.0066406005,0.7572058,0.18502079,0.0018058764,0.0031858592,0.00042744703,0.0006513287,0.0341033],"genre_scores_gemma":[0.19320758,0.0032095003,0.78625774,0.007786379,0.00073551515,0.00444308,0.0003921713,0.00039013714,0.0035779888],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.61164874,0.3120885,0.029551234,0.011210875,0.030783724,0.0047169984],"domain_scores_gemma":[0.4892049,0.31295666,0.02294584,0.10824102,0.046017252,0.020634325],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.44679868,0.0015177066,0.0031155474,0.02550124,0.016992683,0.041537262,0.0068002013,0.014982201,0.003288536],"category_scores_gemma":[0.35149008,0.0027046364,0.0024166545,0.014282142,0.08819296,0.06613578,0.039865635,0.017804239,0.0010902191],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014560396,0.00006374262,0.0017640652,0.00055843045,0.000044315384,0.00012614985,0.018505227,0.00057577103,0.00019397872,0.9322366,0.006502494,0.03941462],"study_design_scores_gemma":[0.000015957377,0.000031524876,0.00050012977,0.0032655022,0.000034371165,0.00013022622,0.012402932,0.0010526503,0.00021042627,0.88827,0.09402981,0.00005636336],"about_ca_topic_score_codex":0.00604694,"about_ca_topic_score_gemma":0.013169038,"teacher_disagreement_score":0.9931998,"about_ca_system_score_codex":0.01907446,"about_ca_system_score_gemma":0.113183744,"threshold_uncertainty_score":0.6821958},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W3126706105","doi":"10.29173/iasl7605","title":"Effective Use of Open Access Resources in a School Library","year":2021,"lang":"en","type":"article","venue":"IASL Annual Conference Proceedings","topic":"Education, Psychology, and Social Research","field":"Social Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Excellence; Educational resources; Quality (philosophy); Process (computing); Computer science; School library; Outcome (game theory); Value (mathematics); Open educational resources; Data collection; World Wide Web; Medical education; Resource (disambiguation); Knowledge management; Multimedia; Psychology; Pedagogy; Medicine; Sociology; Political science; Mathematics","score_opus":0.1637567510523111,"score_gpt":0.4928101036820148,"score_spread":0.3290533526297037,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3126706105","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96079576,0.0006481627,0.0011842509,0.0010524371,0.000024530316,0.00019173049,0.00005878784,0.00020803775,0.03583629],"genre_scores_gemma":[0.9901696,0.00046772003,0.002524748,0.00012744144,0.000028719112,0.00006244854,0.000052671017,0.000037254504,0.0065293247],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9859583,0.0075329286,0.0010506861,0.0007158401,0.0037838789,0.00095833617],"domain_scores_gemma":[0.9727591,0.013791527,0.004882871,0.0033055232,0.002248091,0.0030128905],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.005453897,0.00027859616,0.00062436133,0.004987911,0.004297037,0.010098877,0.0014709226,0.0011170433,0.0053678285],"category_scores_gemma":[0.022026993,0.00038802339,0.00042825378,0.003936626,0.0023677994,0.008510256,0.007987251,0.00081289746,0.0013127695],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005351331,0.005906332,0.17480752,0.0015100632,0.0001326541,0.0038596538,0.23561782,0.0005395056,0.01125374,0.010411961,0.0048211296,0.5506045],"study_design_scores_gemma":[0.00015170395,0.0032004174,0.36297983,0.0012021984,0.00031106928,0.0035871512,0.44821292,0.0012850187,0.0130302515,0.005929348,0.15982176,0.00028833616],"about_ca_topic_score_codex":0.0021231177,"about_ca_topic_score_gemma":0.0047022537,"teacher_disagreement_score":0.9985291,"about_ca_system_score_codex":0.0020220745,"about_ca_system_score_gemma":0.00454639,"threshold_uncertainty_score":0.028843343},"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":["scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W3127346846","doi":"10.3389/fonc.2020.619274","title":"The Essential Role of Open Data and Software for the Future of Ultrasound-Based Neuronavigation","year":2021,"lang":"en","type":"article","venue":"Frontiers in Oncology","topic":"Radiomics and Machine Learning in Medical Imaging","field":"Medicine","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"École de Technologie Supérieure; McGill University; Montreal Neurological Institute and Hospital","funders":"Natural Sciences and Engineering Research Council of Canada; California HIV/AIDS Research Program","keywords":"Neuronavigation; Computer science; Software; Artificial intelligence; Data collection; Graphics; Machine learning; 3D ultrasound; Data science; Medical physics; Information retrieval; Ultrasound; Medicine; Radiology; Magnetic resonance imaging; Computer graphics (images)","score_opus":0.011452106882332131,"score_gpt":0.3337846004590645,"score_spread":0.3223324935767324,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3127346846","genre_codex":"methods","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0045705554,0.003939603,0.8327663,0.03258191,0.004319955,0.0006912824,0.02087652,0.08470672,0.015547181],"genre_scores_gemma":[0.026915837,0.0049737734,0.8049561,0.009692031,0.002498691,0.0016287182,0.096988685,0.045933984,0.006412211],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9616173,0.008395373,0.0064332183,0.005449774,0.016551495,0.0015528515],"domain_scores_gemma":[0.78373456,0.07153756,0.011143282,0.07937274,0.042451866,0.0117600085],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.053481013,0.0026619327,0.00309019,0.008810743,0.0020804685,0.019876055,0.013287096,0.0046871165,0.016963486],"category_scores_gemma":[0.15335622,0.0021530564,0.003691476,0.011027795,0.007070661,0.040553156,0.023178684,0.015273876,0.022013173],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00079310325,0.00047582816,0.005357344,0.0024792538,0.00042958534,0.00050710345,0.0018758276,0.0050429804,0.005424521,0.13387315,0.3123502,0.531391],"study_design_scores_gemma":[0.00013707821,0.00013161892,0.0032716673,0.0021863033,0.000108792454,0.0004210125,0.0006318859,0.013661147,0.006672476,0.19085257,0.78159666,0.00032873068],"about_ca_topic_score_codex":0.0045803688,"about_ca_topic_score_gemma":0.002352793,"teacher_disagreement_score":0.98671293,"about_ca_system_score_codex":0.002814822,"about_ca_system_score_gemma":0.013431616,"threshold_uncertainty_score":0.282838},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W3127643406","doi":"10.36680/j.itcon.2021.002","title":"Open access to Construction IT research articles – developments over the past 25 years","year":2021,"lang":"en","type":"article","venue":"Journal of Information Technology in Construction","topic":"Intellectual Capital and Performance Analysis","field":"Business, Management and Accounting","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Dozen; Publishing; Quarter (Canadian coin); Quality (philosophy); Business model; Engineering; Library science; Political science; Engineering ethics; World Wide Web; Business; History; Computer science; Marketing","score_opus":0.04679358624725741,"score_gpt":0.331948027033694,"score_spread":0.2851544407864366,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3127643406","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.059325866,0.47488362,0.01654699,0.11377872,0.019318124,0.0002687899,0.0017268384,0.0021866956,0.31196436],"genre_scores_gemma":[0.28349486,0.55415606,0.021926973,0.009860802,0.034852423,0.0001769346,0.003234784,0.0010388091,0.09125835],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9800898,0.0028152007,0.0018273807,0.0017477836,0.0122360345,0.0012837871],"domain_scores_gemma":[0.8494637,0.052642107,0.020189404,0.008919766,0.051004454,0.017780589],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.018818583,0.00065624685,0.00071820634,0.020132832,0.003430996,0.021639634,0.0021300935,0.0029138334,0.023139223],"category_scores_gemma":[0.06026116,0.00056410884,0.00075877237,0.039169572,0.008415759,0.014679233,0.006999774,0.0023648716,0.0068485895],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016358182,0.00020576855,0.007493898,0.0061380845,0.000054256056,0.00064382725,0.0041974783,0.000561907,0.0017339591,0.07797736,0.08279933,0.81803054],"study_design_scores_gemma":[0.000012071905,0.00007239227,0.015000258,0.001681508,0.00001586828,0.001024482,0.0029969944,0.00025300324,0.00057226606,0.008600412,0.9697197,0.00005104544],"about_ca_topic_score_codex":0.0034212763,"about_ca_topic_score_gemma":0.0047692684,"teacher_disagreement_score":0.9978699,"about_ca_system_score_codex":0.0058444696,"about_ca_system_score_gemma":0.01596694,"threshold_uncertainty_score":0.099523365},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W3130340903","doi":"10.1002/tax.12442","title":"Growing a Community: The Inaugural #Blackbotanistsweek Recap and Looking Forward","year":2021,"lang":"en","type":"article","venue":"Taxon","topic":"Climate Change Communication and Perception","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Institut National de la Recherche Scientifique","funders":"College of Engineering, Michigan State University; Michigan State University Foundation; Michigan State University; National Science Foundation","keywords":"White (mutation); Indigenous; Social media; The arts; Media studies; Political science; Criminology; Sociology; History; Law; Ecology","score_opus":0.31829697050145933,"score_gpt":0.41575056066798977,"score_spread":0.09745359016653043,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3130340903","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03531571,0.01031653,0.0010681967,0.73545,0.047568236,0.0003655628,0.0011037473,0.00047815006,0.16833375],"genre_scores_gemma":[0.11950138,0.006744298,0.0019360754,0.19951443,0.0075540836,0.0005343428,0.0009145695,0.0007439217,0.66255695],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9972882,0.0006530537,0.00005972437,0.00023794318,0.0006596982,0.0011013888],"domain_scores_gemma":[0.9910097,0.00032884724,0.00026416488,0.00015611776,0.00070682657,0.0075343694],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030685046,0.00079952343,0.0003657467,0.000962616,0.017513078,0.006735145,0.0012530182,0.0051571396,0.06617618],"category_scores_gemma":[0.0049151354,0.0006695486,0.00054784835,0.0005772471,0.0039710966,0.007072366,0.012453899,0.010376145,0.011992284],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004358248,0.00012236352,0.0016786,0.0001293621,0.0000063843295,0.0004170187,0.015381424,0.0000106839225,0.00065845286,0.007883246,0.93284345,0.040825453],"study_design_scores_gemma":[0.0000058396117,0.00004173245,0.002819544,0.00015403028,0.0000019718568,0.000092727816,0.012958217,0.000015803262,0.00008905055,0.0007639591,0.98304313,0.000014007151],"about_ca_topic_score_codex":0.032818418,"about_ca_topic_score_gemma":0.09988419,"teacher_disagreement_score":0.06617618,"about_ca_system_score_codex":0.00426257,"about_ca_system_score_gemma":0.009914442,"threshold_uncertainty_score":0},"labels":[{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"opus","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W3134007699","doi":"10.1177/1745691620979806","title":"Estimating the Prevalence of Transparency and Reproducibility-Related Research Practices in Psychology (2014–2017)","year":2021,"lang":"en","type":"article","venue":"Perspectives on Psychological Science","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":184,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Fonds de Recherche du Québec - Santé; Stiftung Charité; National Institute on Alcohol Abuse and Alcoholism; Einstein Stiftung Berlin; Laura and John Arnold Foundation","keywords":"Transparency (behavior); Credibility; Psychology; Confidence interval; Medicine; Clinical psychology; Social psychology; Political science; Law","score_opus":0.8303977243332353,"score_gpt":0.6710300330829925,"score_spread":0.15936769125024286,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3134007699","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8610749,0.058109183,0.031541325,0.016126849,0.0010051528,0.0021632134,0.017897377,0.0005537924,0.011528219],"genre_scores_gemma":[0.96689713,0.006745595,0.01645031,0.0017771852,0.0004764674,0.001527876,0.005307491,0.00022555115,0.0005924855],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.5471902,0.16803937,0.14503326,0.038260605,0.095672384,0.0058041913],"domain_scores_gemma":[0.099170774,0.50357515,0.2835596,0.05590652,0.05458378,0.0032041431],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.29123622,0.000719533,0.0016835493,0.036187675,0.0029385707,0.0069432296,0.0031823036,0.002954961,0.0019631137],"category_scores_gemma":[0.7000803,0.0016590189,0.0032624712,0.023884937,0.00712554,0.008934429,0.00894888,0.002873614,0.0006784839],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00076558406,0.00013149565,0.83226115,0.015016838,0.001793827,0.0005886025,0.022247303,0.00074454874,0.0017743448,0.006666564,0.011476558,0.106533214],"study_design_scores_gemma":[0.00018856053,0.00037834785,0.8735511,0.021540498,0.0025064629,0.004225715,0.011900329,0.0052356925,0.006311754,0.0071253753,0.066636756,0.00039947702],"about_ca_topic_score_codex":0.0070048235,"about_ca_topic_score_gemma":0.008128649,"teacher_disagreement_score":0.7087638,"about_ca_system_score_codex":0.0047981213,"about_ca_system_score_gemma":0.009592826,"threshold_uncertainty_score":0.8740319},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3134086897","doi":"10.2196/26718","title":"Clinical Trial Data Sharing for COVID-19–Related Research","year":2021,"lang":"en","type":"article","venue":"Journal of Medical Internet Research","topic":"COVID-19 Clinical Research Studies","field":"Medicine","cited_by":32,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Sample size determination; Context (archaeology); Statistical power; Clinical trial; Pandemic; Sample (material); Coronavirus disease 2019 (COVID-19); Data sharing; Medicine; Psychology; Data science; Computer science; Statistics; Alternative medicine; Mathematics; Geography","score_opus":0.775337627386123,"score_gpt":0.7283670039823996,"score_spread":0.04697062340372338,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3134086897","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":"methods","domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003963518,0.048055116,0.3321323,0.43310335,0.04210509,0.03630707,0.029774994,0.005159008,0.069399565],"genre_scores_gemma":[0.090729386,0.029870205,0.5407571,0.14369898,0.028112909,0.1286108,0.024164889,0.0042364136,0.009819404],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.10946121,0.78318995,0.0655835,0.016206456,0.022375615,0.0031833495],"domain_scores_gemma":[0.03363104,0.71490717,0.052695155,0.15823703,0.029708643,0.010820996],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.6997608,0.0026080417,0.01092985,0.018383462,0.0043238257,0.0360629,0.014952604,0.019525332,0.047155097],"category_scores_gemma":[0.8747197,0.005092983,0.01520053,0.030169575,0.0150367655,0.021988759,0.033787932,0.025001314,0.017595468],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0072929594,0.0003165498,0.005118042,0.053044584,0.010914032,0.0006332889,0.0069607454,0.006289141,0.00084738806,0.28413993,0.2842597,0.34018362],"study_design_scores_gemma":[0.0041513853,0.0008645779,0.0032398354,0.058929022,0.002038495,0.0005774645,0.001157788,0.0050314204,0.0007971381,0.36874303,0.5539213,0.0005486325],"about_ca_topic_score_codex":0.002397674,"about_ca_topic_score_gemma":0.0020458729,"teacher_disagreement_score":0.9850474,"about_ca_system_score_codex":0.014595398,"about_ca_system_score_gemma":0.076539464,"threshold_uncertainty_score":0.37024844},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication","open_science"],"domain":"reproducibility","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W3135000317","doi":"10.1186/s41073-021-00108-4","title":"Steps toward preregistration of research on research integrity","year":2021,"lang":"en","type":"article","venue":"Research Integrity and Peer Review","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Women's College Research Institute","keywords":"Research integrity; USable; Psychology; Data integrity; Set (abstract data type); Documentation; Data collection; Qualitative research; Computer science; Data science; Public relations; Sociology; Political science; World Wide Web; Database; Social science","score_opus":0.6142879617107355,"score_gpt":0.5984877638154373,"score_spread":0.015800197895298185,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3135000317","genre_codex":"commentary","genre_gemma":"methods","domain_codex":"methods","domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02022074,0.010702402,0.39842138,0.42332825,0.02507076,0.07312071,0.0011627069,0.007513729,0.040459424],"genre_scores_gemma":[0.1058921,0.005807089,0.7112831,0.06841042,0.008375788,0.07879194,0.0015228532,0.0025455763,0.017371213],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.109699085,0.656961,0.13828017,0.015861616,0.062127016,0.017071085],"domain_scores_gemma":[0.032396212,0.3979704,0.08739,0.20812483,0.25210443,0.022014122],"candidate_categories":["metaresearch","research_integrity"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.8588083,0.0031465655,0.0065403366,0.030585917,0.019678153,0.039076682,0.020606639,0.032024466,0.014491671],"category_scores_gemma":[0.86620355,0.008069797,0.010293717,0.02505548,0.024136676,0.046583995,0.03901731,0.061458793,0.017558744],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0023337498,0.0014841545,0.011655942,0.015331324,0.0007227455,0.0015313134,0.101510614,0.0016038357,0.00716854,0.09452978,0.2577533,0.5043747],"study_design_scores_gemma":[0.0008801081,0.0019995752,0.021384584,0.021890013,0.00048837764,0.001347782,0.03308044,0.0027289984,0.011583338,0.05315878,0.85029954,0.0011585231],"about_ca_topic_score_codex":0.00749143,"about_ca_topic_score_gemma":0.0076143323,"teacher_disagreement_score":0.96797556,"about_ca_system_score_codex":0.025874438,"about_ca_system_score_gemma":0.24135989,"threshold_uncertainty_score":0.18773305},"labels":[{"model":"gemma","categories":["metaresearch","research_integrity"],"domain":"methods","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","research_integrity","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W3136742816","doi":"10.1016/j.respol.2021.104226","title":"The fall of the innovation empire and its possible rise through open science","year":2021,"lang":"en","type":"article","venue":"Research Policy","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":88,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Bill and Melinda Gates Foundation","keywords":"Incentive; Productivity; Intellectual property; General partnership; Open innovation; Economics; Open science; Knowledge creation; Public economics; Business; Economic growth; Political science; Market economy; Management; Law; Operations management","score_opus":0.8515611675883188,"score_gpt":0.720693447368452,"score_spread":0.1308677202198668,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3136742816","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15395771,0.068255894,0.043067854,0.48914433,0.00068097963,0.00007687133,0.00024807482,0.00024207852,0.24432625],"genre_scores_gemma":[0.9568396,0.017436836,0.01033531,0.0058085164,0.0010540924,0.00011927233,0.0000644706,0.00007978759,0.008262068],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9880486,0.0052286955,0.0005697409,0.0015399923,0.0033178022,0.0012951603],"domain_scores_gemma":[0.947058,0.03742595,0.004244613,0.0045893523,0.0042805015,0.0024014849],"candidate_categories":["bibliometrics","open_science"],"consensus_categories":[],"category_scores_codex":[0.02401341,0.0003357236,0.0012773502,0.0039081317,0.0040717307,0.015280093,0.0012512518,0.006484192,0.0064413333],"category_scores_gemma":[0.050356016,0.00059094443,0.00072884734,0.0039763586,0.036397364,0.0312771,0.0072641564,0.006958301,0.00097220065],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000049282044,0.000036544592,0.0019982273,0.00013785606,0.000015067906,0.000066851346,0.000980839,0.0007405308,0.00014341065,0.97081506,0.0014837312,0.02353254],"study_design_scores_gemma":[0.000031862528,0.00004453151,0.0041272086,0.0003109893,0.000007095362,0.000102950755,0.0010676683,0.0015032719,0.00023350008,0.9462921,0.046254057,0.000024672343],"about_ca_topic_score_codex":0.0034285644,"about_ca_topic_score_gemma":0.0025773507,"teacher_disagreement_score":0.9987487,"about_ca_system_score_codex":0.010427501,"about_ca_system_score_gemma":0.008309988,"threshold_uncertainty_score":0.12699658},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"incentives","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3145008074","doi":"10.17150/2308-6203.2021.10(1).22-38","title":"New Media for Science: Key Features of Platforms with Open Peer Review","year":2021,"lang":"en","type":"article","venue":"Theoretical and Practical Issues of Journalism","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Openness to experience; Scope (computer science); Key (lock); Computer science; Open science; Publishing; Process (computing); Peer review; Data science; Engineering ethics; Political science; Public relations; Engineering; Psychology; Computer security","score_opus":0.45154714619089453,"score_gpt":0.6087514323964822,"score_spread":0.15720428620558768,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3145008074","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09355972,0.020704253,0.30530074,0.0477356,0.008685192,0.0051189153,0.0022402906,0.00651159,0.51014376],"genre_scores_gemma":[0.636281,0.008818797,0.28672156,0.0029382594,0.011699689,0.0046144575,0.001102152,0.0020917885,0.04573233],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.9626174,0.016032493,0.0025137884,0.0027253898,0.0145598715,0.001551079],"domain_scores_gemma":[0.8529591,0.078695066,0.019531893,0.026929382,0.015336549,0.006547895],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.023984494,0.0009523546,0.0008883317,0.008527809,0.0042465143,0.024578333,0.0028507418,0.0026999114,0.019248333],"category_scores_gemma":[0.1294653,0.0009162098,0.0012705505,0.011683844,0.005039075,0.027419375,0.013397356,0.0031866385,0.006623646],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004455151,0.0003724037,0.015055596,0.002679231,0.00018573373,0.0011349199,0.011387857,0.0014061152,0.0041494314,0.41750988,0.049229167,0.49644414],"study_design_scores_gemma":[0.00013759256,0.00032282725,0.018666316,0.0016282084,0.00022692904,0.0027949617,0.006174613,0.0056323307,0.0032989073,0.2739711,0.68684036,0.00030571752],"about_ca_topic_score_codex":0.0007703901,"about_ca_topic_score_gemma":0.0010808896,"teacher_disagreement_score":0.9760155,"about_ca_system_score_codex":0.002724537,"about_ca_system_score_gemma":0.006238016,"threshold_uncertainty_score":0.12684363},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","scholarly_communication","open_science"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3154878857","doi":"10.31234/osf.io/vj5tu","title":"Centering inclusivity in the design of online conferences - An OHBM - Open Science perspective","year":2021,"lang":"en","type":"preprint","venue":"","topic":"Conferences and Exhibitions Management","field":"Social Sciences","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Montfort Hospital; Montreal Neurological Institute and Hospital; Polytechnique Montréal; Concordia University; McGill University","funders":"National Institute of Mental Health; Medical Research Council; Directorate for Biological Sciences; National Institutes of Health; Vlaamse regering; Ministero dell’Istruzione, dell’Università e della Ricerca; Fonds Wetenschappelijk Onderzoek; National Science Foundation; University of Queensland; European Commission; AXA Research Fund; King's College London; National Imaging Facility; National Institute for Health and Care Research; Elsass Fonden; Dipartimenti di Eccellenza; Biotechnology and Biological Sciences Research Council; University of Sydney; National Center for Advancing Translational Sciences; Wellcome Trust; National Alliance for Research on Schizophrenia and Depression; Brain and Behavior Research Foundation","keywords":"Public relations; Online community; Political science; Social media; Perspective (graphical); Sociology; Computer science; Law","score_opus":0.17141355045694476,"score_gpt":0.4341449156616235,"score_spread":0.26273136520467877,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3154878857","genre_codex":"methods","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13131724,0.01078788,0.429324,0.12786995,0.004641501,0.0029181833,0.0001588719,0.0016066027,0.29137585],"genre_scores_gemma":[0.7523147,0.0047357157,0.20004773,0.0116235465,0.0012394647,0.0026495834,0.0001766976,0.00074750715,0.02646503],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.8587825,0.11985822,0.0027686732,0.004113543,0.010329593,0.0041474826],"domain_scores_gemma":[0.8290234,0.11481961,0.0096161505,0.015842594,0.011820478,0.018877756],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.09421289,0.0009063191,0.00062257785,0.003306515,0.012137916,0.028745035,0.0055877995,0.0050671063,0.013647539],"category_scores_gemma":[0.12543897,0.00081856555,0.0009247181,0.0019828547,0.024980312,0.019787692,0.021956066,0.0067096413,0.0036255608],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028060994,0.0008049863,0.0072619594,0.0031187714,0.00007774086,0.000727987,0.3492672,0.0019158467,0.0046362435,0.33076754,0.0145246135,0.2866165],"study_design_scores_gemma":[0.00012373934,0.00045202507,0.0028727544,0.0031212287,0.00007221453,0.000592091,0.18266004,0.0018552789,0.0032497128,0.23967458,0.5651872,0.00013915877],"about_ca_topic_score_codex":0.0014951043,"about_ca_topic_score_gemma":0.002166135,"teacher_disagreement_score":0.9944122,"about_ca_system_score_codex":0.0063235867,"about_ca_system_score_gemma":0.014174266,"threshold_uncertainty_score":0.49825132},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3159243089","doi":"10.1136/bmjopen-2020-041530","title":"Living labs for patient engagement and knowledge exchange: an exploratory sequential mixed methods study to develop a living lab in paediatric rehabilitation","year":2021,"lang":"en","type":"article","venue":"BMJ Open","topic":"Innovative Approaches in Technology and Social Development","field":"Business, Management and Accounting","cited_by":16,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Manitoba","funders":"Canadian Institutes of Health Research","keywords":"Thematic analysis; Knowledge translation; Medicine; Medical education; Data collection; Rehabilitation; Qualitative research; Health care; Qualitative property; Independent living; Nursing; Knowledge management; Gerontology; Computer science","score_opus":0.11864072417143366,"score_gpt":0.40954946023727923,"score_spread":0.29090873606584555,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3159243089","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95710355,0.00048467412,0.009694961,0.0024881628,0.00007494229,0.025691498,0.00045811568,0.000067984874,0.0039360365],"genre_scores_gemma":[0.88195145,0.00095812033,0.048893206,0.0027473199,0.000061671315,0.06253575,0.00035023113,0.000057210098,0.0024449714],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.95922536,0.03433235,0.0013560379,0.001097212,0.0018230536,0.0021659385],"domain_scores_gemma":[0.9666939,0.021287138,0.0027834713,0.0015327134,0.0037642499,0.003938468],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.050837435,0.00064280495,0.0011569699,0.0018261106,0.004863308,0.0041589406,0.0028387317,0.002004863,0.0044774804],"category_scores_gemma":[0.041320153,0.00093889976,0.0011364456,0.0014447592,0.0028701923,0.0049451203,0.007594461,0.0028858506,0.0008655764],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00084080105,0.010094468,0.02590757,0.0044730725,0.00009205161,0.0023023526,0.82539886,0.0004716196,0.0022620142,0.0031427292,0.005053594,0.11996086],"study_design_scores_gemma":[0.0007048263,0.009717749,0.024393355,0.002651878,0.00010199679,0.00070199685,0.9281001,0.0013011872,0.0019822938,0.0030577423,0.02714639,0.00014043404],"about_ca_topic_score_codex":0.001800631,"about_ca_topic_score_gemma":0.0045302184,"teacher_disagreement_score":0.050837435,"about_ca_system_score_codex":0.0043942323,"about_ca_system_score_gemma":0.013455168,"threshold_uncertainty_score":0.2688573},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch"],"domain":"methods","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3194760139","doi":"10.1038/s41597-021-00941-8","title":"Open-access quantitative MRI data of the spinal cord and reproducibility across participants, sites and manufacturers","year":2021,"lang":"en","type":"article","venue":"Scientific Data","topic":"Advanced Neuroimaging Techniques and Applications","field":"Medicine","cited_by":77,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Philips (Canada); McGill University; Centre Hospitalier Universitaire de Sherbrooke; International Collaboration On Repair Discoveries; University of British Columbia; Université de Montréal; Université de Sherbrooke; Centre Hospitalier Universitaire Sainte-Justine; Polytechnique Montréal; Montreal Neurological Institute and Hospital; Mila - Quebec Artificial Intelligence Institute","funders":"National Institute of Neurological Disorders and Stroke; Staatssekretariat für Bildung, Forschung und Innovation; Economic and Social Research Council; Engineering and Physical Sciences Research Council; Natural Sciences and Engineering Research Council of Canada; National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; CRIS Cancer Foundation; Regione Puglia; University College London Hospitals NHS Foundation Trust; Ministero dell’Istruzione, dell’Università e della Ricerca; Ministero della Salute; Concordia University; Agentura Pro Zdravotnický Výzkum České Republiky; National Institutes of Health; Rosetrees Trust; European Commission; Multiple Sclerosis Society; Bundesministerium für Bildung und Forschung; National Imaging Facility; National Institute for Health and Care Research; University of Pennsylvania; SpinalCure Australia; University of Minnesota; National Science Foundation; Compute Canada; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; Polytechnique Montréal; Wellcome Trust; Institut de Valorisation des Données; AstraZeneca; Craig H. Neilsen Foundation; McGill University; Canada First Research Excellence Fund; Max-Planck-Gesellschaft","keywords":"Protocol (science); Reproducibility; Computer science; Spinal cord; Documentation; Data mining; Medicine; Statistics; Mathematics; Pathology","score_opus":0.6245720496163727,"score_gpt":0.5767339061992035,"score_spread":0.04783814341716919,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3194760139","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07016648,0.0068310634,0.39781493,0.005560696,0.0033227778,0.0073589515,0.45325306,0.018452251,0.037239723],"genre_scores_gemma":[0.20424879,0.003007717,0.32081735,0.0035610315,0.0011805277,0.03874187,0.39438957,0.016666891,0.017386245],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9727237,0.008445717,0.0047874767,0.0047021736,0.00871706,0.0006239324],"domain_scores_gemma":[0.8460825,0.05748343,0.012050132,0.046009578,0.036949597,0.0014247403],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.045483172,0.0017158035,0.0014566878,0.004205148,0.0014444222,0.0041127563,0.0033708054,0.002448414,0.045224566],"category_scores_gemma":[0.1963131,0.0012397546,0.00200293,0.004819531,0.0018156222,0.0030422937,0.004322847,0.0017663267,0.018530851],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0044510895,0.0006130082,0.060122952,0.012328694,0.0034578694,0.0008517175,0.0034683074,0.003684344,0.017278546,0.013361889,0.47890958,0.401472],"study_design_scores_gemma":[0.0017068062,0.0010428225,0.17642672,0.005143375,0.002153533,0.0038703308,0.0011578329,0.0054794275,0.020277044,0.043906722,0.73797184,0.00086354883],"about_ca_topic_score_codex":0.0030438055,"about_ca_topic_score_gemma":0.006465805,"teacher_disagreement_score":0.9966292,"about_ca_system_score_codex":0.0009324423,"about_ca_system_score_gemma":0.0039415765,"threshold_uncertainty_score":0.24054086},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3200026842","doi":"10.18438/eblip29913","title":"Do Institutional Repository Deposit Guidelines Deter Data Discovery?","year":2021,"lang":"en","type":"article","venue":"Evidence Based Library and Information Practice","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Metadata; Computer science; Data element; Sample (material); Meta Data Services; Documentation; Data discovery; Data science; World Wide Web; Metadata repository; Information retrieval; Database","score_opus":0.466706315802476,"score_gpt":0.5217880096438329,"score_spread":0.05508169384135686,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3200026842","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6964903,0.006069012,0.046586182,0.09719183,0.0007623976,0.0034589884,0.0025538474,0.00088751694,0.14599997],"genre_scores_gemma":[0.95515233,0.0013157312,0.028999403,0.0071872124,0.00037519605,0.0015077131,0.0008646491,0.00016485491,0.004433038],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.55218863,0.22803077,0.08185951,0.012809508,0.115809806,0.009301765],"domain_scores_gemma":[0.14220224,0.52923447,0.18569642,0.059465785,0.076760255,0.0066407807],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.32319552,0.00039963974,0.0012219895,0.011227593,0.0065604444,0.021200832,0.0058391746,0.0039952067,0.0070192763],"category_scores_gemma":[0.6684824,0.0015500026,0.0009208925,0.01910802,0.010412205,0.026434958,0.014709224,0.0026566805,0.0017422364],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005997733,0.0005589436,0.5370034,0.0038394353,0.00033004413,0.0006794054,0.03417989,0.0013243997,0.00143421,0.16920727,0.025203263,0.22563994],"study_design_scores_gemma":[0.00048093047,0.0013404267,0.40571028,0.009018154,0.00055011467,0.002367227,0.14988425,0.007062013,0.01145194,0.12215833,0.28945702,0.00051931496],"about_ca_topic_score_codex":0.008582888,"about_ca_topic_score_gemma":0.011524326,"teacher_disagreement_score":0.97879916,"about_ca_system_score_codex":0.013816627,"about_ca_system_score_gemma":0.035701197,"threshold_uncertainty_score":0.8346204},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"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":"W3201074415","doi":"10.21470/1678-9741-2021-0289","title":"Open Access and Article Processing Charges in Cardiology and Cardiac Surgery Journals: a CrossSectional Analysis","year":2021,"lang":"en","type":"article","venue":"Brazilian Journal of Cardiovascular Surgery","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Medicine; Interquartile range; Cardiology; Internal medicine; Cardiac surgery","score_opus":0.13168469426835763,"score_gpt":0.4190559080871945,"score_spread":0.2873712138188369,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3201074415","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99310726,0.0015418663,0.00026950304,0.00022971124,0.000020941741,0.00012263995,0.0035243286,0.000012591128,0.0011712547],"genre_scores_gemma":[0.9967828,0.00060012465,0.00030760976,0.00012260764,0.000057153295,0.00016090869,0.0016956758,0.000014147013,0.0002590917],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9862113,0.002512161,0.0039140014,0.001778029,0.004666945,0.0009174908],"domain_scores_gemma":[0.8943488,0.032620363,0.05758363,0.0029276165,0.007997814,0.0045217588],"candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.0101729175,0.00033523282,0.0009172261,0.0069079553,0.000826062,0.0024203218,0.0011030555,0.000672757,0.004946129],"category_scores_gemma":[0.04262155,0.0005459717,0.0015881109,0.011171767,0.00093222194,0.0040122857,0.0017391772,0.0012962872,0.0011172625],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000094577765,0.000046918307,0.9977266,0.000103650316,0.00013308482,0.000031291045,0.00022246633,0.000018406443,0.000034306137,0.00005094736,0.0002709028,0.0012668245],"study_design_scores_gemma":[0.000017024566,0.00017420342,0.9971706,0.000103633356,0.000120674886,0.00031049934,0.0011527706,0.00013910765,0.000057922258,0.00006804601,0.00067302043,0.000012345128],"about_ca_topic_score_codex":0.0024114144,"about_ca_topic_score_gemma":0.0027017738,"teacher_disagreement_score":0.99889696,"about_ca_system_score_codex":0.0013229985,"about_ca_system_score_gemma":0.0015641251,"threshold_uncertainty_score":0.053800166},"labels":[{"model":"gemma","categories":["metaresearch","scholarly_communication"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W3207442765","doi":"10.5281/zenodo.4571673","title":"Modèle de plan de gestion des données : La recherche basée sur les arts","year":2021,"lang":"fr","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Cultural Insights and Digital Impacts","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Humanities; Political science; Art","score_opus":0.8796125412529113,"score_gpt":0.3294194173326425,"score_spread":0.5501931239202689,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3207442765","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0035353112,0.000719011,0.9720026,0.0028944353,0.000114451585,0.0003056572,0.0005356297,0.0008501318,0.019042674],"genre_scores_gemma":[0.09010301,0.0018729856,0.88378006,0.0006352483,0.0000929259,0.0008260506,0.0014171487,0.00047218183,0.020800421],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.98540884,0.0055079665,0.0011079338,0.0025317552,0.004932556,0.0005108989],"domain_scores_gemma":[0.9846858,0.008239861,0.0006817632,0.0038048187,0.002119628,0.0004682121],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.011345229,0.0013358631,0.00086474133,0.002954111,0.0019617127,0.012070263,0.002958955,0.0033687411,0.015036243],"category_scores_gemma":[0.02228722,0.0012963809,0.0034191734,0.0030067533,0.006529744,0.012389775,0.005526874,0.0042838487,0.0041908817],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018317616,0.00010117685,0.0014908416,0.0009944893,0.000096165946,0.00029677778,0.004448582,0.027532734,0.0036532988,0.84529185,0.005547044,0.11036383],"study_design_scores_gemma":[0.00016216964,0.00026199786,0.0014936378,0.001257118,0.00023051286,0.00079724827,0.0026759566,0.15587735,0.01294021,0.4647662,0.3593707,0.00016684663],"about_ca_topic_score_codex":0.024007343,"about_ca_topic_score_gemma":0.015421677,"teacher_disagreement_score":0.99704105,"about_ca_system_score_codex":0.0045800866,"about_ca_system_score_gemma":0.010008873,"threshold_uncertainty_score":0},"labels":[{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"grok","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"opus","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W3210376598","doi":"10.1097/mol.0000000000000797","title":"Preprint servers in lipidology: current status and future role.","year":2022,"lang":"en","type":"article","venue":"PubMed","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University; Western University","funders":"","keywords":"Coronavirus disease 2019 (COVID-19); Preprint; 2019-20 coronavirus outbreak; Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); Publishing; Library science; Medicine; World Wide Web; Disease; Computer science; Infectious disease (medical specialty); Internal medicine; Outbreak; Virology; Political science","score_opus":0.07156169814552168,"score_gpt":0.34449311701390817,"score_spread":0.27293141886838646,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3210376598","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0060657775,0.39648014,0.039306294,0.30886513,0.15276676,0.0023949114,0.016448224,0.025620623,0.052052084],"genre_scores_gemma":[0.045895275,0.48198712,0.09660056,0.047880165,0.14617105,0.0025457838,0.037767466,0.015105464,0.12604712],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.91616327,0.024396094,0.019030929,0.0062413444,0.030957587,0.003210776],"domain_scores_gemma":[0.709759,0.07335939,0.038016886,0.026961565,0.11900931,0.032893885],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.101526424,0.0024917794,0.006054118,0.009913426,0.0049149627,0.03778319,0.0057750116,0.007997421,0.08569387],"category_scores_gemma":[0.16377465,0.0022770273,0.0028504368,0.0129761025,0.006645433,0.026168436,0.009059163,0.010452644,0.18221672],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000560093,0.00011822723,0.0028409737,0.013107871,0.00019972863,0.0003990915,0.0009802036,0.000104792474,0.0024475802,0.009224568,0.5518818,0.41813508],"study_design_scores_gemma":[0.00008302589,0.000111541915,0.0022294326,0.0033891671,0.00009958813,0.00079238816,0.0006227556,0.0002891947,0.00091038033,0.0057359994,0.98564863,0.00008802041],"about_ca_topic_score_codex":0.0011728414,"about_ca_topic_score_gemma":0.0009917491,"teacher_disagreement_score":0.99422497,"about_ca_system_score_codex":0.004216599,"about_ca_system_score_gemma":0.0244465,"threshold_uncertainty_score":0.5369295},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"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":"W3214426301","doi":"10.5334/joad.80","title":"Geospatial and Image Data from the “When Computers Dream of Charcoal: Using Deep Learning, Open Tools and Open Data to Identify Relict Charcoal Hearths in and Around State Game Lands in Pennsylvania” Paper","year":2021,"lang":"en","type":"article","venue":"Journal of Open Archaeology Data","topic":"HIV, Drug Use, Sexual Risk","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Charcoal; Geospatial analysis; Dream; Hearth; Computer science; State (computer science); Archaeology; Remote sensing; Geography; Psychology; Chemistry","score_opus":0.23095003284038812,"score_gpt":0.4596432283936033,"score_spread":0.22869319555321516,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W3214426301","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.89922863,0.00080735993,0.011002804,0.008688778,0.0005010058,0.00012805307,0.05103776,0.00093175157,0.027673824],"genre_scores_gemma":[0.9119078,0.00050669746,0.032352086,0.00081293256,0.00019430128,0.000102327496,0.042084668,0.00019266643,0.0118464995],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99970955,0.00006167841,0.000012510419,0.0000638445,0.00010329151,0.00004903926],"domain_scores_gemma":[0.9989035,0.00037284932,0.00013165936,0.00011155587,0.0003148179,0.00016555679],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00037767878,0.00023902344,0.00011527758,0.0010324705,0.000442408,0.0008407392,0.00036882638,0.00072708906,0.0035603954],"category_scores_gemma":[0.002264628,0.00013472003,0.00019918274,0.0009884296,0.00048500602,0.00091002614,0.0013159047,0.0007585038,0.0012456409],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0008000822,0.0005623422,0.37539178,0.00035743284,0.00026972778,0.0013055485,0.0049408707,0.009207898,0.009417357,0.0044321283,0.20506768,0.38824716],"study_design_scores_gemma":[0.000057313937,0.00023844442,0.5719606,0.00035457776,0.00015254025,0.0007922236,0.024957635,0.038297553,0.016804848,0.0128898155,0.33333275,0.00016161476],"about_ca_topic_score_codex":0.023207435,"about_ca_topic_score_gemma":0.08903019,"teacher_disagreement_score":0.023207435,"about_ca_system_score_codex":0.0004371852,"about_ca_system_score_gemma":0.00064749544,"threshold_uncertainty_score":0.046144664},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W350340533","doi":"","title":"Publishing Trends and Open Access Journals at York University","year":2011,"lang":"en","type":"article","venue":"YorkSpace (York University)","topic":"Publishing and Scholarly Communication","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Publishing; Library science; Media studies; Political science; World Wide Web; Computer science; Sociology; Law","score_opus":0.24217321604234088,"score_gpt":0.282328559418204,"score_spread":0.04015534337586313,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W350340533","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8317062,0.024646986,0.00028570424,0.028678443,0.00075422396,0.00006020703,0.012220573,0.00019908465,0.10144855],"genre_scores_gemma":[0.90490794,0.020748995,0.00059000816,0.0013776665,0.0010651194,0.000068053225,0.0064222915,0.00019710152,0.06462276],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99577785,0.00036612773,0.00050806045,0.0003089844,0.0024386854,0.0006002903],"domain_scores_gemma":[0.942845,0.010719717,0.015564962,0.001132954,0.017893314,0.011844045],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0023610385,0.00013839614,0.00036589705,0.0123439105,0.0033319092,0.007996913,0.0006757701,0.0008022873,0.037922207],"category_scores_gemma":[0.027928095,0.00023588535,0.00022595559,0.03406848,0.0012228133,0.0070101027,0.0015211622,0.0015149594,0.004044924],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028601874,0.0002219143,0.7435596,0.00052118365,0.00006224141,0.00044925587,0.015800206,0.00014128281,0.00090567843,0.018585065,0.09261601,0.12685147],"study_design_scores_gemma":[0.000015258069,0.000056535162,0.91512644,0.00025681168,0.000019859473,0.00037954812,0.011070073,0.00011333834,0.00033618245,0.0010018565,0.07157596,0.00004810792],"about_ca_topic_score_codex":0.1449947,"about_ca_topic_score_gemma":0.25885987,"teacher_disagreement_score":0.99763894,"about_ca_system_score_codex":0.005793895,"about_ca_system_score_gemma":0.006476848,"threshold_uncertainty_score":0.28830147},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","scholarly_communication"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4200389440","doi":"10.1101/2021.12.06.21267383","title":"Data and code availability statements in systematic reviews of interventions were often missing or inaccurate: a content analysis","year":2021,"lang":"en","type":"preprint","venue":"medRxiv","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital; University of Ottawa","funders":"National Health and Medical Research Council; Medical Research Council; Australian Government; University of Ottawa","keywords":"Computer science; Citation; Systematic review; Scopus; Code review; Information retrieval; Download; Sample (material); Raw data; Data file; Data science; World Wide Web; Software; MEDLINE; Software quality; Database; Software development","score_opus":0.9316209954698754,"score_gpt":0.6191030025453813,"score_spread":0.3125179929244941,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4200389440","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":"methods","domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3386354,0.12310467,0.31075028,0.037122022,0.0050413786,0.12463233,0.048527606,0.0016300756,0.010556188],"genre_scores_gemma":[0.575737,0.019911153,0.22769687,0.009085631,0.0010675656,0.15432848,0.010475524,0.00053795247,0.0011598417],"study_design_codex":"systematic_review","study_design_gemma":"observational","domain_scores_codex":[0.172735,0.4470631,0.26348814,0.019110007,0.0949736,0.0026301085],"domain_scores_gemma":[0.041958436,0.7423329,0.15361173,0.026885338,0.034598805,0.0006127748],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.5800971,0.0022863473,0.0060083494,0.042672165,0.0034427897,0.006289258,0.0045933165,0.0042124884,0.004515613],"category_scores_gemma":[0.8790527,0.0035420542,0.009471383,0.044255853,0.009894377,0.013129931,0.013682121,0.0041378294,0.0007749762],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0069344942,0.00050452986,0.1506928,0.42858058,0.022233844,0.0011993267,0.06268543,0.0031759802,0.003412715,0.01974441,0.027114382,0.2737215],"study_design_scores_gemma":[0.0045174975,0.0032631494,0.18726191,0.51145023,0.039867625,0.0038170882,0.03283906,0.016378256,0.010998221,0.04890601,0.1389901,0.0017108144],"about_ca_topic_score_codex":0.002929934,"about_ca_topic_score_gemma":0.0033850393,"teacher_disagreement_score":0.41990292,"about_ca_system_score_codex":0.011763621,"about_ca_system_score_gemma":0.021957304,"threshold_uncertainty_score":0.5178151},"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","open_science"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4205376244","doi":"10.31219/osf.io/jm3hf","title":"An infographic of data sharing in North America","year":2022,"lang":"en","type":"preprint","venue":"","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université Laval","funders":"Canadian Institutes of Health Research","keywords":"Infographic; Data sharing; Political science; Geography; Regional science; Computer science; Data mining","score_opus":0.2607468458245488,"score_gpt":0.430734309988772,"score_spread":0.16998746416422317,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4205376244","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.034378055,0.038607817,0.014497731,0.15431547,0.0086999405,0.00031554006,0.047358673,0.0024814084,0.6993454],"genre_scores_gemma":[0.37937093,0.07826907,0.05247276,0.053788055,0.009411026,0.0011189649,0.04445524,0.0021379811,0.37897593],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976077,0.0005879853,0.00015929474,0.0002264514,0.001079575,0.0003389684],"domain_scores_gemma":[0.9851854,0.008287353,0.0015044251,0.0012908395,0.002630895,0.0011011618],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.0027652662,0.0002744688,0.00024889485,0.009212295,0.003614493,0.0054935347,0.00070632884,0.0017057355,0.038583715],"category_scores_gemma":[0.009513485,0.00046837627,0.0002509345,0.02482005,0.0018157656,0.0071503716,0.0021843475,0.0021452683,0.0033245445],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007711914,0.00004703721,0.0026468253,0.0005986228,0.000008724932,0.00016406347,0.0024756999,0.00065314205,0.00068118674,0.117357284,0.7742173,0.101073],"study_design_scores_gemma":[0.0000039439033,0.000011972212,0.0054318225,0.00037097468,0.0000039169827,0.00006277047,0.00095510075,0.000119093005,0.00019597384,0.0044783484,0.98835224,0.000013925687],"about_ca_topic_score_codex":0.09510105,"about_ca_topic_score_gemma":0.11144313,"teacher_disagreement_score":0.9992937,"about_ca_system_score_codex":0.005946095,"about_ca_system_score_gemma":0.009667901,"threshold_uncertainty_score":0.18909502},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4210255074","doi":"10.3138/jsp.53.2.02","title":"Strategies to Increase the Number of Open Access Journals: The Cases of Elsevier and Springer Nature","year":2022,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Competition (biology); Open access journal; Library science; Political science; Business; Computer science; Scopus; Law","score_opus":0.5532761290723078,"score_gpt":0.6122283546242888,"score_spread":0.058952225551980986,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210255074","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6561846,0.004914975,0.005359919,0.07453228,0.00037822057,0.0002749987,0.00015127612,0.00031547344,0.25788826],"genre_scores_gemma":[0.97761565,0.0014275911,0.0031147366,0.003758652,0.00029808067,0.00011213816,0.000057382105,0.00007572849,0.013539925],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.9761788,0.0059288186,0.0010746628,0.0023184242,0.009811495,0.0046878033],"domain_scores_gemma":[0.8646511,0.07723644,0.015236103,0.005599111,0.011866143,0.025411177],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.028757203,0.00058197463,0.00056631624,0.007081808,0.008480751,0.02487966,0.0025230125,0.0061947773,0.008795523],"category_scores_gemma":[0.079424724,0.00065563415,0.00087800395,0.009106689,0.008878733,0.018962586,0.011020059,0.004513523,0.001176539],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00042144235,0.0006453288,0.12279347,0.00047394726,0.00013501236,0.007299469,0.050249916,0.0010362589,0.0037556035,0.61547184,0.033397,0.1643207],"study_design_scores_gemma":[0.00041295728,0.0006948063,0.11103347,0.0008132581,0.0002097497,0.0064690174,0.11666451,0.0044223163,0.0080697965,0.14623797,0.6044602,0.0005119788],"about_ca_topic_score_codex":0.0101364525,"about_ca_topic_score_gemma":0.0134495795,"teacher_disagreement_score":0.997477,"about_ca_system_score_codex":0.013009777,"about_ca_system_score_gemma":0.017849782,"threshold_uncertainty_score":0.15208441},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4210639134","doi":"10.1001/jamaoncol.2021.7639","title":"Development of an Open Database of Genes Included in Hereditary Cancer Genetic Testing Panels Available From Major Sources in the US","year":2022,"lang":"en","type":"article","venue":"JAMA Oncology","topic":"Genomics and Rare Diseases","field":"Biochemistry, Genetics and Molecular Biology","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"National Cancer Institute; National Institute of Mental Health; Canadian Institutes of Health Research; National Institutes of Health; Wellcome Trust","keywords":"Medicine; Hereditary Cancer; Genetic testing; Gene; Cancer; Database; Genetics; Bioinformatics; Computational biology; Internal medicine; Biology; Computer science","score_opus":0.042722726313453314,"score_gpt":0.3039056119699771,"score_spread":0.2611828856565238,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4210639134","genre_codex":"dataset","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16211969,0.0032093355,0.16389996,0.022359949,0.0020407366,0.010511632,0.5726009,0.0239702,0.039287698],"genre_scores_gemma":[0.12345855,0.0011681501,0.2965824,0.00507044,0.0004772215,0.0038189683,0.56198466,0.0020802496,0.005359437],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9774984,0.0058365637,0.0042631673,0.0033017213,0.008025218,0.0010749723],"domain_scores_gemma":[0.8452602,0.052599356,0.013155129,0.02787094,0.052609626,0.008504779],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.03247094,0.00059421174,0.0011595732,0.008752207,0.0014269352,0.004905829,0.003184967,0.001068458,0.009692283],"category_scores_gemma":[0.08571642,0.00088172965,0.0011660069,0.00882171,0.0004932852,0.0038059202,0.0043666963,0.0026261515,0.0034377493],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031286522,0.002651136,0.13945934,0.0014820808,0.000569115,0.00053766044,0.0012927469,0.0031406907,0.011563331,0.008737313,0.3375206,0.4899175],"study_design_scores_gemma":[0.0026158865,0.0011878056,0.25799322,0.0016992622,0.00094970636,0.00090654066,0.0017477054,0.014061732,0.031663526,0.0065965857,0.6801887,0.0003893979],"about_ca_topic_score_codex":0.022642383,"about_ca_topic_score_gemma":0.020268524,"teacher_disagreement_score":0.996815,"about_ca_system_score_codex":0.0028026775,"about_ca_system_score_gemma":0.01287033,"threshold_uncertainty_score":0.1717248},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4224451671","doi":"10.4018/978-1-7998-9702-6.ch009","title":"Building, Sustaining, and Growing Multidisciplinary, Multi-Departmental Partnerships to Teach Open Science Tools","year":2022,"lang":"en","type":"book-chapter","venue":"Advances in library and information science (ALIS) book series","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Thompson Rivers University","funders":"","keywords":"Multidisciplinary approach; Medical education; Engineering management; Service (business); Best practice; Engineering; Knowledge management; Computer science; Political science; Business; Medicine; Marketing","score_opus":0.08664473437372724,"score_gpt":0.37663088145493323,"score_spread":0.28998614708120596,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4224451671","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.055297676,0.019948337,0.14798407,0.08343123,0.0074931877,0.001208561,0.00027432895,0.004437078,0.67992556],"genre_scores_gemma":[0.1389689,0.012073954,0.21590161,0.012791559,0.0012668289,0.00094423804,0.00056479545,0.0025477435,0.6149404],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9952231,0.0013343145,0.0001549062,0.0006845336,0.0020982856,0.0005049163],"domain_scores_gemma":[0.9896735,0.001977784,0.00030337562,0.00064538035,0.0012248306,0.0061751525],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0065391506,0.00062886876,0.000440189,0.0018345113,0.004304292,0.012962361,0.0019546016,0.0019812474,0.023535611],"category_scores_gemma":[0.006530517,0.00063256087,0.00054180785,0.001973905,0.0019113999,0.010507503,0.013336349,0.004851078,0.0124777015],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000027023509,0.00055226515,0.0018991324,0.00041209647,0.000021578846,0.0005398556,0.02371863,0.00050395046,0.0032148066,0.07006852,0.33676568,0.56227654],"study_design_scores_gemma":[0.000011294732,0.00008480943,0.0010011102,0.00024353463,0.0000050372887,0.00043801006,0.010505452,0.0004155363,0.00056156283,0.013355344,0.973361,0.000017305816],"about_ca_topic_score_codex":0.0011887454,"about_ca_topic_score_gemma":0.007527594,"teacher_disagreement_score":0.9980454,"about_ca_system_score_codex":0.002921206,"about_ca_system_score_gemma":0.01007122,"threshold_uncertainty_score":0.07873446},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"design_other","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4226504205","doi":"10.31222/osf.io/9e526","title":"Teaching Open and Reproducible Scholarship: A Critical Review of the Evidence Base for Current Pedagogical Methods and their Outcomes","year":2022,"lang":"en","type":"review","venue":"","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":18,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Economic and Social Research Council; Queen's University; Department for the Economy; UK Research and Innovation; Queen's University Belfast; John Templeton Foundation","keywords":"Scholarship; Credibility; Promotion (chess); Psychology; Engineering ethics; Medical education; Mathematics education; Political science; Medicine; Engineering","score_opus":0.9813179241858248,"score_gpt":0.7638947768303257,"score_spread":0.21742314735549906,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4226504205","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0003624354,0.9929491,0.001234935,0.0034410038,0.0008113469,0.0007401604,0.0001123758,0.000015874051,0.00033282294],"genre_scores_gemma":[0.011150051,0.97605103,0.0069693406,0.0028056193,0.00048182713,0.0022011625,0.00017755391,0.000027880707,0.0001355239],"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","domain_scores_codex":[0.8519196,0.074912846,0.044839922,0.005411655,0.021638107,0.0012778335],"domain_scores_gemma":[0.47158554,0.43383476,0.032306947,0.010836282,0.04971136,0.0017251023],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.17379694,0.0025743123,0.009846304,0.026292924,0.0025893964,0.0095425425,0.005237525,0.0052004154,0.003997868],"category_scores_gemma":[0.4624068,0.0025318132,0.010172951,0.018068131,0.0052754628,0.010962442,0.0051765367,0.0056576175,0.0008500694],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003574781,0.000055753353,0.0008622337,0.793069,0.0063243676,0.00017160586,0.0017867131,0.0002705737,0.00033475147,0.004107416,0.0077005033,0.18495953],"study_design_scores_gemma":[0.00008221642,0.00010411841,0.0006593269,0.9490222,0.010037335,0.00013938364,0.0006348103,0.00011183662,0.0002210622,0.002729534,0.03620868,0.00004954265],"about_ca_topic_score_codex":0.0074227774,"about_ca_topic_score_gemma":0.020634267,"teacher_disagreement_score":0.9947625,"about_ca_system_score_codex":0.014588494,"about_ca_system_score_gemma":0.04858742,"threshold_uncertainty_score":0.91913706},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4229589109","doi":"10.31222/osf.io/d2y43","title":"Transparent, Open, and Reproducible Prevention Science","year":2020,"lang":"en","type":"preprint","venue":"","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Victoria","funders":"","keywords":"Transparency (behavior); Openness to experience; Open science; Best practice; Engineering ethics; Citizen science; Public relations; Open data; Political science; Computer science; Knowledge management; Psychology; Engineering; World Wide Web; Computer security","score_opus":0.9469422615351218,"score_gpt":0.7878863451378618,"score_spread":0.15905591639726002,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4229589109","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023043673,0.0076372847,0.5619343,0.24193487,0.0043305545,0.0023413668,0.001311609,0.0011571912,0.15630919],"genre_scores_gemma":[0.5311258,0.0062910775,0.41952562,0.024600951,0.0033672429,0.003551862,0.000973032,0.00068241835,0.009882051],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.58281744,0.31431624,0.016568404,0.018736036,0.062488783,0.0050731073],"domain_scores_gemma":[0.2731803,0.47885457,0.03440739,0.16384088,0.042853765,0.006863089],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.33465484,0.0008582196,0.0020853519,0.0055873836,0.00858985,0.025721665,0.003787019,0.008239651,0.00631242],"category_scores_gemma":[0.5280445,0.0018281735,0.0025105353,0.0057426565,0.04792339,0.025518928,0.022212328,0.013830038,0.002372193],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000920072,0.00016848094,0.0041069835,0.0013742917,0.00014308898,0.00013474206,0.017716419,0.0016222955,0.00061007205,0.89103603,0.011899143,0.07109639],"study_design_scores_gemma":[0.00007990494,0.000064712316,0.0022319895,0.0018616533,0.00006578431,0.000117566175,0.0029430757,0.0012193325,0.0011606087,0.91009057,0.080088355,0.000076428405],"about_ca_topic_score_codex":0.0032042498,"about_ca_topic_score_gemma":0.0029350468,"teacher_disagreement_score":0.99621296,"about_ca_system_score_codex":0.0092444215,"about_ca_system_score_gemma":0.052206375,"threshold_uncertainty_score":0},"labels":[{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"grok","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"opus","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4232529136","doi":"10.1139/cjc-2015-0201","title":"Appuyer le libre accès et offrir un plus grand choix aux auteurs","year":2015,"lang":"fr","type":"article","venue":"Canadian Journal of Chemistry","topic":"Health, Medicine and Society","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Chemistry; Humanities; Philosophy","score_opus":0.06687742783372794,"score_gpt":0.36168792800451727,"score_spread":0.29481050017078936,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4232529136","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013931503,0.029580688,0.025785975,0.5369838,0.14910084,0.00026879236,0.0016293717,0.009493162,0.23322599],"genre_scores_gemma":[0.054441772,0.011062366,0.03281275,0.07057988,0.030334365,0.00021979444,0.0012243901,0.0041431524,0.79518163],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.97758377,0.0063794805,0.0020703666,0.0019753864,0.009898341,0.002092701],"domain_scores_gemma":[0.8917245,0.016150437,0.0042754626,0.009087044,0.057861716,0.020900778],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.012055806,0.0019641109,0.0022244249,0.0046805292,0.00890416,0.020563815,0.0023090162,0.008740508,0.17491773],"category_scores_gemma":[0.08887699,0.0011625787,0.0018252765,0.0030349323,0.003752915,0.01714867,0.0095242625,0.011611423,0.11865169],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000520924,0.00019801971,0.0030048832,0.00067880086,0.000079554186,0.0035201202,0.004704852,0.0002599172,0.005213693,0.029103246,0.7876757,0.16504037],"study_design_scores_gemma":[0.000028525028,0.00002913947,0.0006709785,0.00023194346,0.000020757587,0.0016752784,0.0028162391,0.0001910032,0.0007652543,0.0028278697,0.9906765,0.00006637039],"about_ca_topic_score_codex":0.022615194,"about_ca_topic_score_gemma":0.022194669,"teacher_disagreement_score":0.997691,"about_ca_system_score_codex":0.005430352,"about_ca_system_score_gemma":0.015004886,"threshold_uncertainty_score":0.5851581},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4236731814","doi":"10.1139/cjes-2015-0075","title":"Appuyer le libre accès et offrir un plus grand choix aux auteurs","year":2015,"lang":"fr","type":"article","venue":"Canadian Journal of Earth Sciences","topic":"Health, Medicine and Society","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Humanities; Geology; Philosophy","score_opus":0.13608875446860333,"score_gpt":0.39643653520902067,"score_spread":0.2603477807404173,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4236731814","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016753068,0.025778417,0.026660427,0.5762948,0.10826455,0.00031637817,0.0014568851,0.007454544,0.23702097],"genre_scores_gemma":[0.07071838,0.011048974,0.041997682,0.07526806,0.025538715,0.00028078022,0.0013010284,0.003925996,0.76992035],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.97375417,0.007855632,0.002449952,0.0020545032,0.011462513,0.0024231442],"domain_scores_gemma":[0.8848812,0.017662719,0.004509595,0.008880449,0.06269653,0.021369496],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.013634264,0.0018328496,0.0019311135,0.005190092,0.010160058,0.021852404,0.0023721266,0.009684689,0.13156587],"category_scores_gemma":[0.094555296,0.0010591571,0.0017409174,0.0034678879,0.0039627426,0.019614622,0.010118141,0.011569105,0.082471244],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00046007358,0.00023005292,0.00450207,0.0009231717,0.000097285105,0.0042397487,0.0085905325,0.0003461519,0.005792853,0.040333502,0.73150396,0.20298053],"study_design_scores_gemma":[0.000023648869,0.00002712344,0.00077753846,0.00027484845,0.000021049313,0.0014661368,0.0045757806,0.00020067865,0.0006748443,0.0029105213,0.9889774,0.000070574],"about_ca_topic_score_codex":0.035495415,"about_ca_topic_score_gemma":0.034953333,"teacher_disagreement_score":0.99762785,"about_ca_system_score_codex":0.0059772753,"about_ca_system_score_gemma":0.016812062,"threshold_uncertainty_score":0.44013166},"labels":[{"model":"gemma","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4237547470","doi":"10.21203/rs.3.rs-185385/v1","title":"Incorporating and Evaluating Citizen Engagement in Health Research: a Scoping Review Protocol","year":2021,"lang":"en","type":"review","venue":"Research Square","topic":"Mental Health and Patient Involvement","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Calgary General Hospital; Dalhousie University; University of Calgary","funders":"","keywords":"Protocol (science); Political science; Sociology; Medicine; Alternative medicine","score_opus":0.9242265114782438,"score_gpt":0.7574416992370914,"score_spread":0.16678481224115238,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4237547470","genre_codex":"protocol","genre_gemma":"protocol","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"protocol","genre_consensus":"protocol","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0005493269,0.0016887402,0.0029123032,0.0005310123,0.00025171024,0.99130285,0.001880024,0.000116406336,0.0007676487],"genre_scores_gemma":[0.0008168118,0.0008865263,0.010334552,0.00021014456,0.000026649066,0.98715186,0.0002941057,0.0000143506,0.00026494954],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.7212672,0.14223625,0.095076926,0.012652055,0.02198262,0.0067849956],"domain_scores_gemma":[0.710377,0.15300705,0.035171114,0.029258564,0.0645791,0.007607227],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.32273751,0.0056162616,0.016515099,0.021883618,0.010228645,0.016504453,0.006712696,0.015583543,0.047471646],"category_scores_gemma":[0.34148473,0.011419673,0.020960264,0.020360257,0.010930933,0.010551386,0.014928569,0.014547127,0.012404931],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.009140425,0.001009379,0.0030961465,0.76797986,0.0034834887,0.000719019,0.020436835,0.0017360499,0.0036570106,0.013351655,0.036148597,0.13924143],"study_design_scores_gemma":[0.032688804,0.0024056428,0.008676571,0.6912278,0.013959329,0.00046901882,0.011428427,0.0019075671,0.004744214,0.014201375,0.21744612,0.0008450127],"about_ca_topic_score_codex":0.010492783,"about_ca_topic_score_gemma":0.037935708,"teacher_disagreement_score":0.6772625,"about_ca_system_score_codex":0.025614286,"about_ca_system_score_gemma":0.17365986,"threshold_uncertainty_score":0.8351852},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"methods","study_design":"systematic_review","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"methods","study_design":"not_applicable","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4242799417","doi":"10.21203/rs.3.rs-39520/v2","title":"CogTale: An online platform for the evaluation, synthesis and dissemination of evidence from cognitive interventions studies","year":2020,"lang":"en","type":"preprint","venue":"","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"Dementia Australia; University of Melbourne","keywords":"Psychological intervention; Cognition; Computer science; Psychology; Neuroscience; Psychiatry","score_opus":0.9698762387674531,"score_gpt":0.6849343831019687,"score_spread":0.28494185566548436,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4242799417","genre_codex":"dataset","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0061066495,0.0108379135,0.19440576,0.011573176,0.0042779143,0.010779139,0.4707219,0.26233312,0.028964303],"genre_scores_gemma":[0.02148299,0.008964282,0.587829,0.0039411034,0.0018588227,0.026175352,0.27492583,0.05336651,0.021456199],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.98492163,0.006437049,0.0039841817,0.0012527832,0.002883558,0.00052083784],"domain_scores_gemma":[0.79867965,0.1499193,0.0139109045,0.019402552,0.012246402,0.0058411523],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.042568773,0.0035124295,0.0040638857,0.022099808,0.0010279366,0.01100216,0.005396699,0.0038911558,0.19582361],"category_scores_gemma":[0.18951899,0.0022819778,0.00434788,0.01612964,0.001251464,0.0068589533,0.011520105,0.0045588915,0.07780383],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002799197,0.00026343463,0.001342412,0.034948796,0.0020678788,0.00048539112,0.0012547297,0.0015508753,0.0020292483,0.010986418,0.6473309,0.29494074],"study_design_scores_gemma":[0.008225092,0.00040601622,0.0054251887,0.01397571,0.0017514494,0.00044455688,0.00025858864,0.0048393137,0.0035652642,0.039570443,0.92101914,0.0005193329],"about_ca_topic_score_codex":0.0015577949,"about_ca_topic_score_gemma":0.003045871,"teacher_disagreement_score":0.9574312,"about_ca_system_score_codex":0.0021646884,"about_ca_system_score_gemma":0.014069275,"threshold_uncertainty_score":0.6550952},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4253351240","doi":"10.31234/osf.io/bzr82","title":"Integrating open science in the teaching of cognitive research methods: Comparing virtual vs. face-to-face delivery","year":2020,"lang":"en","type":"preprint","venue":"","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta; Dalhousie University","funders":"","keywords":"Openness to experience; Open science; Computer science; Medical education; Transparency (behavior); Psychology; Deliverable; Flexibility (engineering); Mathematics education; Medicine; Engineering; Management","score_opus":0.6015923280553885,"score_gpt":0.6006039930250974,"score_spread":0.0009883350302911653,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4253351240","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.87189704,0.004660434,0.074100934,0.002903777,0.0011826294,0.013769697,0.00029103362,0.00094345806,0.030250909],"genre_scores_gemma":[0.8940695,0.002069684,0.0811625,0.0017207812,0.00030990242,0.017207433,0.00020869233,0.00036338114,0.002888149],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.88761735,0.07946246,0.008203748,0.006357965,0.015859654,0.0024988065],"domain_scores_gemma":[0.6743728,0.27386966,0.015690831,0.017675374,0.010828295,0.0075630527],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.07671565,0.0011471267,0.0020383727,0.00210499,0.0020132535,0.009067094,0.0037402909,0.0031035894,0.007945741],"category_scores_gemma":[0.3075028,0.001161499,0.0018223719,0.0016078592,0.0057001747,0.009044031,0.0087441355,0.005463359,0.0017599654],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.047819342,0.038032595,0.014141121,0.007317346,0.0010882003,0.00023802297,0.043282818,0.007059572,0.008597908,0.014454127,0.0031046094,0.81486434],"study_design_scores_gemma":[0.06692085,0.3744085,0.17272934,0.016900128,0.0049702823,0.0013501514,0.03959082,0.056894153,0.044256452,0.114210375,0.10585867,0.0019102625],"about_ca_topic_score_codex":0.001624462,"about_ca_topic_score_gemma":0.0012954219,"teacher_disagreement_score":0.9962597,"about_ca_system_score_codex":0.004563085,"about_ca_system_score_gemma":0.0051442236,"threshold_uncertainty_score":0.405716},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4255689085","doi":"10.4000/communiquer.2207","title":"Open data &amp; Big data ; Nouveaux défis pour la vie privée","year":2017,"lang":"fr","type":"article","venue":"Communiquer Revue de communication sociale et publique","topic":"Healthcare Systems and Practices","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Humanities; Art","score_opus":0.5537774310923645,"score_gpt":0.5648092961091161,"score_spread":0.011031865016751574,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4255689085","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0045909686,0.060476802,0.087700725,0.78958404,0.018599454,0.00010367547,0.00022684966,0.00034066054,0.03837687],"genre_scores_gemma":[0.30353516,0.20109265,0.18661225,0.19545339,0.049001597,0.0012276507,0.0010344571,0.0018543673,0.06018846],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9011986,0.05050171,0.005908522,0.0072834576,0.031130454,0.0039773155],"domain_scores_gemma":[0.81350386,0.13661791,0.0060337307,0.014436844,0.023481026,0.0059266174],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.07261333,0.0010510425,0.0015486305,0.004086773,0.009802368,0.040046837,0.0031223109,0.0106837535,0.00704138],"category_scores_gemma":[0.119554915,0.0010339308,0.0018187924,0.006009708,0.027981078,0.054491438,0.019422876,0.030748954,0.0027602226],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000054784832,0.00004817678,0.0010750456,0.0008210924,0.00003674114,0.00024570376,0.008883204,0.00038551915,0.00060809834,0.8356304,0.06299141,0.089219734],"study_design_scores_gemma":[0.000012883076,0.000036557314,0.000756775,0.0021097034,0.000019594461,0.00062347105,0.007986592,0.001171235,0.0008433998,0.31753963,0.66878515,0.000115100454],"about_ca_topic_score_codex":0.010111842,"about_ca_topic_score_gemma":0.009425563,"teacher_disagreement_score":0.9968777,"about_ca_system_score_codex":0.010453833,"about_ca_system_score_gemma":0.016358234,"threshold_uncertainty_score":0.38402057},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4282962995","doi":"10.1038/s41597-022-01401-7","title":"A large, curated, open-source stroke neuroimaging dataset to improve lesion segmentation algorithms","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Acute Ischemic Stroke Management","field":"Medicine","cited_by":149,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Research Institute for Aging; Heart and Stroke Foundation; University of Waterloo; Sunnybrook Health Science Centre; Ontario Brain Institute; University of Toronto; University of British Columbia","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; National Institute of Neurological Disorders and Stroke; Medical Research Council; National Institute of General Medical Sciences; Wellcome Trust; U.S. Department of Veterans Affairs","keywords":"Generalizability theory; Segmentation; Computer science; Artificial intelligence; Neuroimaging; Atlas (anatomy); Pattern recognition (psychology); Stroke (engine); Machine learning; Medicine; Mathematics; Statistics","score_opus":0.06612173964472086,"score_gpt":0.35520756388031677,"score_spread":0.2890858242355959,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4282962995","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06313531,0.0045764693,0.12696743,0.0029031541,0.0017017265,0.0022896035,0.7443144,0.04143211,0.012679912],"genre_scores_gemma":[0.03402878,0.00073566684,0.08418107,0.0006717822,0.00024626154,0.0021101413,0.87154615,0.0035725834,0.0029076603],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99564505,0.00082094263,0.0007849007,0.0014533984,0.0010115695,0.00028415048],"domain_scores_gemma":[0.98929733,0.0031473471,0.0008642033,0.0032621736,0.0029533973,0.00047558302],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0061153774,0.0024252199,0.0018835632,0.005143421,0.0017793855,0.0029839466,0.0035447532,0.002502169,0.013295459],"category_scores_gemma":[0.02557747,0.00079299876,0.0024840438,0.0046691364,0.001018811,0.0022918372,0.003557612,0.0025666188,0.018072834],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00094938895,0.0006041335,0.019608336,0.003539988,0.0009490201,0.0006642142,0.0005762145,0.010119229,0.009748481,0.0036093676,0.7664796,0.18315212],"study_design_scores_gemma":[0.00123103,0.0006009301,0.08778549,0.0020132337,0.0010832532,0.003393831,0.00084198755,0.059634298,0.026733536,0.02815184,0.78796077,0.00056986016],"about_ca_topic_score_codex":0.01693151,"about_ca_topic_score_gemma":0.03640953,"teacher_disagreement_score":0.99645525,"about_ca_system_score_codex":0.0019966804,"about_ca_system_score_gemma":0.0054630404,"threshold_uncertainty_score":0.04447776},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4283690014","doi":"10.1101/2022.06.27.22276964","title":"Establishing a core set of open science practices in biomedicine: a modified Delphi study","year":2022,"lang":"en","type":"preprint","venue":"medRxiv","topic":"Delphi Technique in Research","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Montreal Neurological Institute and Hospital; Diabetes Canada; Toronto General Hospital; University of Toronto; Ottawa Hospital; University Health Network; University of Calgary; Princess Margaret Cancer Centre; McGill University; University of Ottawa; Douglas College; Simon Fraser University; Ted Rogers Centre for Heart Research; University of British Columbia; Vector Institute","funders":"Wellcome Trust","keywords":"Delphi method; Open science; Voting; Set (abstract data type); Medical education; Operationalization; Best practice; Inclusion (mineral); Psychology; Political science; Knowledge management; Computer science; Public relations; Medicine; Social psychology","score_opus":0.47831850580845475,"score_gpt":0.5800500123148565,"score_spread":0.10173150650640178,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283690014","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9654528,0.0000786745,0.013235939,0.0014952603,0.000063238556,0.014852209,0.00013700324,0.000044336488,0.0046405764],"genre_scores_gemma":[0.93511415,0.00018573106,0.02599169,0.0012477066,0.000031213436,0.03624648,0.00014663629,0.000036761092,0.000999626],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.8601367,0.1168864,0.0064272196,0.0040274384,0.0054059727,0.0071162577],"domain_scores_gemma":[0.8401346,0.11629662,0.0054732184,0.008082346,0.02448836,0.005524864],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.17307101,0.0010532913,0.0009828091,0.0028846504,0.0064175767,0.00444531,0.0025799444,0.0029338843,0.0038960977],"category_scores_gemma":[0.14008306,0.0014632423,0.0013649688,0.0018594001,0.007332143,0.0064832857,0.011450296,0.0036103087,0.0008812644],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012654813,0.002016602,0.023602938,0.0017510671,0.00009987633,0.00078994554,0.8855527,0.0017711405,0.0048055667,0.011011571,0.002949533,0.06438349],"study_design_scores_gemma":[0.0005867357,0.004693799,0.032345112,0.0014262855,0.00009907089,0.00037212946,0.91030914,0.007533374,0.0043825977,0.011641047,0.0262827,0.0003279837],"about_ca_topic_score_codex":0.0029572616,"about_ca_topic_score_gemma":0.0028363424,"teacher_disagreement_score":0.9974201,"about_ca_system_score_codex":0.008053192,"about_ca_system_score_gemma":0.013068921,"threshold_uncertainty_score":0.9152979},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"methods","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"methods","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4283727562","doi":"10.5195/ijms.2022.1581","title":"Thinking Globally in the Pursuit of Individual Identity: Diversity, Equity, and Inclusion in the International Journal of Medical Students (IJMS)","year":2022,"lang":"en","type":"article","venue":"International Journal of Medical Students","topic":"Diversity and Career in Medicine","field":"Social Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"","keywords":"Equity (law); Inclusion (mineral); Public relations; Diversity (politics); Publishing; Health equity; Population; Political science; Health care; Sociology; Psychology; Social science; Law","score_opus":0.04830136152758891,"score_gpt":0.4304827608843238,"score_spread":0.3821813993567349,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283727562","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005593826,0.022733403,0.0025373031,0.7148796,0.24339367,0.00008426062,0.00005110299,0.00010646909,0.010620437],"genre_scores_gemma":[0.22894262,0.053557638,0.009009162,0.25816703,0.42493463,0.00024402881,0.00013415,0.00044318827,0.024567517],"study_design_codex":"not_applicable","study_design_gemma":"qualitative","domain_scores_codex":[0.9537659,0.022690108,0.0049997824,0.001980453,0.014030215,0.0025335348],"domain_scores_gemma":[0.8365467,0.08483161,0.011916498,0.005496293,0.038454868,0.022754],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.04016726,0.00059521693,0.0010605362,0.003374362,0.007685553,0.031918712,0.0017692114,0.007161405,0.0039201896],"category_scores_gemma":[0.13241033,0.00052299444,0.0008098659,0.002309228,0.013931991,0.009600833,0.0073057893,0.016878841,0.0013429162],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009874469,0.000104465405,0.0029754757,0.0021155092,0.00008862161,0.00051369116,0.02489358,0.00014589874,0.00089439814,0.07372487,0.7529186,0.1415262],"study_design_scores_gemma":[0.000014802735,0.00008028855,0.0017764936,0.0031986465,0.000050788636,0.00069874845,0.010993863,0.00024126764,0.0006519845,0.026688835,0.955533,0.00007116998],"about_ca_topic_score_codex":0.0015310162,"about_ca_topic_score_gemma":0.0027539204,"teacher_disagreement_score":0.04016726,"about_ca_system_score_codex":0.005449691,"about_ca_system_score_gemma":0.02955861,"threshold_uncertainty_score":0.21242732},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","scholarly_communication","open_science"],"domain":"evaluation","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4283802616","doi":"10.5206/cjilsrcsib.v45i1.14149","title":"Measuring the prevalence of open access in Canada: A national comparison","year":2022,"lang":"en","type":"article","venue":"Canadian Journal of Information and Library Science","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"Université de Montréal","funders":"Université du Québec à Montréal","keywords":"Lagging; Political science; Compliance (psychology); Discipline; Momentum (technical analysis); Library science; Medicine; Business; Psychology; Computer science","score_opus":0.5376084805342523,"score_gpt":0.5023064140048592,"score_spread":0.03530206652939305,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4283802616","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8159461,0.037241872,0.0013650255,0.010475166,0.0002928709,0.0003634956,0.07378078,0.00027563822,0.060259018],"genre_scores_gemma":[0.97274405,0.010208326,0.00095485453,0.0007160572,0.000056224908,0.00009479457,0.011331476,0.000065401044,0.0038287593],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98554426,0.00079597934,0.0017386217,0.0015827441,0.008320133,0.0020181583],"domain_scores_gemma":[0.8997086,0.012057792,0.013242199,0.003340332,0.063776866,0.007874212],"candidate_categories":["metaresearch","bibliometrics","open_science"],"consensus_categories":[],"category_scores_codex":[0.008453885,0.00037991605,0.0013942297,0.01974176,0.004487503,0.0053155855,0.0021416435,0.0007789729,0.0044174283],"category_scores_gemma":[0.04100651,0.00033967674,0.0011370011,0.049306788,0.0023053237,0.0020110912,0.0030785685,0.0007405458,0.00040731404],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005168918,0.00008598525,0.87688535,0.0021741951,0.00083147053,0.00037470055,0.009583224,0.0005888271,0.0009405052,0.008922811,0.023946358,0.07514958],"study_design_scores_gemma":[0.00001799187,0.000031467112,0.97543305,0.00031417768,0.00014000441,0.00014650637,0.0038155557,0.00024027762,0.00020112126,0.00023242002,0.019371241,0.00005635883],"about_ca_topic_score_codex":0.99196845,"about_ca_topic_score_gemma":0.9931566,"teacher_disagreement_score":0.99785835,"about_ca_system_score_codex":0.09314594,"about_ca_system_score_gemma":0.1316307,"threshold_uncertainty_score":0.6758242},"labels":[{"model":"gemma","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4289243806","doi":"10.5281/zenodo.3338482","title":"Publishing (and Forgetting) the Small or Medium-sized Scholarly Edition or Cultural Heritage Collection as Linked Open Data: Using Zenodo and Github to Publish the Visionary Cross Project (Abstract)","year":2018,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Lethbridge College; University of Lethbridge","funders":"","keywords":"Publication; Publishing; Forgetting; Cultural heritage; Computer science; World Wide Web; Library science; Data science; History; Political science; Archaeology; Linguistics; Philosophy; Law","score_opus":0.16804431484735488,"score_gpt":0.3341543346348667,"score_spread":0.16611001978751183,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4289243806","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.042060044,0.0044623995,0.47570956,0.056987204,0.014694663,0.0011828439,0.010240299,0.016491897,0.3781711],"genre_scores_gemma":[0.28315443,0.0057384814,0.4763484,0.008765841,0.005268629,0.0018891143,0.0100939255,0.024373027,0.18436807],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9820369,0.008685561,0.0018475695,0.0014988581,0.0052551404,0.0006759453],"domain_scores_gemma":[0.90725404,0.026203612,0.004265174,0.049119666,0.009056106,0.0041014994],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.04255985,0.00048382694,0.00072839356,0.009490401,0.0050645736,0.026444277,0.002378505,0.0021072258,0.02577387],"category_scores_gemma":[0.09148789,0.0007450197,0.00076364545,0.013218008,0.009631532,0.020514624,0.021622676,0.0030759398,0.011686317],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026059608,0.000111911366,0.0044955057,0.00091106375,0.00008991979,0.0010160366,0.033894476,0.0009943314,0.0049080076,0.4712789,0.19843285,0.2836064],"study_design_scores_gemma":[0.000033271106,0.000053641073,0.0017131374,0.00045003716,0.000027946122,0.000418406,0.0070661968,0.00081303116,0.0031502852,0.058977257,0.9271867,0.00011004586],"about_ca_topic_score_codex":0.004613359,"about_ca_topic_score_gemma":0.007125418,"teacher_disagreement_score":0.9976215,"about_ca_system_score_codex":0.003181434,"about_ca_system_score_gemma":0.009087845,"threshold_uncertainty_score":0.22508067},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W4289782436","doi":"10.1101/2022.08.03.22278384","title":"A cross-sectional audit and survey of Open Science and Data Sharing practices at The Montreal Neurological Institute-Hospital","year":2022,"lang":"en","type":"preprint","venue":"medRxiv","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Ottawa; Ottawa Hospital","funders":"Montreal Neurological Institute and Hospital","keywords":"Audit; Open science; Data sharing; Psychology; Medicine; Medical education; Family medicine; Alternative medicine; Accounting; Business","score_opus":0.6950374542905322,"score_gpt":0.6156771880701942,"score_spread":0.07936026622033798,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4289782436","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98609585,0.0016456846,0.0009720884,0.0013497759,0.000022217017,0.0020808105,0.005826849,0.000091074304,0.001915625],"genre_scores_gemma":[0.9897064,0.0012760914,0.002368955,0.0006581523,0.000043560467,0.0033120448,0.001995074,0.000029814559,0.00060999073],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.93493396,0.025215568,0.014239366,0.005953318,0.015895396,0.0037623919],"domain_scores_gemma":[0.72413224,0.06546877,0.11630677,0.013691597,0.06854288,0.011857771],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.050006688,0.0004344398,0.0010672199,0.009642054,0.0029630782,0.0031474526,0.002201841,0.0010172059,0.0022652498],"category_scores_gemma":[0.11526562,0.0010357068,0.0006742535,0.01360817,0.0023981063,0.0038208554,0.004019203,0.0013108242,0.00064480526],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021863921,0.00019064103,0.9656789,0.0010244496,0.0000833104,0.00023609957,0.0140429465,0.000043853124,0.00039235473,0.00018782992,0.0031120835,0.014788932],"study_design_scores_gemma":[0.000034506855,0.00044056072,0.9758244,0.00066075503,0.000045283377,0.00020577072,0.017742766,0.00017990157,0.00030384562,0.0000590651,0.004447881,0.000055265464],"about_ca_topic_score_codex":0.08153437,"about_ca_topic_score_gemma":0.09071824,"teacher_disagreement_score":0.99779814,"about_ca_system_score_codex":0.01315108,"about_ca_system_score_gemma":0.02749432,"threshold_uncertainty_score":0.26446378},"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","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W4290546285","doi":"10.29173/cais1262","title":"The oligopoly’s shift to open access publishing","year":2022,"lang":"en","type":"article","venue":"Proceedings of the Annual Conference of CAIS / Actes du congrès annuel de l ACSI","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":34,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"Université de Montréal; Wilfrid Laurier University; University of Ottawa; Bureau de Coopération Interuniversitaire; Université du Québec à Montréal; Dalhousie University","funders":"Social Sciences and Humanities Research Council of Canada; Natural Sciences and Engineering Research Council of Canada; Canadian Institutes of Health Research","keywords":"Oligopoly; Agency (philosophy); Library science; Publishing; Political science; Economics; Business; Computer science; Sociology; Social science; Law","score_opus":0.40441269756483084,"score_gpt":0.5025465979353135,"score_spread":0.0981339003704827,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4290546285","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6525743,0.0054021985,0.010232476,0.012798319,0.00024776754,0.00014908907,0.0019141327,0.0002056723,0.31647614],"genre_scores_gemma":[0.98770124,0.0011749297,0.0012027224,0.0007463936,0.000096111995,0.000018982948,0.0003430116,0.000028882392,0.008687817],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9928598,0.0010027093,0.0003123559,0.0011227598,0.0036170862,0.0010852779],"domain_scores_gemma":[0.98243284,0.0050548883,0.005358767,0.0011167303,0.0031062264,0.0029305848],"candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.0032629643,0.00019236282,0.0003652317,0.0047488427,0.0017872559,0.008993339,0.00052133924,0.0009196448,0.010084944],"category_scores_gemma":[0.019877387,0.0003111114,0.00053128425,0.008953131,0.0033408597,0.00468949,0.0021786625,0.0011999398,0.0010885892],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00074847176,0.00024135505,0.36409608,0.0005594072,0.00033939586,0.0017747711,0.004408025,0.013260617,0.0029157947,0.39163154,0.03282427,0.18720023],"study_design_scores_gemma":[0.000083383464,0.00028900374,0.6574284,0.00032859662,0.00013081905,0.0013891526,0.007330412,0.014751614,0.0014183607,0.061584506,0.25513956,0.00012620365],"about_ca_topic_score_codex":0.07467023,"about_ca_topic_score_gemma":0.098352134,"teacher_disagreement_score":0.99947864,"about_ca_system_score_codex":0.009979971,"about_ca_system_score_gemma":0.0077375183,"threshold_uncertainty_score":0.14847118},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4290547061","doi":"10.29173/cais1245","title":"Factors influencing Canadian HASS researchers’ open access publishing practices","year":2022,"lang":"en","type":"article","venue":"Proceedings of the Annual Conference of CAIS / Actes du congrès annuel de l ACSI","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"McGill University","funders":"","keywords":"Publishing; Altruism (biology); Prestige; Test (biology); Variance (accounting); Psychology; Visibility; Social psychology; Sociology; Public relations; Political science; Business; Geography; Biology; Law","score_opus":0.6978815989921062,"score_gpt":0.5439937719415102,"score_spread":0.153887827050596,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4290547061","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"incentives","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"incentives","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96009815,0.002352872,0.00030643248,0.008458101,0.00009054733,0.00011143195,0.0013941368,0.000047523623,0.027140785],"genre_scores_gemma":[0.9920677,0.0018835337,0.00052772515,0.00044036272,0.00003171328,0.00003176395,0.00033979723,0.000014063479,0.0046633603],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9902146,0.0012904125,0.00082486746,0.00070739805,0.0050120833,0.001950629],"domain_scores_gemma":[0.93055797,0.015118842,0.009091185,0.0019318784,0.028591704,0.014708434],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.009089593,0.00026412425,0.00049616006,0.007120606,0.011466354,0.0071308375,0.0014471903,0.00073153793,0.0052555073],"category_scores_gemma":[0.035499856,0.00036863497,0.0004806185,0.018803129,0.0032120005,0.0013727956,0.0027423028,0.00073955924,0.00045821548],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00029350526,0.00012522223,0.7970792,0.00091608544,0.00012007664,0.0012826168,0.09855364,0.00032667772,0.0020520976,0.007498198,0.01297497,0.078777745],"study_design_scores_gemma":[0.00002842454,0.00009461763,0.8534405,0.00037206916,0.00007213757,0.0004977936,0.094345786,0.00043684038,0.0007274264,0.00083002716,0.04903071,0.00012373035],"about_ca_topic_score_codex":0.962222,"about_ca_topic_score_gemma":0.97710615,"teacher_disagreement_score":0.9985528,"about_ca_system_score_codex":0.04638428,"about_ca_system_score_gemma":0.11645755,"threshold_uncertainty_score":0.33654302},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"medium"},{"model":"gpt","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W4291982736","doi":"","title":"A Very Long Embargo: Journal Choice Reveals Active Non-Compliance with Funder Open Access Policies by Australian and Canadian Neuroscientists","year":2018,"lang":"en","type":"article","venue":"Spectrum Research Repository (Concordia University)","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Concordia University","funders":"","keywords":"Compliance (psychology); Political science; Public economics; Psychology; Public administration; Public relations; Economics; Social psychology","score_opus":0.4513140322787154,"score_gpt":0.5260346446697138,"score_spread":0.07472061239099842,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4291982736","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95681125,0.0036635878,0.0005293563,0.0066620787,0.00018114864,0.00018427843,0.0054100445,0.0000720754,0.026486324],"genre_scores_gemma":[0.9948698,0.00068598485,0.00045157413,0.0004384707,0.00008594505,0.00007267801,0.0011934554,0.00003698083,0.002165055],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9329384,0.014214653,0.014446311,0.0059775733,0.025369536,0.0070534996],"domain_scores_gemma":[0.4340181,0.2057469,0.24441928,0.03631778,0.060024537,0.019473435],"candidate_categories":["metaresearch","bibliometrics","open_science"],"consensus_categories":[],"category_scores_codex":[0.037437357,0.00032798588,0.0010925079,0.013888717,0.0049328026,0.008725281,0.002518883,0.0016674387,0.010758857],"category_scores_gemma":[0.3354836,0.0004946589,0.0007960052,0.039143905,0.0039621214,0.005422989,0.0049016248,0.0018273653,0.0010735047],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00079351553,0.00011306096,0.92830414,0.00079062063,0.00042830492,0.0007325125,0.015174798,0.00018851845,0.00071085757,0.005802464,0.014953311,0.032007862],"study_design_scores_gemma":[0.00002348958,0.000057556743,0.9687047,0.00036100298,0.00011440512,0.00042535938,0.00984708,0.00029787645,0.00041375306,0.0015409942,0.018136274,0.00007750589],"about_ca_topic_score_codex":0.14888176,"about_ca_topic_score_gemma":0.170739,"teacher_disagreement_score":0.9974811,"about_ca_system_score_codex":0.006793194,"about_ca_system_score_gemma":0.017498247,"threshold_uncertainty_score":0.29603034},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W4293078186","doi":"10.21203/rs.3.rs-1736011/v3","title":"OSeMOSYS Global: An open-source, open data global electricity system model generator","year":2022,"lang":"en","type":"preprint","venue":"Research Square","topic":"Integrated Energy Systems Optimization","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"University College Cork; Loughborough University; Imperial College London; Government of the United Kingdom; Science Foundation Ireland; Foreign, Commonwealth and Development Office; Colorado School of Mines","keywords":"Open source; Electricity system; Electricity; Generator (circuit theory); Open data; Computer science; Electricity generation; Electrical engineering; Engineering; World Wide Web; Operating system; Physics; Power (physics); Software","score_opus":0.11671104922805549,"score_gpt":0.4007202894978045,"score_spread":0.28400924026974905,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4293078186","genre_codex":"methods","genre_gemma":"software","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"software","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.032518625,0.00042159227,0.4237964,0.0010116599,0.0009750365,0.0004533216,0.1288657,0.34918585,0.06277187],"genre_scores_gemma":[0.3614119,0.0006862644,0.26304927,0.0008026248,0.0003064726,0.0011549273,0.21182486,0.120879255,0.039884448],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99969375,0.00008078411,0.000020807433,0.000056501056,0.000113932336,0.000034180255],"domain_scores_gemma":[0.9991504,0.0003082199,0.000050105016,0.00023532994,0.00016826726,0.00008764],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0008302986,0.0012254502,0.00082529057,0.0008985551,0.0003256686,0.0010165432,0.002526012,0.00078133703,0.035735335],"category_scores_gemma":[0.0033970948,0.00064718234,0.0008002483,0.0011333806,0.0003041716,0.0021349113,0.0010191659,0.0016244682,0.010988118],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006221641,0.00035681247,0.004008244,0.0008180276,0.00034630302,0.00030580678,0.00022234661,0.34825823,0.007394105,0.026285144,0.512134,0.099248774],"study_design_scores_gemma":[0.0004553661,0.0000795526,0.0013164397,0.000044711014,0.00004266558,0.00007598937,0.000043266435,0.8205222,0.0059810127,0.02835537,0.14301717,0.00006633792],"about_ca_topic_score_codex":0.0045442563,"about_ca_topic_score_gemma":0.006482917,"teacher_disagreement_score":0.997474,"about_ca_system_score_codex":0.00049054355,"about_ca_system_score_gemma":0.0011599316,"threshold_uncertainty_score":0.11954659},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"simulation_or_modeling","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4299584860","doi":"","title":"Goldener, grüner und \"anderer\" Weg zu Open Access: Statistiken für Österreich","year":2015,"lang":"en","type":"article","venue":"E-LIS Repository (University of Naples Federico II)","topic":"Libraries and Information Services","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Political science","score_opus":0.10093271775872135,"score_gpt":0.2991196745373948,"score_spread":0.19818695677867343,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4299584860","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2639217,0.3659157,0.047498785,0.110946335,0.0040621907,0.00024220305,0.010385541,0.00055108074,0.19647646],"genre_scores_gemma":[0.87696147,0.07708955,0.021312753,0.005324272,0.001376713,0.00025377868,0.0027644727,0.0002767146,0.014640255],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.986925,0.003406495,0.0017001309,0.0014269156,0.0055202357,0.0010211292],"domain_scores_gemma":[0.9658111,0.023363085,0.0038540931,0.0020971554,0.0035983608,0.0012762469],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.017145896,0.0005949172,0.00092328485,0.013169285,0.0023499706,0.0075756526,0.00057596085,0.001944184,0.0057504075],"category_scores_gemma":[0.03429326,0.00045811394,0.00069920724,0.018055245,0.005869396,0.011466872,0.004581645,0.0033549136,0.0013886422],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036041913,0.00010472683,0.061967935,0.0016461982,0.0002943575,0.00045768198,0.020773273,0.00072272954,0.000736573,0.46836478,0.05138719,0.39318419],"study_design_scores_gemma":[0.00006898799,0.000136713,0.12184191,0.004170086,0.00025629776,0.0011259827,0.014813167,0.0011505272,0.0025909694,0.2834003,0.57018846,0.00025660542],"about_ca_topic_score_codex":0.014642607,"about_ca_topic_score_gemma":0.015490174,"teacher_disagreement_score":0.99942404,"about_ca_system_score_codex":0.005129673,"about_ca_system_score_gemma":0.00410234,"threshold_uncertainty_score":0.09067726},"labels":[{"model":"gemma","categories":["open_science","bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4301396488","doi":"10.5334/dsj-2022-017","title":"A Survey on Publicly Available Open Datasets Derived From Electronic Health Records (EHRs) of Patients with Neuroblastoma","year":2022,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Neuroblastoma Research and Treatments","field":"Medicine","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Neuroblastoma; Computer science; Health records; Open science; License; The Internet; Electronic health record; Open data; World Wide Web; Internet privacy; Medicine; Information retrieval; Data mining; Political science; Mathematics; Statistics; Health care","score_opus":0.07305559528067233,"score_gpt":0.3504970068041415,"score_spread":0.27744141152346913,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4301396488","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10244699,0.017764082,0.005774887,0.010402432,0.00051519886,0.0003713639,0.8513907,0.0009353296,0.010399079],"genre_scores_gemma":[0.08552543,0.008559856,0.0073492927,0.003592142,0.00027250784,0.0008958807,0.89253575,0.00022177181,0.001047476],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9803949,0.005603066,0.004827285,0.0022522719,0.0060142432,0.0009082492],"domain_scores_gemma":[0.9168954,0.048961394,0.012480886,0.006739492,0.012790139,0.0021327105],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.010214086,0.0006292441,0.0009444067,0.008501466,0.000736537,0.0022419856,0.0014861313,0.0013646354,0.0031232387],"category_scores_gemma":[0.06579199,0.0003146983,0.0010362383,0.015663758,0.00060855603,0.0030394024,0.0037584193,0.00096511486,0.0016585814],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00097261486,0.00020320005,0.3698104,0.010269303,0.0007011017,0.00043078617,0.0019072355,0.0013101127,0.0019689866,0.0065938234,0.43834662,0.16748573],"study_design_scores_gemma":[0.00008956951,0.00011690805,0.42255384,0.0047224453,0.00035440028,0.0014055448,0.0036305594,0.001526093,0.0020329498,0.0035194948,0.5598889,0.00015931293],"about_ca_topic_score_codex":0.0060037975,"about_ca_topic_score_gemma":0.008588018,"teacher_disagreement_score":0.9985139,"about_ca_system_score_codex":0.0012749704,"about_ca_system_score_gemma":0.0030222752,"threshold_uncertainty_score":0.0540179},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4302774140","doi":"10.1016/j.jshs.2022.09.005","title":"Observations regarding open access publishing in hybrid journals in sport sciences","year":2022,"lang":"en","type":"editorial","venue":"Journal of sport and health science/Journal of Sport and Health Science","topic":"Sports Analytics and Performance","field":"Economics, Econometrics and Finance","cited_by":6,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Publishing; Open science; Psychology; Open peer review; Political science; Plant biology; Biology; Mathematics; Law; Statistics","score_opus":0.23355663079733974,"score_gpt":0.4196860091467736,"score_spread":0.18612937834943388,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4302774140","genre_codex":"editorial","genre_gemma":"editorial","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00061071554,0.004934202,0.00029808682,0.21153687,0.77597374,0.000024117486,0.00022811526,0.00012301588,0.006271277],"genre_scores_gemma":[0.005927341,0.003163245,0.0002535981,0.05800018,0.9231988,0.00004053075,0.000101779406,0.00011995735,0.009194578],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9674164,0.0040577627,0.0038633058,0.0025781717,0.020180259,0.0019041564],"domain_scores_gemma":[0.67767304,0.17624633,0.015834352,0.009333065,0.09932881,0.02158448],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.038981162,0.0015095896,0.0017837835,0.007258726,0.00434769,0.017045502,0.0034512202,0.016328935,0.011298989],"category_scores_gemma":[0.18603581,0.00082948763,0.0013242421,0.0068311617,0.004957042,0.0073630502,0.0032576618,0.016566455,0.004542037],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000078095916,0.00001501737,0.00024658142,0.00028858022,0.000027512355,0.00011499671,0.00012040546,0.000029883924,0.00009412415,0.0017870043,0.9874313,0.00976646],"study_design_scores_gemma":[0.000121801895,0.00006816405,0.003894822,0.00094042614,0.000086163316,0.00023974579,0.0005387406,0.0002570662,0.00025376037,0.0037211357,0.98982435,0.00005378984],"about_ca_topic_score_codex":0.0024915908,"about_ca_topic_score_gemma":0.008099923,"teacher_disagreement_score":0.9829545,"about_ca_system_score_codex":0.004927142,"about_ca_system_score_gemma":0.004289228,"threshold_uncertainty_score":0.20615453},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4306253448","doi":"10.1002/pra2.641","title":"Investigating Open Access Publishing Practices of Early and <scp>Mid‐Career</scp> Researchers in Humanities and Social Sciences Disciplines","year":2022,"lang":"en","type":"article","venue":"Proceedings of the Association for Information Science and Technology","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University","funders":"","keywords":"Publishing; Audience measurement; Descriptive statistics; Library science; Scholarly communication; Open peer review; Citation; Psychology; Political science; Advertising; Computer science; Plant biology; Business","score_opus":0.6736601966620449,"score_gpt":0.5621824102859028,"score_spread":0.11147778637614214,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4306253448","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reporting","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reporting","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99172956,0.0014286385,0.0003993884,0.0016110158,0.000031610907,0.000049498278,0.0001605103,0.000018307763,0.0045714467],"genre_scores_gemma":[0.99644345,0.0011590233,0.00058155897,0.0001787771,0.00002529167,0.000026892565,0.000084254374,0.000006511608,0.0014941689],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98554975,0.0047276253,0.0017076802,0.00091930496,0.005732135,0.0013634758],"domain_scores_gemma":[0.7988784,0.085319065,0.04448499,0.0074199266,0.038749572,0.025148084],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.021376228,0.00013380975,0.0004197479,0.0055297213,0.004298306,0.006616555,0.0013365536,0.0007720359,0.003017503],"category_scores_gemma":[0.08737719,0.00026021944,0.0002696786,0.0087638395,0.002309161,0.0034165713,0.003288797,0.00074530276,0.0004859408],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001879745,0.0002438745,0.73469937,0.00064683415,0.000057696358,0.00070354814,0.1399308,0.00010449324,0.0024134,0.0021106373,0.002053112,0.116848156],"study_design_scores_gemma":[0.000020141737,0.000289311,0.75256115,0.00045205778,0.00004004765,0.00083854824,0.21915391,0.00042227347,0.001443782,0.0010529493,0.023656018,0.00006978262],"about_ca_topic_score_codex":0.07282599,"about_ca_topic_score_gemma":0.12192014,"teacher_disagreement_score":0.9986634,"about_ca_system_score_codex":0.0061733583,"about_ca_system_score_gemma":0.02130706,"threshold_uncertainty_score":0.14480418},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4309236697","doi":"10.38126/jspg210202","title":"Equitable Research Capacity Towards the Sustainable Development Goals: The Case for Open Science Hardware","year":2022,"lang":"en","type":"article","venue":"Journal of Science Policy & Governance","topic":"Open Source Software Innovations","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"Natural Sciences and Engineering Research Council of Canada; Alfred P. Sloan Foundation","keywords":"Procurement; Capacity building; Open science; Sustainable development; Open source hardware; Service (business); Business; Open research; Computer science; Engineering management; Software; Environmental economics; Political science; Open source; Economic growth; Economics; Engineering; Marketing; World Wide Web","score_opus":0.13395092984454793,"score_gpt":0.4049255625573325,"score_spread":0.27097463271278455,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4309236697","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.034151703,0.005155285,0.06875165,0.6990681,0.0010918964,0.0005598438,0.0001705414,0.00023536618,0.19081564],"genre_scores_gemma":[0.8691129,0.0034870869,0.042333785,0.06585689,0.0015910427,0.0018254794,0.0001311863,0.00021423839,0.015447424],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.8765007,0.07342614,0.0051200846,0.011871137,0.019876054,0.0132059045],"domain_scores_gemma":[0.73093975,0.17684738,0.017372586,0.032095905,0.02175442,0.020989968],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.15298839,0.0011165806,0.0016007563,0.004227992,0.009663013,0.029095985,0.0051075513,0.020757226,0.011025762],"category_scores_gemma":[0.1827032,0.0011672727,0.0018723807,0.004263519,0.06322267,0.04660153,0.050602574,0.017519552,0.0015269556],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019418883,0.00003068773,0.00046558835,0.00011658778,0.00001858261,0.000069073074,0.0008645134,0.00090745714,0.00012943201,0.9879497,0.003039585,0.006389418],"study_design_scores_gemma":[0.00005013044,0.00003847155,0.0004970907,0.00037712028,0.000014677713,0.00007063412,0.00089592714,0.0010955783,0.00030260472,0.94551563,0.051111408,0.00003071985],"about_ca_topic_score_codex":0.004812764,"about_ca_topic_score_gemma":0.002753978,"teacher_disagreement_score":0.9948925,"about_ca_system_score_codex":0.016970865,"about_ca_system_score_gemma":0.062936366,"threshold_uncertainty_score":0.8090896},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"design_other","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4313888162","doi":"10.20933/100001250","title":"IKT for Research Stage 3: Proposal Development","year":2022,"lang":"en","type":"report","venue":"","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Simon Fraser University; University of Dundee","keywords":"General partnership; Knowledge translation; Stakeholder; Stakeholder engagement; Knowledge transfer; Public relations; Best practice; Medical education; Translational research; Dialog box; Political science; Knowledge management; Engineering; Business; Sociology; Medicine; Computer science","score_opus":0.9721307628349205,"score_gpt":0.8474813893636434,"score_spread":0.12464937347127714,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4313888162","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.03147666,0.003571374,0.19716273,0.115385406,0.040494625,0.23997048,0.021899303,0.00936004,0.34067944],"genre_scores_gemma":[0.06153761,0.002095034,0.30642858,0.023643948,0.003329599,0.27722895,0.009834315,0.0064002597,0.3095017],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.92008567,0.04448577,0.0058000293,0.005906624,0.016477138,0.007244771],"domain_scores_gemma":[0.785083,0.10180558,0.0057934746,0.021449324,0.0579298,0.027938843],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.12676091,0.0014589459,0.0017575877,0.0057368437,0.0077560497,0.015160011,0.00614552,0.00842947,0.24288982],"category_scores_gemma":[0.24594063,0.0021436398,0.0030298664,0.0036362442,0.004923176,0.00869271,0.017003855,0.012157766,0.104451396],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00167102,0.000741388,0.0011721706,0.0054156855,0.000059053615,0.002046668,0.055036794,0.00075499364,0.004183611,0.074574955,0.55641377,0.2979299],"study_design_scores_gemma":[0.0003167268,0.00032913557,0.00091017416,0.0022789144,0.00003234078,0.0002698917,0.011804207,0.0004818914,0.0015969881,0.018145539,0.9637338,0.00010043623],"about_ca_topic_score_codex":0.005382497,"about_ca_topic_score_gemma":0.0061964802,"teacher_disagreement_score":0.24288982,"about_ca_system_score_codex":0.013224571,"about_ca_system_score_gemma":0.080704615,"threshold_uncertainty_score":0.8125474},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4316495494","doi":"10.12688/wellcomeopenres.18681.1","title":"Pf7: an open dataset of Plasmodium falciparum genome variation in 20,000 worldwide samples","year":2023,"lang":"en","type":"preprint","venue":"Wellcome Open Research","topic":"Malaria Research and Control","field":"Medicine","cited_by":148,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary","funders":"Faculty of Tropical Medicine, Mahidol University; Addis Ababa University; National Institutes of Health; Universiteit Antwerpen; National Institute for Medical Research; Ministère de la Santé; Universidad Nacional de Colombia; Bundesministerium für Gesundheit; World Health Organization; University of Ghana; Bill and Melinda Gates Foundation; Mahidol University; University College London; National Center for Advancing Translational Sciences; National Human Genome Research Institute; Wellcome Trust; National Institute of Allergy and Infectious Diseases; Medical Research Council; London School of Hygiene and Tropical Medicine","keywords":"Plasmodium falciparum; Variation (astronomy); Biology; Genome; Computational biology; Structural variation; Genetics; Malaria; Evolutionary biology; Gene; Immunology","score_opus":0.2762833932571187,"score_gpt":0.45507050550302575,"score_spread":0.17878711224590704,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4316495494","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038227695,0.00027545053,0.000937634,0.00011553819,0.000028342056,0.000026926748,0.9930842,0.00087990385,0.00082918676],"genre_scores_gemma":[0.0038689638,0.00010280399,0.0012793054,0.000041697745,0.000014212036,0.00007648769,0.99420667,0.00017762734,0.0002322967],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.998396,0.000275985,0.00017867275,0.00050483935,0.00043779565,0.00020683395],"domain_scores_gemma":[0.9976349,0.00069165154,0.0004084217,0.0005499208,0.0003907118,0.00032445294],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0017115934,0.0016859785,0.0014828283,0.0040912605,0.00067843514,0.0015775949,0.0013945888,0.0013605276,0.025377795],"category_scores_gemma":[0.00701324,0.0008090687,0.0008966416,0.008466877,0.0004252073,0.000916853,0.0019281611,0.0014318642,0.02133715],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009331374,0.00010183151,0.024472695,0.0023368471,0.0005317835,0.0005648796,0.0003359705,0.0035396037,0.006657445,0.0018043973,0.91877705,0.039944407],"study_design_scores_gemma":[0.00082221784,0.00014731969,0.08532767,0.00036691956,0.0002398978,0.00093177124,0.0002565408,0.0026396497,0.0035454852,0.003926647,0.90167403,0.000121869954],"about_ca_topic_score_codex":0.011653115,"about_ca_topic_score_gemma":0.010402996,"teacher_disagreement_score":0.025377795,"about_ca_system_score_codex":0.0007471136,"about_ca_system_score_gemma":0.0020563575,"threshold_uncertainty_score":0.08489716},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4317525596","doi":"10.2196/40814","title":"Understanding Public Attitudes and Willingness to Share Commercial Data for Health Research: Survey Study in the United Kingdom","year":2023,"lang":"en","type":"article","venue":"JMIR Public Health and Surveillance","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":20,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Peter Sowerby Foundation; University College London; Cancer Research UK","keywords":"Public health; Survey research; Environmental health; Survey data collection; Internet privacy; Psychology; Business; Medicine; Computer science; Applied psychology; Nursing; Statistics","score_opus":0.9634509391386481,"score_gpt":0.6910705090113438,"score_spread":0.2723804301273043,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4317525596","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9985732,0.00023627083,0.00003671587,0.0002876122,0.000003238207,0.000021270052,0.00018620073,0.000001217333,0.000654269],"genre_scores_gemma":[0.99918836,0.00023946339,0.00005994554,0.00019266487,0.0000022748843,0.000030960313,0.000081615464,0.0000014244283,0.00020327723],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9964128,0.0016983974,0.0006018778,0.00030027566,0.00048877794,0.00049791415],"domain_scores_gemma":[0.98848,0.0032597538,0.0046884394,0.0005176974,0.0019177712,0.0011362521],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.003072922,0.00023653782,0.00044420396,0.0010932325,0.00070155907,0.0018323341,0.0004867159,0.00094510784,0.0034197469],"category_scores_gemma":[0.01050964,0.00049303623,0.00057555916,0.0015348267,0.0010176213,0.0012432092,0.0017231135,0.00076171843,0.0006818183],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014709447,0.00013584724,0.9710585,0.0001744496,0.000038654995,0.0005129127,0.021545315,0.000028237262,0.00040946138,0.000091380345,0.00081912504,0.0050390796],"study_design_scores_gemma":[0.000011412395,0.00015176005,0.97343814,0.0001431149,0.000016121754,0.00036060426,0.024554692,0.000068839465,0.000067762,0.000019807772,0.0011538952,0.000013923517],"about_ca_topic_score_codex":0.09399267,"about_ca_topic_score_gemma":0.08338501,"teacher_disagreement_score":0.9969271,"about_ca_system_score_codex":0.0025432792,"about_ca_system_score_gemma":0.0017991902,"threshold_uncertainty_score":0.18689114},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4319443012","doi":"10.1007/978-981-19-8590-4_6","title":"Reflections on Remixing Open Access Content into Open Educational Resources: A New Paradigm for Sustainable Data-Driven Language Learning Systems Design in Higher Education","year":2023,"lang":"en","type":"book-chapter","venue":"Future education and learning spaces","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"University of Waikato","keywords":"Open educational resources; Computer science; Open education; Multimedia; World Wide Web","score_opus":0.16203991385795963,"score_gpt":0.4263413651949371,"score_spread":0.26430145133697747,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4319443012","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010942099,0.009579267,0.04939676,0.49331224,0.006064378,0.00012011431,0.00013520007,0.00036146137,0.4300884],"genre_scores_gemma":[0.2870531,0.016669855,0.053137697,0.07323198,0.0025405998,0.00035439187,0.00014199427,0.0015095893,0.5653607],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99455255,0.0025555291,0.00015102014,0.0004605129,0.0017937704,0.000486519],"domain_scores_gemma":[0.9920359,0.0061122533,0.0001219029,0.0004748231,0.00078227103,0.0004728569],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.010498658,0.00062006497,0.0003486925,0.0006756825,0.003980001,0.015166077,0.0022862486,0.009044095,0.014748459],"category_scores_gemma":[0.008772994,0.00053808995,0.0005621235,0.001250266,0.017714787,0.03090543,0.006371297,0.011766615,0.0032521791],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001848867,0.000043223656,0.00006649269,0.0000852783,0.0000027632962,0.00022011762,0.014822808,0.0005367278,0.0007555271,0.8737418,0.084960684,0.0247461],"study_design_scores_gemma":[0.000008068827,0.000016359467,0.000084246865,0.0001772657,0.0000030648632,0.00016885188,0.010777631,0.00058996223,0.0012791053,0.15268481,0.83418936,0.000021289123],"about_ca_topic_score_codex":0.012035786,"about_ca_topic_score_gemma":0.018977206,"teacher_disagreement_score":0.99771374,"about_ca_system_score_codex":0.0073836343,"about_ca_system_score_gemma":0.005279052,"threshold_uncertainty_score":0.05552286},"labels":[{"model":"gemma","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W4323352270","doi":"10.1111/jnc.15776","title":"Journal of Neurochemistry embraces Transparent Peer Review","year":2023,"lang":"en","type":"editorial","venue":"Journal of Neurochemistry","topic":"Academic Writing and Publishing","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Western University","funders":"","keywords":"Neurochemistry; Neuroscience; Psychology; Cognitive science; Neurology","score_opus":0.059789098445692995,"score_gpt":0.30817777420418985,"score_spread":0.24838867575849685,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4323352270","genre_codex":"editorial","genre_gemma":"editorial","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00015630994,0.0056656213,0.0037361912,0.087334886,0.8947563,0.00012935408,0.000044042932,0.0003026488,0.007874605],"genre_scores_gemma":[0.0063600168,0.018194005,0.0071799443,0.06081285,0.8724716,0.00019538193,0.00012963587,0.00066602934,0.033990674],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9372561,0.009987962,0.00691401,0.0028368195,0.04107594,0.0019291668],"domain_scores_gemma":[0.70909476,0.058210924,0.012700097,0.013729786,0.18341032,0.022854164],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.043450512,0.0012688839,0.0019831567,0.0030449834,0.0052398783,0.019386636,0.0028634856,0.010762923,0.008831446],"category_scores_gemma":[0.17272235,0.0009828489,0.0012139753,0.0016193305,0.0077796127,0.007446561,0.004180777,0.020375911,0.018016672],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003265532,0.000027162416,0.00015958726,0.00048318057,0.000022668872,0.00028573142,0.00020504097,0.000060658684,0.00040198228,0.007416354,0.9537348,0.037170213],"study_design_scores_gemma":[0.0000097645325,0.000015096902,0.00017704917,0.00028213402,0.000009002606,0.00043830284,0.000081091246,0.000108743,0.00021428894,0.004409276,0.9942316,0.000023778273],"about_ca_topic_score_codex":0.0009388421,"about_ca_topic_score_gemma":0.0023307854,"teacher_disagreement_score":0.95654947,"about_ca_system_score_codex":0.002995159,"about_ca_system_score_gemma":0.01845075,"threshold_uncertainty_score":0.22979105},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"evaluation","study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"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":"W4323354418","doi":"10.1145/3564623","title":"OER for Ethics and Computing Open Access Collection","year":2023,"lang":"en","type":"book","venue":"ACM eBooks","topic":"Ethics and Social Impacts of AI","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Syllabus; Computer science; Context (archaeology); Process (computing); Task (project management); Set (abstract data type); Engineering ethics; Data science; World Wide Web; Mathematics education; Engineering; Psychology","score_opus":0.45128180107448523,"score_gpt":0.554812576344325,"score_spread":0.10353077526983978,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4323354418","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001808112,0.0036592765,0.014846854,0.011756189,0.0076606735,0.0036100338,0.35320026,0.011279584,0.5921791],"genre_scores_gemma":[0.009733842,0.0065957783,0.022083675,0.0056429384,0.0050229505,0.0053275363,0.43126312,0.012228061,0.50210214],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9773679,0.003111866,0.0034679275,0.0015398405,0.013030283,0.0014821324],"domain_scores_gemma":[0.89662063,0.017868316,0.005341924,0.022218697,0.048506703,0.009443756],"candidate_categories":["open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.014684378,0.0013916953,0.0023730542,0.032153115,0.00500425,0.017380891,0.0033554616,0.0031187392,0.4101564],"category_scores_gemma":[0.08416294,0.0010361341,0.0013036043,0.028671147,0.0019808742,0.0125137465,0.011602801,0.003923524,0.37747008],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004770693,0.00005422602,0.0004509461,0.0010370667,0.00001133621,0.000093211114,0.00023463774,0.000049342852,0.0003206032,0.009721481,0.92581064,0.06216886],"study_design_scores_gemma":[0.000006253627,0.000008083955,0.0006839738,0.00030719308,0.0000035761627,0.000053993226,0.00007296489,0.00003976058,0.00012257147,0.0013263226,0.9973615,0.000013674349],"about_ca_topic_score_codex":0.0050319387,"about_ca_topic_score_gemma":0.007149437,"teacher_disagreement_score":0.99664456,"about_ca_system_score_codex":0.004027999,"about_ca_system_score_gemma":0.014488204,"threshold_uncertainty_score":0.8413403},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4377020955","doi":"10.12688/f1000research.133710.1","title":"Making science public: a review of journalists’ use of Open Science research","year":2023,"lang":"en","type":"review","venue":"F1000Research","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"Social Sciences and Humanities Research Council of Canada; Arts and Humanities Research Council; Conselho Nacional de Desenvolvimento Científico e Tecnológico; Fundação de Amparo à Pesquisa do Estado de São Paulo; Deutsche Forschungsgemeinschaft","keywords":"Open peer review; Open science; Plant biology; Open data; Physiology; Citizen science; Neuroscience; Engineering ethics; Political science; Medicine; Psychology; Biology; Engineering; Botany; Physics","score_opus":0.9515545007136529,"score_gpt":0.7551723002125759,"score_spread":0.19638220050107702,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4377020955","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0014469111,0.97282696,0.00047378908,0.019617189,0.0023592585,0.000040635445,0.00005684772,0.000016844395,0.0031615458],"genre_scores_gemma":[0.01570929,0.970414,0.0011216052,0.009704157,0.00252551,0.00010350453,0.00006957168,0.00004890285,0.00030341098],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.9075096,0.047348484,0.017790478,0.005194503,0.019791182,0.0023656986],"domain_scores_gemma":[0.3902825,0.52841073,0.028830111,0.007660686,0.039654456,0.005161427],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.10087683,0.0009895783,0.0024173441,0.027631072,0.0046094717,0.020137716,0.0029431486,0.0069631725,0.0034339223],"category_scores_gemma":[0.29714608,0.001596137,0.002373184,0.031050164,0.009611685,0.017142339,0.0059224954,0.0059413,0.0012234825],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000262276,0.00009627124,0.005944424,0.16270834,0.00072158553,0.0006948003,0.038013756,0.00027062013,0.00078134733,0.026403062,0.062009253,0.7020943],"study_design_scores_gemma":[0.00004501517,0.00014489134,0.008474229,0.21197073,0.0006030619,0.0011956951,0.016274143,0.00008251134,0.00032429292,0.00611472,0.7546602,0.00011053911],"about_ca_topic_score_codex":0.004263153,"about_ca_topic_score_gemma":0.009705795,"teacher_disagreement_score":0.99705684,"about_ca_system_score_codex":0.0076618167,"about_ca_system_score_gemma":0.029291227,"threshold_uncertainty_score":0.53349406},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4378782503","doi":"10.1146/annurev-genom-110122-084756","title":"Federated Analysis for Privacy-Preserving Data Sharing: A Technical and Legal Primer","year":2023,"lang":"en","type":"review","venue":"Annual Review of Genomics and Human Genetics","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":42,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"National Human Genome Research Institute; National Cancer Institute; National Heart, Lung, and Blood Institute; National Institutes of Health","keywords":"Data sharing; Information privacy; Internet privacy; Computer science; Data Protection Act 1998; Data science; Computer security","score_opus":0.1519487372290538,"score_gpt":0.39968609060483273,"score_spread":0.24773735337577893,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4378782503","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00035819408,0.91911185,0.038853854,0.030257935,0.0013457562,0.00008853724,0.00007971791,0.00010333949,0.009800881],"genre_scores_gemma":[0.00741134,0.94729096,0.033259556,0.007898454,0.0018408309,0.00020704391,0.00015325277,0.00005903334,0.0018796738],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9853161,0.007531788,0.0010402134,0.0009580633,0.0046538883,0.00050000736],"domain_scores_gemma":[0.96295565,0.029373981,0.001497949,0.0024768105,0.003251756,0.000443842],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.025898902,0.0010253267,0.0017050405,0.006976004,0.0015661535,0.006980758,0.0036627338,0.0055196374,0.0034887814],"category_scores_gemma":[0.02619612,0.00077900407,0.0021052584,0.008917542,0.0080075,0.014329203,0.004714454,0.00928478,0.0021538946],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000048486952,0.00008252795,0.00041232305,0.0065305936,0.00012836895,0.00032463158,0.0004615345,0.0010592863,0.0005912934,0.47198865,0.02742395,0.49094832],"study_design_scores_gemma":[0.000015967948,0.00005689045,0.00031000335,0.009080913,0.00006785093,0.0013114449,0.00025789943,0.00080623396,0.00092503487,0.12817384,0.8589376,0.000056209752],"about_ca_topic_score_codex":0.0023155734,"about_ca_topic_score_gemma":0.0015407761,"teacher_disagreement_score":0.9963373,"about_ca_system_score_codex":0.0045227455,"about_ca_system_score_gemma":0.008806708,"threshold_uncertainty_score":0.13696814},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4378806110","doi":"10.1136/bmjebm-2022-112200","title":"Plain language summaries: enhancing patient-centred care and improving accessibility of health research","year":2023,"lang":"en","type":"article","venue":"BMJ evidence-based medicine","topic":"Health Literacy and Information Accessibility","field":"Health Professions","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario Stroke Network; University Health Network; University of British Columbia","funders":"","keywords":"Plain language; Health care; Medicine; Computer science; Nursing; Political science","score_opus":0.264962570399348,"score_gpt":0.5450406684158028,"score_spread":0.2800780980164548,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4378806110","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reporting","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reporting","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.30494156,0.073936746,0.09416919,0.33962178,0.008811609,0.0046969145,0.009851387,0.008189313,0.15578142],"genre_scores_gemma":[0.61969906,0.061211605,0.24283749,0.040313095,0.00859252,0.0027209842,0.0045002205,0.0009844977,0.019140553],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.97594255,0.016503336,0.003943908,0.00058029726,0.0027178382,0.00031209373],"domain_scores_gemma":[0.8163881,0.1363764,0.02446415,0.005856057,0.01051154,0.0064037116],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.02136286,0.00046334983,0.0011016504,0.0024017377,0.0007962184,0.005294188,0.0011034596,0.001981834,0.022817198],"category_scores_gemma":[0.20306848,0.0006100397,0.0010019109,0.0017491749,0.00052718335,0.006600228,0.0039477516,0.0018747787,0.004209535],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0019345201,0.0013770934,0.016554443,0.01058473,0.00029563697,0.0006382447,0.007008804,0.00044327977,0.0016990632,0.0033927173,0.15901418,0.7970573],"study_design_scores_gemma":[0.0063049584,0.0076732505,0.20214352,0.030256825,0.0033412839,0.008204596,0.016403066,0.0042324727,0.010412769,0.05639978,0.65365857,0.00096890796],"about_ca_topic_score_codex":0.0008482824,"about_ca_topic_score_gemma":0.0024242639,"teacher_disagreement_score":0.97863716,"about_ca_system_score_codex":0.00079716166,"about_ca_system_score_gemma":0.0038545288,"threshold_uncertainty_score":0.112978935},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4379016279","doi":"10.1038/d41586-023-01819-y","title":"Preprint clubs: why it takes a village to do peer review","year":2023,"lang":"en","type":"article","venue":"Nature","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":3,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ontario College of Art and Design","funders":"","keywords":"Preprint; Peer review; Library science; Political science; Medical education; Sociology; Medicine; World Wide Web; Computer science","score_opus":0.09212102075893597,"score_gpt":0.4533360865420113,"score_spread":0.36121506578307533,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379016279","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.04415187,0.0070582163,0.05623661,0.5252565,0.06911891,0.0009557852,0.0011484155,0.0086372,0.28743643],"genre_scores_gemma":[0.505205,0.0033194388,0.040643636,0.067825936,0.0306155,0.0013500374,0.0015819113,0.006781413,0.34267718],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9091499,0.043821372,0.003757152,0.0059340266,0.03190744,0.0054300185],"domain_scores_gemma":[0.5967406,0.13358863,0.01942615,0.058565676,0.11697238,0.0747065],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.07690642,0.0006380508,0.0012816619,0.0034287912,0.011895978,0.01642592,0.0038196724,0.007649547,0.05651749],"category_scores_gemma":[0.24401258,0.0009805817,0.0008526662,0.0029634896,0.011233099,0.012528981,0.010448847,0.009532408,0.028206814],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001834205,0.00014525442,0.0058104102,0.0004954243,0.00015724482,0.000589511,0.005912003,0.00028573614,0.0018121558,0.081492096,0.78669345,0.11642335],"study_design_scores_gemma":[0.000094962415,0.00011454741,0.005456354,0.00027486988,0.00003469394,0.00034748,0.005153334,0.0008465195,0.000798119,0.10534019,0.881408,0.00013088963],"about_ca_topic_score_codex":0.005894996,"about_ca_topic_score_gemma":0.011252236,"teacher_disagreement_score":0.92309356,"about_ca_system_score_codex":0.0038076818,"about_ca_system_score_gemma":0.022591716,"threshold_uncertainty_score":0.40672487},"labels":[{"model":"gemma","categories":["metaresearch","scholarly_communication"],"domain":"evaluation","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication","open_science"],"domain":"evaluation","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4379389436","doi":"10.3138/jsp-2022-0073","title":"Toward a Potential Solution of the Crisis in Scholarly Publishing: An Academic Research Community Alliance Model","year":2023,"lang":"en","type":"article","venue":"Journal of Scholarly Publishing","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Publishing; Alliance; Public relations; Perspective (graphical); Academic community; Sociology; State (computer science); Set (abstract data type); Political science; Computer science; Social science; Law","score_opus":0.4659224511730276,"score_gpt":0.4779319057638813,"score_spread":0.012009454590853674,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379389436","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08748783,0.001227878,0.29992512,0.17590277,0.0009200135,0.00091453316,0.0001394694,0.00047264982,0.4330098],"genre_scores_gemma":[0.8522178,0.0010657237,0.1025403,0.003597696,0.00028946053,0.0008386025,0.00011550634,0.0000920391,0.039242897],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.98974633,0.0064184624,0.00038451797,0.00074153417,0.0016807858,0.0010282828],"domain_scores_gemma":[0.9808544,0.0072807353,0.0016369169,0.0016081068,0.00329602,0.0053238887],"candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.015439918,0.00053159456,0.0006742053,0.0027370448,0.0068957875,0.018701429,0.00484037,0.009408385,0.015066393],"category_scores_gemma":[0.028261252,0.0005249725,0.0013670224,0.0029887483,0.00942604,0.022352831,0.012423241,0.0054907296,0.0029582784],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003117162,0.00013998075,0.000661917,0.00004368504,0.000010515371,0.00013739771,0.0020650877,0.0029294423,0.00008167201,0.98398846,0.0025202413,0.007390537],"study_design_scores_gemma":[0.00012720603,0.00010718721,0.0003123745,0.00014120701,0.000026161555,0.000221304,0.009304118,0.04554133,0.00017114377,0.89335984,0.05064391,0.000044227378],"about_ca_topic_score_codex":0.0025080242,"about_ca_topic_score_gemma":0.0033525946,"teacher_disagreement_score":0.9951596,"about_ca_system_score_codex":0.005590251,"about_ca_system_score_gemma":0.015226525,"threshold_uncertainty_score":0.081655025},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4379599895","doi":"10.1111/hex.13779","title":"Exploring the impact of engagement in mental health and substance use research: A scoping review and thematic analysis","year":2023,"lang":"en","type":"review","venue":"Health Expectations","topic":"Mental Health and Patient Involvement","field":"Health Professions","cited_by":49,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Centre for Addiction and Mental Health; Trillium Health Centre; University of Toronto","funders":"Canadian Institutes of Health Research; Institute of Neurosciences, Mental Health and Addiction; Centre for Addiction and Mental Health","keywords":"Thematic analysis; Mental health; Substance use; Psychology; Thematic map; Applied psychology; Qualitative research; Sociology; Psychotherapist; Psychiatry; Geography; Social science","score_opus":0.9341418747535614,"score_gpt":0.6724996206695402,"score_spread":0.2616422540840212,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4379599895","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.035544485,0.87833834,0.030579453,0.01325198,0.0017221061,0.023088874,0.0025316677,0.00010395072,0.014839235],"genre_scores_gemma":[0.14063217,0.7622586,0.049012557,0.0025688703,0.00035498597,0.041776177,0.0018000841,0.00009403761,0.0015025241],"study_design_codex":"systematic_review","study_design_gemma":"qualitative","domain_scores_codex":[0.89183015,0.064611435,0.02521691,0.0033006982,0.013311745,0.0017290413],"domain_scores_gemma":[0.751283,0.20830867,0.014261338,0.0046692523,0.020421792,0.0010559814],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.13407248,0.0018128416,0.0048515247,0.053179167,0.004370419,0.009888171,0.0033710585,0.003284478,0.0024001023],"category_scores_gemma":[0.23749739,0.0014748415,0.005956671,0.054161586,0.0057521607,0.011133074,0.010507195,0.002660085,0.0003495089],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018929239,0.000066103974,0.004279234,0.5616811,0.0021415676,0.0008383277,0.1178575,0.0006343407,0.0012231552,0.017851437,0.007252607,0.2859853],"study_design_scores_gemma":[0.000043777804,0.00008568826,0.0026968834,0.8738708,0.003286636,0.00047461886,0.051617164,0.00036963832,0.0005962449,0.0067355507,0.06015235,0.000070620685],"about_ca_topic_score_codex":0.0064750197,"about_ca_topic_score_gemma":0.011109004,"teacher_disagreement_score":0.8659275,"about_ca_system_score_codex":0.013546975,"about_ca_system_score_gemma":0.04134817,"threshold_uncertainty_score":0.7090515},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"methods","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"methods","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4381480929","doi":"10.2196/preprints.49521","title":"Caregiver needs and perception of data sharing for research through mHealth in pediatric asthma: a cross-sectional survey (Preprint)","year":2023,"lang":"en","type":"preprint","venue":"","topic":"Mobile Health and mHealth Applications","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Asthma; mHealth; Likert scale; Medicine; Descriptive statistics; Family medicine; Cross-sectional study; Medical education; Psychology; Nursing; Psychological intervention; Developmental psychology","score_opus":0.5963285604469837,"score_gpt":0.6088969951805573,"score_spread":0.012568434733573608,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4381480929","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9989511,0.00010112285,0.000039152696,0.00017222736,0.000004182449,0.000031056003,0.0004308631,0.0000013901462,0.00026884975],"genre_scores_gemma":[0.9987669,0.00028488162,0.00018834915,0.0001760946,0.0000062621825,0.0000999427,0.00035011212,0.0000014295529,0.00012594399],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99832064,0.0005234063,0.000408069,0.00014364767,0.00036598468,0.00023819858],"domain_scores_gemma":[0.99379325,0.0019427503,0.0023799478,0.00017950498,0.0010041723,0.0007003549],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00375215,0.00013689947,0.00030073733,0.0007618629,0.00083150883,0.00096096884,0.00030704236,0.00046947814,0.0020268653],"category_scores_gemma":[0.008536893,0.00039561818,0.00053915166,0.0013645855,0.00029744333,0.0012033347,0.00093299284,0.0006833363,0.0002529964],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000025761943,0.0000999607,0.9907741,0.00008843021,0.000017348746,0.00008763133,0.0061639263,0.000017043478,0.00009873348,0.000022282964,0.00046786945,0.002136983],"study_design_scores_gemma":[0.0000048782913,0.00017457416,0.97950417,0.000088798864,0.000014556028,0.00017391812,0.019367324,0.0000742777,0.000058032303,0.00001214367,0.00052083284,0.000006604483],"about_ca_topic_score_codex":0.015161249,"about_ca_topic_score_gemma":0.019389937,"teacher_disagreement_score":0.015161249,"about_ca_system_score_codex":0.00081399694,"about_ca_system_score_gemma":0.0010647101,"threshold_uncertainty_score":0.030146003},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4384662027","doi":"10.21203/rs.3.rs-3155064/v1","title":"Developing Global Open Access COVID-19 Education for Frontline Healthcare Workers","year":2023,"lang":"en","type":"preprint","venue":"Research Square","topic":"Dental Research and COVID-19","field":"Dentistry","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"Health care; Massive open online course; Medical education; Coronavirus disease 2019 (COVID-19); Health equity; Distance education; Knowledge translation; Population; Medicine; Public relations; Business; Nursing; Political science; Psychology; Computer science; Knowledge management; Pedagogy; Environmental health","score_opus":0.512217479331562,"score_gpt":0.6500916815296953,"score_spread":0.1378742021981333,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4384662027","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.68017066,0.0020335237,0.035500266,0.10522116,0.0022002172,0.0048417915,0.00090366165,0.001045996,0.16808276],"genre_scores_gemma":[0.8827806,0.0021160946,0.075883865,0.011521185,0.0004935107,0.0019085587,0.00073110603,0.000087604676,0.024477351],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.996792,0.0013321354,0.00011032848,0.00018334926,0.00057362014,0.0010086615],"domain_scores_gemma":[0.98899466,0.0019877884,0.0007471187,0.00045202667,0.0013345067,0.006484011],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.010480014,0.00038171068,0.00018109835,0.0006730603,0.0018598852,0.003185754,0.0014720416,0.0015222526,0.021227485],"category_scores_gemma":[0.01276992,0.00020304252,0.00042031307,0.00028471902,0.00109754,0.003063373,0.011343449,0.0020213076,0.0021114824],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002802568,0.009315542,0.055898692,0.002033516,0.00004350215,0.00065883313,0.014058606,0.0012476282,0.0046063173,0.01480029,0.09640903,0.80064774],"study_design_scores_gemma":[0.0006701137,0.0058947965,0.1698106,0.0069563678,0.000105537656,0.00088950474,0.09060622,0.0056253904,0.012129581,0.028332908,0.6788184,0.00016061691],"about_ca_topic_score_codex":0.0041513364,"about_ca_topic_score_gemma":0.008697243,"teacher_disagreement_score":0.99852794,"about_ca_system_score_codex":0.0027559104,"about_ca_system_score_gemma":0.020493198,"threshold_uncertainty_score":0.071012974},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4385278485","doi":"10.31234/osf.io/758dx","title":"L’importance de la science ouverte en recherche en psychologie","year":2023,"lang":"fr","type":"preprint","venue":"","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Humanities; Political science; Philosophy","score_opus":0.9349024534882003,"score_gpt":0.6576264802495904,"score_spread":0.27727597323861,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4385278485","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":"methods","domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014198883,0.2736445,0.26724288,0.35371807,0.044597097,0.00308842,0.0017023294,0.0010104203,0.04079735],"genre_scores_gemma":[0.3499107,0.12918131,0.34910053,0.11523169,0.024560628,0.012118333,0.0014237054,0.0018377069,0.016635463],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.2416532,0.61392915,0.049774114,0.026851619,0.065235935,0.0025560039],"domain_scores_gemma":[0.058181543,0.78516906,0.022144033,0.09486711,0.03794092,0.0016974607],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.6146902,0.002156407,0.007556015,0.017746978,0.0077817435,0.030402044,0.006866789,0.01422738,0.0088473465],"category_scores_gemma":[0.8184905,0.0034199418,0.005527123,0.016548986,0.05508915,0.039564654,0.01668874,0.022099493,0.0042011547],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0011259257,0.00024669405,0.009133912,0.051069193,0.0056537623,0.0008350508,0.046068437,0.0015269665,0.0023023798,0.57035834,0.04137886,0.27030054],"study_design_scores_gemma":[0.00044719497,0.0006890368,0.0062601254,0.045323774,0.0022286887,0.001169677,0.009125718,0.0022533282,0.0031339726,0.60474163,0.324138,0.0004889331],"about_ca_topic_score_codex":0.0071280063,"about_ca_topic_score_gemma":0.006439887,"teacher_disagreement_score":0.9931332,"about_ca_system_score_codex":0.012807606,"about_ca_system_score_gemma":0.043948255,"threshold_uncertainty_score":0.47515565},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4386830764","doi":"10.5334/dsj-2023-035","title":"Umbrella Data Management Plans to Integrate FAIR Data: Lessons From the ISIDORe and BY-COVID Consortia for Pandemic Preparedness","year":2023,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Institute for Research in Immunology and Cancer","funders":"Biotechnology and Biological Sciences Research Council","keywords":"Preparedness; Data management; Computer science; Multidisciplinary approach; Plan (archaeology); Download; Process (computing); Data management plan; Data science; Process management; Knowledge management; Engineering management; World Wide Web; Business; Engineering; Political science; Geography; Database","score_opus":0.43034568833391706,"score_gpt":0.48834636250226554,"score_spread":0.058000674168348476,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386830764","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013593497,0.004080885,0.55167115,0.36067122,0.0039092596,0.002617746,0.0017507434,0.0032384417,0.05846701],"genre_scores_gemma":[0.05936563,0.0019945945,0.8888759,0.022858659,0.0010506524,0.0023331714,0.004062475,0.0019117341,0.017547186],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.76529217,0.12479608,0.026450597,0.014907302,0.058572818,0.009981086],"domain_scores_gemma":[0.504484,0.18143174,0.012480418,0.19348618,0.080146484,0.02797123],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.48550367,0.0016429588,0.0018081985,0.007506551,0.015122446,0.060581878,0.01693589,0.015645757,0.0070758355],"category_scores_gemma":[0.35521564,0.0026188493,0.0043473593,0.012535607,0.03126967,0.06660323,0.03338017,0.026089603,0.003295065],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009111515,0.00022393055,0.002087382,0.00038560922,0.00006440939,0.00028199158,0.009332332,0.0029758597,0.00071278005,0.87653524,0.052421864,0.05488738],"study_design_scores_gemma":[0.000104203515,0.0001451129,0.0016060306,0.0016447548,0.000043169104,0.0003428887,0.0060958597,0.0056121503,0.0015523643,0.38656014,0.59606844,0.00022483438],"about_ca_topic_score_codex":0.040897354,"about_ca_topic_score_gemma":0.020341108,"teacher_disagreement_score":0.9830641,"about_ca_system_score_codex":0.02448364,"about_ca_system_score_gemma":0.11329206,"threshold_uncertainty_score":0.6344656},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4386882067","doi":"10.32920/24171420.v1","title":"Open data and injuries in urban areas—A spatial analytical framework of Toronto using machine learning and spatial regressions","year":2023,"lang":"en","type":"preprint","venue":"","topic":"Urban Transport and Accessibility","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto; Toronto Metropolitan University","funders":"Canadian Institutes of Health Research","keywords":"Metropolitan area; Geography; Injury prevention; Poison control; Occupational safety and health; Suicide prevention; Human factors and ergonomics; Scale (ratio); Public health; Environmental health; Medicine; Cartography","score_opus":0.09782308978445509,"score_gpt":0.41524915415466185,"score_spread":0.31742606437020676,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4386882067","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.80032337,0.0070183957,0.1374714,0.010507762,0.00014517766,0.00027121708,0.024215462,0.00042843932,0.019618686],"genre_scores_gemma":[0.9744275,0.0013003768,0.018185562,0.00009362302,0.000043908316,0.00007354662,0.0038144244,0.000034077068,0.0020268874],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99839693,0.00081735005,0.000106919746,0.00027863224,0.0002215304,0.00017872691],"domain_scores_gemma":[0.9932579,0.0038939137,0.0011779804,0.00047752156,0.0008588919,0.00033374835],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024580294,0.0005295055,0.00046596,0.0047114776,0.0011296753,0.0035660467,0.0013517548,0.0005631373,0.0017998532],"category_scores_gemma":[0.012072009,0.00033002748,0.0009449211,0.008416781,0.0018714321,0.0012317314,0.0024813474,0.0007301439,0.00018132731],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010993523,0.00007593908,0.5184304,0.00036104448,0.00049496006,0.0010343156,0.003770724,0.2819684,0.00034346257,0.138701,0.008150845,0.046559054],"study_design_scores_gemma":[0.000013996747,0.000046754034,0.18557392,0.00032635423,0.00017177268,0.000141841,0.006142093,0.75397176,0.00027337763,0.034584764,0.018667126,0.00008630044],"about_ca_topic_score_codex":0.8794021,"about_ca_topic_score_gemma":0.81557703,"teacher_disagreement_score":0.1205979,"about_ca_system_score_codex":0.01228266,"about_ca_system_score_gemma":0.00718864,"threshold_uncertainty_score":0.24261618},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4387874040","doi":"10.12688/f1000research.140556.1","title":"Open science interventions proposed or implemented to assess researcher impact: a scoping review","year":2023,"lang":"en","type":"review","venue":"F1000Research","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; University of Waterloo; University of Toronto; Ottawa Hospital; University of British Columbia","funders":"","keywords":"Rigour; Open science; Psychological intervention; Promotion (chess); Biomedicine; Political science; Psychology; Engineering ethics; Medical education; Management science; Medicine; Engineering; Nursing","score_opus":0.9934818852275966,"score_gpt":0.8386701618145955,"score_spread":0.1548117234130011,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387874040","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0073373704,0.8934354,0.015680129,0.019669678,0.005186215,0.041073885,0.0033284365,0.0002712481,0.014017606],"genre_scores_gemma":[0.037285358,0.87087506,0.031592306,0.0054514855,0.0007894509,0.05065779,0.0018756494,0.00010703672,0.0013658243],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.8247585,0.08804914,0.05121044,0.0046560983,0.028803105,0.0025226837],"domain_scores_gemma":[0.4984459,0.38252226,0.038516615,0.010996891,0.06734318,0.0021752436],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.1752923,0.0023021717,0.0050432174,0.029918991,0.0044240453,0.011476887,0.004704738,0.0062170215,0.01071201],"category_scores_gemma":[0.4426066,0.0025726527,0.008951655,0.030061131,0.004080265,0.013641788,0.008997823,0.0041931528,0.0022220216],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036723816,0.0000993879,0.0014988226,0.7671934,0.0020320825,0.00018473133,0.003725554,0.00034534576,0.00045811484,0.003975935,0.014019624,0.20609984],"study_design_scores_gemma":[0.000097075754,0.00009197226,0.0011610244,0.9562555,0.0031571072,0.00008985149,0.0015070502,0.00013024706,0.00028614965,0.0011595406,0.036034208,0.000030166397],"about_ca_topic_score_codex":0.006713164,"about_ca_topic_score_gemma":0.017222257,"teacher_disagreement_score":0.9952953,"about_ca_system_score_codex":0.01455235,"about_ca_system_score_gemma":0.07357855,"threshold_uncertainty_score":0.92704535},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"incentives","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"evaluation","study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4387874083","doi":"10.1186/s12916-023-03113-0","title":"A 10-year update to the principles for clinical trial data sharing by pharmaceutical companies: perspectives based on a decade of literature and policies","year":2023,"lang":"en","type":"article","venue":"BMC Medicine","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":25,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto; St. Michael's Hospital","funders":"Cancer Council South Australia; National Health and Medical Research Council; Health Canada","keywords":"Data sharing; Medicine; Transparency (behavior); Scope (computer science); Multidisciplinary approach; Clinical trial; Health care; Data management; Process (computing); Pharmaceutical industry; Knowledge management; Public relations; Alternative medicine; Computer science; Political science","score_opus":0.8122559443649875,"score_gpt":0.6759175419285655,"score_spread":0.13633840243642203,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387874083","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001049471,0.018856551,0.026441587,0.927698,0.016044568,0.00040490663,0.000321586,0.00031475784,0.008868608],"genre_scores_gemma":[0.020692471,0.02210249,0.106216654,0.8223746,0.014565985,0.002426184,0.00090226275,0.00058973813,0.010129574],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.49785927,0.23319982,0.10480847,0.023291282,0.12267643,0.018164776],"domain_scores_gemma":[0.25078657,0.50266725,0.03294639,0.04999297,0.13341309,0.030193767],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.53882414,0.0023067817,0.004502766,0.006403457,0.010864748,0.039563295,0.015346645,0.06509379,0.0037410052],"category_scores_gemma":[0.53212154,0.005286592,0.008794346,0.008392496,0.031828526,0.039777335,0.029810466,0.123947114,0.0037477377],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017415544,0.0003746245,0.0012687466,0.002570984,0.0002452196,0.0006669586,0.009845258,0.0016918816,0.0010188541,0.4762858,0.40857482,0.09728268],"study_design_scores_gemma":[0.00012589988,0.00018875857,0.0014068522,0.007733829,0.00015758604,0.00057319045,0.0014063625,0.0009782962,0.0007639577,0.06848662,0.9178608,0.0003177708],"about_ca_topic_score_codex":0.02564804,"about_ca_topic_score_gemma":0.022632191,"teacher_disagreement_score":0.98465335,"about_ca_system_score_codex":0.037028696,"about_ca_system_score_gemma":0.17220889,"threshold_uncertainty_score":0.568712},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","scholarly_communication","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4387925098","doi":"10.1016/j.geoforum.2023.103881","title":"Implementing EDI across a large formal research network: Contributing to equitable and sustainable water solutions for a changing climate","year":2023,"lang":"en","type":"article","venue":"Geoforum","topic":"Indigenous Studies and Ecology","field":"Health Professions","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Global Institute for Water Security; University of Saskatchewan","funders":"Canada First Research Excellence Fund","keywords":"Operationalization; Equity (law); Sociology; Business; Public relations; Knowledge management; Political science; Computer science","score_opus":0.09833064087190231,"score_gpt":0.47514370360701985,"score_spread":0.37681306273511755,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4387925098","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07511203,0.0029890975,0.1906847,0.5567561,0.0018736967,0.0035828708,0.00013516177,0.0008752216,0.16799113],"genre_scores_gemma":[0.65727663,0.0028982912,0.24681862,0.05122921,0.0005057344,0.003791833,0.00027035366,0.0004011122,0.036808204],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.856555,0.10230575,0.004196985,0.007212227,0.015408624,0.014321459],"domain_scores_gemma":[0.83034456,0.08236163,0.009666336,0.018301902,0.022944327,0.036381233],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.23855019,0.0008439839,0.0007366637,0.0040232977,0.021495936,0.024108259,0.008457943,0.008590232,0.007850535],"category_scores_gemma":[0.13333935,0.0013278864,0.0016383256,0.0031618536,0.027507264,0.028679648,0.06482163,0.01565303,0.0015363048],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010029421,0.0010216436,0.010841469,0.0013104911,0.00010196303,0.0019288303,0.15293096,0.00284737,0.0031851209,0.45482042,0.041845966,0.32906547],"study_design_scores_gemma":[0.00012785371,0.00056593405,0.0072535304,0.0019948676,0.00006360465,0.0006995708,0.19283321,0.0038942979,0.0026032955,0.26649785,0.5232087,0.00025738226],"about_ca_topic_score_codex":0.019531418,"about_ca_topic_score_gemma":0.03989894,"teacher_disagreement_score":0.99154204,"about_ca_system_score_codex":0.043057915,"about_ca_system_score_gemma":0.19566907,"threshold_uncertainty_score":0.93900317},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","sts"],"domain":"incentives","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4388006999","doi":"10.1136/bmjebm-2023-112475","title":"How to use the regulatory data from Health Canada for secondary analyses on new drugs, biologics and vaccines","year":2023,"lang":"en","type":"article","venue":"BMJ evidence-based medicine","topic":"Clinical practice guidelines implementation","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Dalhousie University","funders":"Dalhousie University","keywords":"Medicine; Public health; Internet privacy; Business; Computer science","score_opus":0.7468377814715961,"score_gpt":0.5827913881962639,"score_spread":0.16404639327533221,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388006999","genre_codex":"commentary","genre_gemma":"methods","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0041891797,0.01403573,0.107490554,0.3253722,0.0131196715,0.0942071,0.28931007,0.012661047,0.13961443],"genre_scores_gemma":[0.026214005,0.013798913,0.7113831,0.07368306,0.003054955,0.09133817,0.04629453,0.0052906754,0.028942572],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.63187444,0.18025571,0.083856896,0.007515632,0.09033855,0.006158773],"domain_scores_gemma":[0.19900413,0.40354058,0.04280556,0.07751775,0.26811874,0.009013189],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.31042376,0.0023337689,0.0067219925,0.029423425,0.004439494,0.01978562,0.0068896315,0.009567846,0.06469373],"category_scores_gemma":[0.73580444,0.005446836,0.008098961,0.028549466,0.0054984814,0.011891642,0.0077721034,0.0076710926,0.04787107],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00056085055,0.00012489373,0.0025112503,0.008774149,0.00047199466,0.00016456771,0.0014763928,0.0008049454,0.0004221741,0.005486634,0.83598745,0.14321466],"study_design_scores_gemma":[0.0032273687,0.0001226986,0.011424112,0.02795721,0.0007559967,0.00016473909,0.0014733404,0.002548979,0.0022100299,0.019805923,0.929512,0.0007974498],"about_ca_topic_score_codex":0.52047825,"about_ca_topic_score_gemma":0.6435565,"teacher_disagreement_score":0.9813545,"about_ca_system_score_codex":0.018645542,"about_ca_system_score_gemma":0.20340805,"threshold_uncertainty_score":0.9646915},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"methods","about_ca_system":true,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"methods","study_design":"design_other","genre":"methods","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W4388947055","doi":"10.32920/24625011","title":"Examining the Value of Geospatial Open Data","year":2023,"lang":"en","type":"preprint","venue":"","topic":"Geographic Information Systems Studies","field":"Social Sciences","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Geospatial analysis; Open data; Open government; Transparency (behavior); Volunteered geographic information; Public participation GIS; Thematic map; Accountability; Data science; Geographic information system; Government (linguistics); Computer science; Geography; World Wide Web; Political science; GIS and public health; Computer security; Cartography","score_opus":0.3887944497600715,"score_gpt":0.42343375981140424,"score_spread":0.03463931005133275,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388947055","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.85291547,0.0015886661,0.0041904445,0.02681404,0.000099925615,0.00008382447,0.0016265712,0.000041618026,0.11263928],"genre_scores_gemma":[0.99639374,0.00041898518,0.0014693057,0.00026545193,0.000043495016,0.000019571433,0.0003231537,0.000021934167,0.0010443105],"study_design_codex":"observational","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.97384983,0.013980245,0.0007296959,0.0011326611,0.008842952,0.0014646908],"domain_scores_gemma":[0.6357152,0.2905421,0.021390956,0.013877546,0.0332668,0.005207393],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.025948161,0.00024567544,0.00028271895,0.007929302,0.002920556,0.0118995495,0.0016048064,0.0013141431,0.0041769845],"category_scores_gemma":[0.17754741,0.00030373866,0.00036614126,0.016727855,0.007309148,0.012032095,0.004810684,0.002080438,0.00040829097],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00031145292,0.00029599346,0.5876563,0.00055937096,0.00017477892,0.0006575456,0.030320128,0.003044397,0.00073534966,0.21921995,0.009540185,0.14748466],"study_design_scores_gemma":[0.000051532894,0.00031488933,0.45004117,0.0017626837,0.00028860438,0.0007000427,0.22882332,0.022375494,0.0035363296,0.15669678,0.13523398,0.00017516481],"about_ca_topic_score_codex":0.066456534,"about_ca_topic_score_gemma":0.09202137,"teacher_disagreement_score":0.9983952,"about_ca_system_score_codex":0.009460058,"about_ca_system_score_gemma":0.007060863,"threshold_uncertainty_score":0.13722867},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W4388951762","doi":"10.1139/facets-2022-0208","title":"Evaluating prospective study registration and result reporting of trials conducted in Canada from 2009 to 2019","year":2023,"lang":"en","type":"article","venue":"FACETS","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"Women's College Hospital; University of Toronto; University of British Columbia; Ottawa Hospital; Institute of Health Services and Policy Research; University of Ottawa","funders":"Faculty of Medicine, University of Ottawa; University of Ottawa","keywords":"Clinical trial; Medicine; Trial registration; Family medicine; Odds; Odds ratio; Demographics; MEDLINE; Demography; Internal medicine; Logistic regression","score_opus":0.9090883366815484,"score_gpt":0.6316454442496872,"score_spread":0.27744289243186127,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4388951762","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reporting","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reporting","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.12638368,0.37045822,0.04371982,0.033803016,0.0047566635,0.081704535,0.29832822,0.0016616572,0.03918426],"genre_scores_gemma":[0.67392886,0.072987795,0.087908894,0.006889578,0.0010142531,0.06372385,0.089669205,0.0006718255,0.00320577],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.31497604,0.27291885,0.28455263,0.016666234,0.10574211,0.00514413],"domain_scores_gemma":[0.10094097,0.29070157,0.23322976,0.08332594,0.28656963,0.005232218],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.57813776,0.0017511894,0.007093513,0.02315532,0.00424675,0.010880891,0.007402502,0.0023691931,0.00392165],"category_scores_gemma":[0.7759013,0.0029314726,0.008394504,0.04957392,0.0037886256,0.004609082,0.00535722,0.0031686784,0.00091981125],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.011501506,0.00024145906,0.29608336,0.2554449,0.057698548,0.00047236585,0.0056030136,0.0036031753,0.0009750225,0.0070437277,0.0842223,0.2771107],"study_design_scores_gemma":[0.006721699,0.001669709,0.62831986,0.13403279,0.058099106,0.00053684483,0.001898546,0.0044572433,0.0038567975,0.0040110312,0.15533943,0.00105695],"about_ca_topic_score_codex":0.6129137,"about_ca_topic_score_gemma":0.7133364,"teacher_disagreement_score":0.9476627,"about_ca_system_score_codex":0.05233729,"about_ca_system_score_gemma":0.2563179,"threshold_uncertainty_score":0.7787318},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W4390636406","doi":"10.5206/mt.v3i4.17132","title":"Articles, software, data: An Open Science ethological study","year":2024,"lang":"en","type":"article","venue":"Maple Transactions","topic":"Big Data and Business Intelligence","field":"Business, Management and Accounting","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"Erasmus+; Universidad de Cantabria; Université Gustave Eiffel; European Commission","keywords":"Open science; Context (archaeology); Data science; Relevance (law); Computer science; Software; Open data; Work (physics); Open research; Citizen science; Knowledge management; World Wide Web; Engineering ethics; Political science; Engineering","score_opus":0.2815450316315971,"score_gpt":0.40026884134982843,"score_spread":0.11872380971823132,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390636406","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24988392,0.1411643,0.05629985,0.18941149,0.002940522,0.0009768029,0.001441121,0.0001460803,0.35773596],"genre_scores_gemma":[0.9124544,0.039619252,0.02435799,0.008452499,0.0022905509,0.0010783691,0.00042487666,0.00020124376,0.011120857],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9286637,0.048180025,0.004468615,0.004545048,0.012718487,0.0014241056],"domain_scores_gemma":[0.7083541,0.2524516,0.016815811,0.00831261,0.010370572,0.0036953266],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.052148066,0.00050804706,0.0008421996,0.032465234,0.012134565,0.031693503,0.001448094,0.004105324,0.006212369],"category_scores_gemma":[0.07687582,0.00047829046,0.0005673638,0.04031971,0.045048833,0.036762483,0.009525486,0.003362259,0.00087951944],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000067952635,0.00010855052,0.009749597,0.0027608492,0.00003800817,0.00063842785,0.17801455,0.00016486946,0.0004845406,0.71092033,0.006881157,0.09017125],"study_design_scores_gemma":[0.000024364364,0.00013289526,0.010762683,0.007911679,0.00004866717,0.0013472667,0.34609088,0.00047761903,0.00048873003,0.22700615,0.4056403,0.00006883662],"about_ca_topic_score_codex":0.0016901521,"about_ca_topic_score_gemma":0.0023528691,"teacher_disagreement_score":0.9985519,"about_ca_system_score_codex":0.0074753817,"about_ca_system_score_gemma":0.010503919,"threshold_uncertainty_score":0.2757886},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"evaluation","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4390772842","doi":"10.59942/2995-9063.1002","title":"More Obstacles for the Graduate Student Author: Open Access ETDs Trigger Plagiarism Detectors","year":2023,"lang":"en","type":"article","venue":"Journal of Graduate Librarianship","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Suspect; Inclusion (mineral); Plagiarism detection; Publishing; Institution; Library science; Graduate students; Open source software; World Wide Web; Computer science; Software; Political science; Medical education; Sociology; Medicine; Law","score_opus":0.7616027830279153,"score_gpt":0.5593803035575363,"score_spread":0.20222247947037897,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390772842","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.52868915,0.012012149,0.09049807,0.2530441,0.012447279,0.0010447678,0.0018276494,0.013363371,0.08707346],"genre_scores_gemma":[0.84039277,0.004867411,0.06875364,0.04367047,0.005519369,0.0003288432,0.0016361829,0.0039882367,0.03084316],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.88927054,0.042189855,0.016485142,0.008490386,0.039991666,0.0035723646],"domain_scores_gemma":[0.3443753,0.32199782,0.12092231,0.07414529,0.11739851,0.02116067],"candidate_categories":["metaresearch","open_science","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.06922171,0.0006302014,0.0010261739,0.007655305,0.00766119,0.02310465,0.0034666918,0.006035813,0.011123964],"category_scores_gemma":[0.44484487,0.0011192029,0.0010909624,0.009786791,0.0056934794,0.017513447,0.0081822155,0.0052465606,0.009436178],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00077519385,0.00066348724,0.2824981,0.002576725,0.00022806943,0.007026845,0.07165335,0.00075224775,0.012929831,0.02134149,0.19027367,0.40928096],"study_design_scores_gemma":[0.00013050894,0.0008258907,0.09468864,0.0036840432,0.00036707358,0.01341884,0.06719173,0.004607835,0.024629096,0.027176313,0.76243925,0.00084075355],"about_ca_topic_score_codex":0.0026343937,"about_ca_topic_score_gemma":0.0043286323,"teacher_disagreement_score":0.99653333,"about_ca_system_score_codex":0.0052701156,"about_ca_system_score_gemma":0.008922656,"threshold_uncertainty_score":0.3660838},"labels":[{"model":"gemma","categories":["open_science","research_integrity"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["research_integrity","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":"W4390817161","doi":"10.1038/s41597-023-02689-9","title":"The Pelagic Species Trait Database, an open data resource to support trait-based ocean research","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Marine and fisheries research","field":"Environmental Science","cited_by":9,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"Government of Canada; Marine Environmental Observation Prediction and Response Network","keywords":"Pelagic zone; Trait; Food web; Ecology; Biology; Habitat; Predation; Apex predator; Population; Invertebrate; Computer science","score_opus":0.2983281128450758,"score_gpt":0.41775939093007336,"score_spread":0.11943127808499754,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390817161","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012301053,0.00025836666,0.0013210912,0.000064218715,0.000025405403,0.000041611864,0.9947805,0.0010712699,0.0012074576],"genre_scores_gemma":[0.0026193338,0.00020662592,0.0038334075,0.00006451305,0.0000101156265,0.00016088589,0.9925184,0.00025872048,0.00032797156],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980286,0.00022755511,0.0006283292,0.00049229735,0.00047759025,0.00014572752],"domain_scores_gemma":[0.9900372,0.0031270247,0.001984459,0.001763738,0.001947891,0.0011396293],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0026110266,0.0014909968,0.0015008014,0.0121413125,0.00076135056,0.0022528514,0.0024619645,0.0015555538,0.018984975],"category_scores_gemma":[0.016202688,0.00065159664,0.00090278586,0.014107444,0.00041356916,0.002638421,0.00437892,0.0015734492,0.01802951],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00062067993,0.00013001985,0.042357408,0.00894867,0.00054951577,0.00053643854,0.00058894034,0.0016826516,0.005262182,0.007573778,0.8761247,0.055625066],"study_design_scores_gemma":[0.00019725985,0.00006285862,0.04938746,0.0011930593,0.00019093408,0.00027735924,0.0003404143,0.001991402,0.0019787173,0.005161374,0.9390916,0.00012746333],"about_ca_topic_score_codex":0.010048881,"about_ca_topic_score_gemma":0.019602256,"teacher_disagreement_score":0.99753803,"about_ca_system_score_codex":0.0009515451,"about_ca_system_score_gemma":0.0032348623,"threshold_uncertainty_score":0.06351107},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4390842479","doi":"10.1093/gigascience/giad114","title":"Open Data Governance at the Canadian Open Neuroscience Platform (CONP): From the Walled Garden to the Arboretum","year":2024,"lang":"en","type":"article","venue":"GigaScience","topic":"Research Data Management Practices","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University and Génome Québec Innovation Centre; McGill University Health Centre; Terry Fox Research Institute; University of Toronto; Montreal Neurological Institute and Hospital; McGill University","funders":"Fondation Brain Canada","keywords":"Open science; Open data; Citizen science; Data science; World Wide Web; Computer science; Neuroscience; Biology; Botany; Physics","score_opus":0.2758934476905706,"score_gpt":0.4059608317831537,"score_spread":0.13006738409258306,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390842479","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.054321937,0.0038954546,0.100488774,0.47280136,0.002290676,0.0008821435,0.0016604803,0.0016726316,0.36198658],"genre_scores_gemma":[0.61016023,0.006595426,0.15218142,0.054944996,0.0006787501,0.000996824,0.0017893022,0.0015525502,0.17110042],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.96783745,0.009498008,0.00087337097,0.0034377878,0.013709793,0.004643628],"domain_scores_gemma":[0.9415204,0.016863761,0.0021760068,0.009648585,0.0146390945,0.015152194],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.04031665,0.0005412311,0.00045630048,0.0027535183,0.029753968,0.02152114,0.0042372937,0.0037454471,0.00897362],"category_scores_gemma":[0.05097666,0.0007271258,0.00068694283,0.006308781,0.028908445,0.013106115,0.021376112,0.0070896246,0.0012679796],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012205011,0.000047220787,0.0042922427,0.00017612547,0.000041468473,0.0011786581,0.03874458,0.0011768733,0.0013633689,0.64146465,0.19185254,0.11954016],"study_design_scores_gemma":[0.00002379428,0.000030334597,0.0039861454,0.00037538834,0.000014946749,0.00032220583,0.015413925,0.0011535631,0.000613559,0.09798659,0.8799065,0.00017304378],"about_ca_topic_score_codex":0.86823547,"about_ca_topic_score_gemma":0.9310411,"teacher_disagreement_score":0.9957627,"about_ca_system_score_codex":0.07636572,"about_ca_system_score_gemma":0.3260282,"threshold_uncertainty_score":0.55407465},"labels":[{"model":"gemma","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"medium"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4390858342","doi":"10.1016/j.jclinepi.2024.111263","title":"Accessibility of clinical study reports supporting medicine approvals: a cross-sectional evaluation","year":2024,"lang":"en","type":"article","venue":"Journal of Clinical Epidemiology","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Toronto; St. Michael's Hospital","funders":"Cancer Council South Australia; National Health and Medical Research Council; Medical Research Council","keywords":"Cross-sectional study; Medicine; MEDLINE; Family medicine; Environmental health; Pathology; Political science","score_opus":0.9754818601916025,"score_gpt":0.808792759575731,"score_spread":0.1666891006158715,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4390858342","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reporting","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reporting","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93834245,0.006754576,0.0070952293,0.0017452681,0.0001649198,0.012232247,0.024070304,0.00020767587,0.00938737],"genre_scores_gemma":[0.97166926,0.0020296122,0.00682415,0.0011321671,0.00020699762,0.011653448,0.005538223,0.00008911563,0.0008570592],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.74479365,0.11024352,0.0860925,0.01423128,0.04127329,0.0033658277],"domain_scores_gemma":[0.27860373,0.34986776,0.28417212,0.031163488,0.050736543,0.005456316],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.20798522,0.00085672596,0.0016736167,0.010939985,0.0018717091,0.0042125005,0.0026107074,0.002067605,0.008296435],"category_scores_gemma":[0.3895038,0.0013410476,0.0035216142,0.0094826445,0.0030042615,0.0080737835,0.006473096,0.0022193708,0.0024959785],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018165221,0.00067936856,0.96643806,0.0030516011,0.0010556888,0.00018689292,0.0044099838,0.00021235745,0.00024339683,0.000608246,0.0026539657,0.018644024],"study_design_scores_gemma":[0.0005168702,0.0034788828,0.96208936,0.0024825055,0.0011216742,0.0017442721,0.008188059,0.0009831331,0.00091029075,0.00054923084,0.017781671,0.00015403375],"about_ca_topic_score_codex":0.0013819298,"about_ca_topic_score_gemma":0.0013918151,"teacher_disagreement_score":0.7920148,"about_ca_system_score_codex":0.0022684399,"about_ca_system_score_gemma":0.0040716166,"threshold_uncertainty_score":0.97669524},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","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":"W4392861763","doi":"10.22541/au.171052473.32741331/v1","title":"Fostering diversity, equity, and inclusion in interdisciplinary marine science","year":2024,"lang":"en","type":"preprint","venue":"","topic":"Coastal and Marine Management","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of British Columbia; Fisheries and Oceans Canada","funders":"National Science Foundation","keywords":"Marine research; Geography; Oceanography; Library science; Geology","score_opus":0.032027343601153245,"score_gpt":0.3100130854940937,"score_spread":0.27798574189294045,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392861763","genre_codex":"other","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.27210698,0.016314317,0.0397686,0.09492109,0.0015322512,0.00032312825,0.00014391512,0.00021713726,0.57467246],"genre_scores_gemma":[0.975266,0.0030218235,0.007292741,0.0032324179,0.00046220183,0.00019074901,0.000060557475,0.00006830486,0.010405176],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.98923254,0.005684241,0.00026142685,0.00085572887,0.002270299,0.0016957568],"domain_scores_gemma":[0.98295814,0.0073318062,0.0012855307,0.0015278319,0.001788377,0.0051083514],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.017131882,0.00047366938,0.0004911794,0.0034533923,0.008498223,0.012653292,0.0014355999,0.0020062306,0.010948605],"category_scores_gemma":[0.030154273,0.0002798228,0.00039481078,0.0023306357,0.010691251,0.008084735,0.0557408,0.0022364107,0.0013136338],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021882061,0.0003625083,0.02398564,0.0008427285,0.00012189137,0.0003587349,0.020485783,0.0023491357,0.001449635,0.439264,0.020370835,0.49019018],"study_design_scores_gemma":[0.000048498372,0.00014061092,0.013200788,0.0013360243,0.000054245145,0.00021899794,0.014302477,0.0017318506,0.0012310832,0.82618517,0.14150307,0.00004711309],"about_ca_topic_score_codex":0.002307223,"about_ca_topic_score_gemma":0.005778838,"teacher_disagreement_score":0.9985644,"about_ca_system_score_codex":0.0042618695,"about_ca_system_score_gemma":0.013623226,"threshold_uncertainty_score":0.09060311},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4392914547","doi":"10.1016/j.eclinm.2024.102551","title":"Health data sharing attitudes towards primary and secondary use of data: a systematic review","year":2024,"lang":"en","type":"review","venue":"EClinicalMedicine","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":56,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"CINAHL; PsycINFO; Medicine; Data sharing; MEDLINE; Data extraction; Family medicine; Qualitative research; Health care; Public health; Medical education; Alternative medicine; Psychological intervention; Nursing; Pathology","score_opus":0.9055626380658633,"score_gpt":0.7122511745234683,"score_spread":0.19331146354239503,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4392914547","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0057087983,0.9897169,0.0004051187,0.00062065653,0.00012432583,0.0017027832,0.0009032101,0.000015941214,0.00080222235],"genre_scores_gemma":[0.06272736,0.9291158,0.0020545102,0.0011716175,0.00009055799,0.0040621306,0.0005879683,0.000013056732,0.00017705369],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.9542101,0.019779116,0.015962372,0.0021806648,0.0070511242,0.0008165367],"domain_scores_gemma":[0.8101904,0.1535557,0.023150854,0.0026706285,0.009209341,0.0012231215],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.038419146,0.0013618156,0.007852612,0.013464311,0.0011946731,0.0049097296,0.0025146406,0.0023399063,0.0051394063],"category_scores_gemma":[0.13856879,0.0014638654,0.007713083,0.01785832,0.00221306,0.005420513,0.0033345243,0.0016851096,0.00032719228],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015295109,0.000021560501,0.0012963824,0.97234815,0.005170231,0.00006413816,0.0011386974,0.000040944306,0.00006449828,0.00020102243,0.0006046489,0.018896688],"study_design_scores_gemma":[0.00013223666,0.0001327237,0.003757911,0.9649128,0.022674154,0.00018900739,0.001476626,0.00003849756,0.00010843973,0.00018719093,0.0063602054,0.000030199939],"about_ca_topic_score_codex":0.009513863,"about_ca_topic_score_gemma":0.02458322,"teacher_disagreement_score":0.99748534,"about_ca_system_score_codex":0.007178802,"about_ca_system_score_gemma":0.023970792,"threshold_uncertainty_score":0.20318228},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"systematic_review","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4393558576","doi":"10.5281/zenodo.3826037","title":"Open Access of COVID-19 related publications in the first quarter of 2020: a preliminary study based in PubMed","year":2020,"lang":"en","type":"dataset","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Coronavirus disease 2019 (COVID-19); Quarter (Canadian coin); Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2); 2019-20 coronavirus outbreak; Virology; History; Medicine; Infectious disease (medical specialty)","score_opus":0.2023763604307022,"score_gpt":0.4203691591748092,"score_spread":0.217992798744107,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4393558576","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006323851,0.0006431142,0.00007004847,0.0002406478,0.00004489224,0.00007810344,0.9976265,0.00006741206,0.00059686624],"genre_scores_gemma":[0.0035668134,0.0011980651,0.0008717564,0.0002549045,0.000066603585,0.0009286633,0.99221283,0.000090660986,0.00080971996],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9908456,0.0012517674,0.003828248,0.0014038751,0.001860399,0.0008102341],"domain_scores_gemma":[0.9610286,0.01789204,0.007320157,0.0033993002,0.007892704,0.0024672428],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.0068554683,0.0011515998,0.0024747541,0.020045614,0.0011863619,0.00474048,0.0017103368,0.0017429224,0.059160497],"category_scores_gemma":[0.045795668,0.00074554246,0.0022073446,0.026934242,0.0007776113,0.002611357,0.0032335771,0.0016540342,0.026854884],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0017477948,0.00011993454,0.013281497,0.05132428,0.0007811651,0.00034323655,0.00039446782,0.0003346346,0.0005719927,0.0025953876,0.9069798,0.021525774],"study_design_scores_gemma":[0.0015169247,0.00020905021,0.06472377,0.019458847,0.0014400631,0.000629456,0.0008607374,0.00029902338,0.00069394853,0.0025996258,0.907392,0.00017653489],"about_ca_topic_score_codex":0.010764106,"about_ca_topic_score_gemma":0.022207286,"teacher_disagreement_score":0.99828964,"about_ca_system_score_codex":0.0019938357,"about_ca_system_score_gemma":0.008843873,"threshold_uncertainty_score":0.19791162},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"reporting","study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"design_other","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4398541587","doi":"10.7910/dvn/hvuu6x","title":"Replication Data for: \"The Implications of Cohabitation Between Working Age Children and Parents for Political Opinions\"","year":2023,"lang":"en","type":"dataset","venue":"Harvard Dataverse","topic":"Psychosocial Factors Impacting Youth","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Cohabitation; Replication (statistics); Politics; Psychology; Sociology; Political science; Biology; Law; Virology","score_opus":0.15855909686570022,"score_gpt":0.4182733440481281,"score_spread":0.25971424718242786,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4398541587","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0012154363,0.00006900446,0.0000885175,0.0002565896,0.00008442662,0.00005045057,0.9972187,0.0001522264,0.00086451136],"genre_scores_gemma":[0.0030378208,0.00005809436,0.0004033892,0.0001223105,0.000031057585,0.00043649125,0.9937216,0.000045681092,0.002143599],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978405,0.0005711361,0.00034137935,0.00043874,0.00046874754,0.00033943512],"domain_scores_gemma":[0.99077564,0.0022813538,0.0014189916,0.0019921372,0.0027958741,0.0007360701],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0040664724,0.0012254186,0.0009151624,0.0025827903,0.0011809629,0.0022335853,0.0028071483,0.0020075028,0.053904314],"category_scores_gemma":[0.021852935,0.0006068896,0.0013181155,0.0050971503,0.00056434085,0.0011812936,0.0025333394,0.0021674847,0.03682889],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016889973,0.000052632386,0.0084676165,0.00054065755,0.000072005416,0.000054103897,0.000104151426,0.00016313906,0.000092382405,0.0005116392,0.9872114,0.0025612907],"study_design_scores_gemma":[0.0019685458,0.0001003108,0.10093057,0.00066694414,0.00015708878,0.00018781281,0.00085245393,0.0008454902,0.00055456854,0.0012323615,0.8924211,0.00008274335],"about_ca_topic_score_codex":0.06469579,"about_ca_topic_score_gemma":0.11831653,"teacher_disagreement_score":0.06469579,"about_ca_system_score_codex":0.0017523658,"about_ca_system_score_gemma":0.0038006343,"threshold_uncertainty_score":0.18032789},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4398828514","doi":"10.7910/dvn/tfinur","title":"Replication Data for: Political Land Corruption: Evidence from Malta - the European Union's Smallest member State","year":2017,"lang":"en","type":"dataset","venue":"Harvard Dataverse","topic":"Land Rights and Reforms","field":"Agricultural and Biological Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"World Federation of Science Journalists","funders":"","keywords":"Replication (statistics); Politics; State (computer science); Member state; Language change; Political science; European union; Member states; International trade; Law; Biology; Business; Computer science; Art","score_opus":0.09512606426945908,"score_gpt":0.29974719464620064,"score_spread":0.20462113037674157,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4398828514","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0040799044,0.00018366305,0.00012137857,0.0005184692,0.000050175404,0.00004296779,0.99310946,0.00024918286,0.0016447568],"genre_scores_gemma":[0.006347229,0.00009727434,0.00021940394,0.000092817434,0.000021310785,0.00021471712,0.990928,0.00008242798,0.0019968566],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99797505,0.00041644118,0.00023296136,0.00034204134,0.00064459513,0.0003889099],"domain_scores_gemma":[0.98921704,0.002188918,0.0018199316,0.002322712,0.0035255875,0.00092581793],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002224699,0.0010177832,0.00091772433,0.0054305196,0.0012533547,0.0032699427,0.0021927925,0.0017394852,0.032571208],"category_scores_gemma":[0.016808216,0.00044731397,0.00065003295,0.010150787,0.0009813154,0.0013934844,0.0029976733,0.0018901592,0.023222063],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014355459,0.00006741306,0.010959191,0.00038828477,0.000060019855,0.0000943173,0.00017967013,0.00030329585,0.000099242854,0.0012113389,0.9827983,0.0036953792],"study_design_scores_gemma":[0.00053776195,0.00002880699,0.06286936,0.00050407875,0.000069133515,0.00012762957,0.0010773722,0.00075853657,0.0007245217,0.0012672998,0.9319752,0.0000603389],"about_ca_topic_score_codex":0.10988635,"about_ca_topic_score_gemma":0.1532082,"teacher_disagreement_score":0.10988635,"about_ca_system_score_codex":0.0021520446,"about_ca_system_score_gemma":0.0045864256,"threshold_uncertainty_score":0.21849346},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4399390779","doi":"10.1038/s41597-024-03443-5","title":"The Pan-Canadian Chemical Library: A Mechanism to Open Academic Chemistry to High-Throughput Virtual Screening","year":2024,"lang":"en","type":"article","venue":"Scientific Data","topic":"Computational Drug Discovery Methods","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Winnipeg; University of Alberta; Structural Genomics Consortium; University of Toronto","funders":"National Institute of General Medical Sciences; Genentech; European Federation of Pharmaceutical Industries and Associations; University of Toronto; Ontario Genomics Institute; Merck KGaA; Alliance de recherche numérique du Canada; National Institutes of Health; Ontario Genomics; Genome Canada; McGill University; Bayer; Pfizer; Bristol-Myers Squibb","keywords":"Chemical space; Drug discovery; Workflow; Virtual screening; Computer science; Chemical library; Data science; Cheminformatics; Chemical database; Process (computing); Mechanism (biology); Chemistry; Chemical biology; Resource (disambiguation); Nanotechnology; Small molecule; Database; Physics; Materials science; Computational chemistry; Organic chemistry","score_opus":0.06922418333834772,"score_gpt":0.3512207175254694,"score_spread":0.2819965341871217,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4399390779","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.059777386,0.0046266234,0.6669582,0.0056391796,0.0009583761,0.0068510245,0.04047036,0.06443249,0.15028642],"genre_scores_gemma":[0.135583,0.004342275,0.7690748,0.0025715856,0.00012887575,0.0029512602,0.03862142,0.0032329408,0.0434938],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981862,0.00023552966,0.00006848502,0.00024763177,0.0010404144,0.00022180218],"domain_scores_gemma":[0.9983789,0.0002402283,0.000086367545,0.00042840437,0.0005594063,0.00030674724],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0019952205,0.0010435092,0.0009690819,0.0031210065,0.0018222859,0.0020837192,0.00296503,0.00084665,0.01122046],"category_scores_gemma":[0.004194284,0.0006553793,0.0010656756,0.0045972923,0.0015534298,0.0014775919,0.003858282,0.0019102031,0.0048129284],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0018173648,0.0008281171,0.0038416474,0.0013110221,0.00032146214,0.0008059817,0.0006500719,0.022623237,0.13282844,0.11331404,0.22304715,0.49861145],"study_design_scores_gemma":[0.0005718056,0.0005198029,0.0027207506,0.00012735986,0.00017698007,0.000600948,0.00015474857,0.040460702,0.10229088,0.02007047,0.8320172,0.0002883824],"about_ca_topic_score_codex":0.096684575,"about_ca_topic_score_gemma":0.20846032,"teacher_disagreement_score":0.99703497,"about_ca_system_score_codex":0.003262876,"about_ca_system_score_gemma":0.014673689,"threshold_uncertainty_score":0.19224364},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":true,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W4400303628","doi":"10.1515/9782760634961-088","title":"Autres titres en libre accès aux Presses de l’Université de Montréal","year":2015,"lang":"fr","type":"book-chapter","venue":"Les Presses de l'Université de Montréal eBooks","topic":"Cultural Insights and Digital Impacts","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Art","score_opus":0.06024510160698509,"score_gpt":0.21945110746084157,"score_spread":0.15920600585385647,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4400303628","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004209559,0.11237826,0.0018403316,0.03617659,0.031308934,0.00009300659,0.0063421964,0.0010108887,0.80664027],"genre_scores_gemma":[0.0075541805,0.010262071,0.00049013074,0.0021976118,0.0014979866,0.000026574957,0.00073632406,0.0002413304,0.97699374],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9977835,0.00016717249,0.00008122913,0.00025096445,0.0013547643,0.00036239004],"domain_scores_gemma":[0.9974287,0.00033651883,0.000122021396,0.0002060672,0.0014261061,0.00048055933],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0012975457,0.0012159114,0.0008583447,0.00814272,0.004607429,0.011088221,0.0015752,0.002027611,0.24100159],"category_scores_gemma":[0.00507158,0.0006464386,0.00055329734,0.01258473,0.0022232654,0.002897872,0.0026580999,0.0020674986,0.03829721],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002745599,0.000017453904,0.0003903677,0.00024405462,0.000009983495,0.000078880315,0.0009194838,0.00009025605,0.00031297244,0.017077724,0.92059916,0.060232252],"study_design_scores_gemma":[0.0000023141838,0.0000022123265,0.0006084612,0.00007064475,0.000003553117,0.000013740925,0.00016076287,0.000010751641,0.00006461476,0.00019466023,0.99886334,0.0000049398172],"about_ca_topic_score_codex":0.6518943,"about_ca_topic_score_gemma":0.8376504,"teacher_disagreement_score":0.9984248,"about_ca_system_score_codex":0.018582016,"about_ca_system_score_gemma":0.016921131,"threshold_uncertainty_score":0.80623066},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W4401072327","doi":"10.5195/jmla.2024.1967","title":"Diversity, equity, and inclusion initiatives in the Medical Library Association: a look back at the last twenty-five years","year":2024,"lang":"en","type":"article","venue":"Journal of the Medical Library Association JMLA","topic":"Primary Care and Health Outcomes","field":"Health Professions","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia Hospital; University of Alberta","funders":"","keywords":"Equity (law); Diversity (politics); Inclusion (mineral); Association (psychology); Medical library; Library science; Political science; Sociology; Psychology; Social science; Computer science; Law","score_opus":0.02846691663219778,"score_gpt":0.37151114954227654,"score_spread":0.34304423291007874,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401072327","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0672449,0.15196577,0.0004821046,0.74815285,0.008058491,0.00003187011,0.000053043368,0.000044259417,0.023966754],"genre_scores_gemma":[0.66279507,0.1596155,0.0023985312,0.14450455,0.013619547,0.00013465066,0.0001488339,0.00006458625,0.016718704],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.98310447,0.007654971,0.0009564745,0.00058823347,0.004266712,0.0034292464],"domain_scores_gemma":[0.9471542,0.018587787,0.0044836323,0.00082976365,0.007767322,0.021177324],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02391238,0.0002585319,0.0004992661,0.0030136209,0.017274693,0.016453369,0.0014882471,0.006411957,0.0034191469],"category_scores_gemma":[0.026699202,0.00039169006,0.00046698184,0.004751148,0.011904305,0.010683628,0.016170569,0.0075678127,0.00036692605],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00014400922,0.0008722892,0.031630684,0.0039730817,0.000064860906,0.001085931,0.20723398,0.00018638656,0.0007274236,0.08167289,0.15708116,0.5153273],"study_design_scores_gemma":[0.000017964314,0.00026366967,0.040082745,0.0041816966,0.0000235047,0.0004944608,0.14768161,0.000075036995,0.0003003063,0.006417532,0.800357,0.00010447784],"about_ca_topic_score_codex":0.017615328,"about_ca_topic_score_gemma":0.048900716,"teacher_disagreement_score":0.02391238,"about_ca_system_score_codex":0.011366107,"about_ca_system_score_gemma":0.038105637,"threshold_uncertainty_score":0.12646228},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4401288674","doi":"10.1007/s11136-024-03727-8","title":"Patient engagement in quality of life research","year":2024,"lang":"en","type":"article","venue":"Quality of Life Research","topic":"Mental Health and Patient Involvement","field":"Health Professions","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Calgary; Alberta Health Services","funders":"Cumming School of Medicine, University of Calgary; Australian and New Zealand Urogenital and Prostate Cancer Trials Group; University of Queensland","keywords":"Quality of Life Research; Public health; Quality of life (healthcare); Psychology; Public engagement; Quality (philosophy); Sociology; Environmental health; Public relations; Medicine; Political science; Nursing; Psychotherapist","score_opus":0.9085348704858113,"score_gpt":0.696182493808275,"score_spread":0.21235237667753626,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4401288674","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7852162,0.010158089,0.009264865,0.095077984,0.0010182118,0.0019645365,0.0010048925,0.000095370095,0.09619971],"genre_scores_gemma":[0.98498017,0.0022687945,0.0037311236,0.0051277005,0.00045365244,0.0013995277,0.00036261178,0.000029703495,0.0016466501],"study_design_codex":"observational","study_design_gemma":"qualitative","domain_scores_codex":[0.6788678,0.29050088,0.007863709,0.0025243366,0.013104662,0.0071385927],"domain_scores_gemma":[0.76500034,0.16558912,0.024625868,0.009300437,0.01470928,0.020774858],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.13711897,0.00035352976,0.0012418376,0.002341238,0.0031973468,0.007782506,0.0013116198,0.002555312,0.008413554],"category_scores_gemma":[0.19942775,0.0005083668,0.0013047892,0.0034526738,0.0021214404,0.0047963257,0.008408614,0.0049120532,0.0009345903],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0021820832,0.0082809,0.5167073,0.002323006,0.0013094441,0.0002779239,0.035407577,0.0007465062,0.0007228634,0.011718099,0.01386188,0.40646243],"study_design_scores_gemma":[0.0021320842,0.014207145,0.66933566,0.01004508,0.001207491,0.0019432667,0.08083725,0.004603574,0.0030114327,0.044457193,0.16788541,0.00033434175],"about_ca_topic_score_codex":0.0009144717,"about_ca_topic_score_gemma":0.0018454214,"teacher_disagreement_score":0.86288106,"about_ca_system_score_codex":0.0029874016,"about_ca_system_score_gemma":0.008673936,"threshold_uncertainty_score":0.7251631},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"methods","study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4402328208","doi":"10.1101/2024.09.06.24313221","title":"Knowledge mobilization with and for equity-deserving communities invested in research: A scoping review protocol","year":2024,"lang":"en","type":"review","venue":"medRxiv","topic":"Health Policy Implementation Science","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Wilfrid Laurier University","funders":"","keywords":"Mobilization; Protocol (science); Equity (law); Business; Community mobilization; Knowledge management; Public economics; Political science; Economics; Computer science; Economic growth; Medicine","score_opus":0.9715157305682416,"score_gpt":0.82871523998628,"score_spread":0.14280049058196154,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402328208","genre_codex":"protocol","genre_gemma":"protocol","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"protocol","genre_consensus":"protocol","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00050090323,0.00082388945,0.004932844,0.0010225513,0.00023318722,0.9863051,0.0046281903,0.0001289964,0.0014243728],"genre_scores_gemma":[0.00047374537,0.00048576863,0.0075034616,0.00014380124,0.000017707916,0.990672,0.00039914672,0.000009986946,0.00029445102],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.8356739,0.083752654,0.05291818,0.0075073517,0.015734416,0.004413479],"domain_scores_gemma":[0.78596175,0.08992702,0.025462698,0.027960028,0.06425209,0.0064364267],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.27673236,0.0039198752,0.0068722405,0.021457225,0.0060676676,0.0109366365,0.006262138,0.009217279,0.06776874],"category_scores_gemma":[0.24614662,0.0050054914,0.010167637,0.020325435,0.00709572,0.008105012,0.009188531,0.00730434,0.01881048],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.006190874,0.00088753324,0.002292796,0.5451281,0.001769121,0.0018022659,0.022018202,0.0030652226,0.003988744,0.056201406,0.10936324,0.2472925],"study_design_scores_gemma":[0.010386252,0.0013458149,0.0073165107,0.43144545,0.0029542604,0.0006417429,0.012024066,0.0023630434,0.004777211,0.041426126,0.48463717,0.0006822593],"about_ca_topic_score_codex":0.0052306447,"about_ca_topic_score_gemma":0.010390814,"teacher_disagreement_score":0.9937379,"about_ca_system_score_codex":0.017625984,"about_ca_system_score_gemma":0.13558982,"threshold_uncertainty_score":0},"labels":[{"model":"gpt","categories":["metaresearch","scholarly_communication"],"domain":"methods","study_design":"design_other","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"grok","categories":["metaresearch","scholarly_communication","open_science"],"domain":"methods","study_design":"systematic_review","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"opus","categories":["metaresearch","scholarly_communication"],"domain":"reporting","study_design":"systematic_review","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4402404259","doi":"10.1371/journal.pbio.3002815","title":"A how-to guide for code sharing in biology","year":2024,"lang":"en","type":"review","venue":"PLoS Biology","topic":"Scientific Computing and Data Management","field":"Decision Sciences","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia, Okanagan Campus","funders":"University of British Columbia","keywords":"Code (set theory); Biology; Set (abstract data type); Data science; Best practice; Computer science; Source code; Programming language","score_opus":0.5141643496662349,"score_gpt":0.5401220480212572,"score_spread":0.025957698355022263,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402404259","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00070662866,0.3135781,0.35082105,0.21411705,0.019996203,0.0029635408,0.0056897677,0.016026456,0.07610117],"genre_scores_gemma":[0.004738833,0.25661096,0.56843287,0.08217783,0.004896854,0.006635513,0.005218698,0.007255789,0.064032614],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9815185,0.007708303,0.0035702826,0.0011499838,0.0054618367,0.00059096614],"domain_scores_gemma":[0.9283003,0.039903402,0.005467499,0.007403285,0.015401435,0.0035241272],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.020114461,0.0023075445,0.002020294,0.0071245376,0.002702907,0.0055616535,0.0062945024,0.0066310666,0.030764962],"category_scores_gemma":[0.07887841,0.0020179872,0.0025169775,0.008977589,0.005542703,0.014061544,0.007300096,0.013105048,0.055031456],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000026026533,0.00006361156,0.00015608402,0.0051407106,0.000042242653,0.00021634324,0.0008482734,0.00048691276,0.00045645286,0.04018661,0.5852856,0.3670911],"study_design_scores_gemma":[0.00000856101,0.000018053277,0.000088093504,0.0034034047,0.0000092683895,0.00024910824,0.0001299085,0.00009277556,0.00013361158,0.014838989,0.9810069,0.00002128454],"about_ca_topic_score_codex":0.0053072586,"about_ca_topic_score_gemma":0.01182587,"teacher_disagreement_score":0.9937055,"about_ca_system_score_codex":0.0034564,"about_ca_system_score_gemma":0.01797407,"threshold_uncertainty_score":0.10637677},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4402695722","doi":"10.5376/jmr.2024.14.0018","title":"Global Collaboration in Research and Data Sharing for Mosquito-Borne Diseases","year":2024,"lang":"en","type":"article","venue":"Journal of Mosquito Research","topic":"Mosquito-borne diseases and control","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Data sharing; Data science; Computer science; Medicine; Pathology","score_opus":0.2797435021467741,"score_gpt":0.5623340285362383,"score_spread":0.28259052638946414,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402695722","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13487068,0.06971636,0.27981162,0.2759435,0.0050273295,0.0029605546,0.002565024,0.0010930395,0.22801198],"genre_scores_gemma":[0.7839066,0.036861014,0.15060031,0.013259417,0.0016378324,0.002118305,0.0019105363,0.00032503917,0.009380986],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.8621731,0.114047244,0.0057421215,0.005422972,0.008632219,0.0039823283],"domain_scores_gemma":[0.8010948,0.13008578,0.015605966,0.029550133,0.012615927,0.011047375],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.11400243,0.00068068603,0.001115833,0.005468346,0.0061906483,0.014843086,0.0029680931,0.0028758065,0.0073950994],"category_scores_gemma":[0.12571324,0.00059540675,0.0014977219,0.009585683,0.0061317044,0.015488,0.028717749,0.003924685,0.0010467571],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026286312,0.00025903524,0.049071874,0.0046182163,0.00066695234,0.0011096745,0.0636944,0.0038073973,0.0016119843,0.3672224,0.035486646,0.47218865],"study_design_scores_gemma":[0.00007715318,0.00036044096,0.01704148,0.0061890627,0.00030498937,0.00107091,0.060569335,0.0030931607,0.0014257334,0.31645387,0.59323823,0.00017562286],"about_ca_topic_score_codex":0.0030219879,"about_ca_topic_score_gemma":0.0037280223,"teacher_disagreement_score":0.8859976,"about_ca_system_score_codex":0.0044543725,"about_ca_system_score_gemma":0.033482257,"threshold_uncertainty_score":0.6029097},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4402890012","doi":"10.1111/jebm.12651","title":"International Traditional Medicine Clinical Trial Registry: A meaningful initiative and its future development","year":2024,"lang":"en","type":"letter","venue":"Journal of Evidence-Based Medicine","topic":"Complementary and Alternative Medicine Studies","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"China Academy of Chinese Medical Sciences; World Health Organization","keywords":"Clinical trial; Alternative medicine; Medicine; Family medicine; Internal medicine; Pathology","score_opus":0.565957851305087,"score_gpt":0.46926807889027355,"score_spread":0.09668977241481347,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4402890012","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"reporting","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":"reporting","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0033658862,0.18597229,0.15005197,0.5248,0.042148296,0.014840431,0.022260338,0.013628833,0.04293199],"genre_scores_gemma":[0.029770428,0.13248093,0.66500497,0.07168629,0.02085326,0.026155207,0.039957926,0.00345598,0.010635041],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.8416474,0.09443816,0.030454654,0.007360838,0.02172105,0.0043779584],"domain_scores_gemma":[0.4892381,0.14876147,0.06364787,0.057516787,0.16416708,0.07666872],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.3182781,0.0021815826,0.0071032424,0.01714168,0.0025919222,0.023900818,0.0092257,0.011349755,0.025552943],"category_scores_gemma":[0.31523806,0.0018078699,0.0040554376,0.024115587,0.0057933982,0.02708843,0.014965034,0.014260018,0.02342122],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012982538,0.00066246465,0.005265776,0.017643753,0.0003785237,0.00030715263,0.0008789943,0.0006282281,0.0011086474,0.09001578,0.38463536,0.49717706],"study_design_scores_gemma":[0.00087434234,0.00068564265,0.0057695596,0.02767349,0.00053468684,0.0006006643,0.00085598987,0.0018181996,0.00079151284,0.033717528,0.9264559,0.00022235031],"about_ca_topic_score_codex":0.0023563784,"about_ca_topic_score_gemma":0.0012343639,"teacher_disagreement_score":0.6817219,"about_ca_system_score_codex":0.007598347,"about_ca_system_score_gemma":0.110327125,"threshold_uncertainty_score":0.8406845},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4403162971","doi":"10.1186/s41073-024-00152-w","title":"An evaluation of the preprints produced at the beginning of the 2022 mpox public health emergency","year":2024,"lang":"en","type":"article","venue":"Research Integrity and Peer Review","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Guelph; Public Health Agency of Canada","funders":"","keywords":"Observational study; Quality (philosophy); Preprint; Scale (ratio); Critical appraisal; Public health; Medicine; Psychology; Computer science; World Wide Web; Alternative medicine; Geography; Nursing; Internal medicine; Pathology","score_opus":0.7197556863314704,"score_gpt":0.6343461286640703,"score_spread":0.0854095576674001,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403162971","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6237333,0.11915195,0.05573661,0.037492,0.015070298,0.054489806,0.06884967,0.0016886977,0.023787627],"genre_scores_gemma":[0.82182384,0.025295805,0.07829918,0.005007889,0.003619231,0.046361107,0.016140228,0.00082721276,0.0026255066],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.5601577,0.15969141,0.16556768,0.011587824,0.099603556,0.0033918624],"domain_scores_gemma":[0.09639234,0.47782326,0.19220503,0.035858545,0.19306616,0.004654609],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.46344754,0.0014536966,0.004422414,0.026345292,0.003770913,0.010180232,0.0040757568,0.0026492632,0.004954505],"category_scores_gemma":[0.7572896,0.0015784259,0.0065322155,0.018994343,0.0051183333,0.005744909,0.008324206,0.002484059,0.0013233798],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.011873225,0.0005593102,0.31683537,0.2114919,0.0071165944,0.002598806,0.068873204,0.0012848637,0.0047134855,0.0041955924,0.05270182,0.31775582],"study_design_scores_gemma":[0.001725112,0.0059725777,0.6112637,0.11298319,0.011742925,0.0031851681,0.031918213,0.0025800022,0.00954463,0.0050473516,0.20295054,0.0010866042],"about_ca_topic_score_codex":0.0031685699,"about_ca_topic_score_gemma":0.0042055175,"teacher_disagreement_score":0.5365524,"about_ca_system_score_codex":0.00951718,"about_ca_system_score_gemma":0.024935476,"threshold_uncertainty_score":0.6616647},"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","open_science"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4403332823","doi":"10.1177/09636625241268881","title":"Public understanding of preprints: How audiences make sense of unreviewed research in the news","year":2024,"lang":"en","type":"article","venue":"Public Understanding of Science","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University; University of British Columbia","funders":"Social Sciences and Humanities Research Council of Canada","keywords":"Preprint; Coronavirus disease 2019 (COVID-19); Population; Pandemic; Term (time); Media studies; Sociology; Public relations; Psychology; Political science; World Wide Web; Computer science; Medicine","score_opus":0.7307891558044088,"score_gpt":0.5019906325918755,"score_spread":0.22879852321253336,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403332823","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9126844,0.0027234533,0.008474625,0.04022113,0.00058851135,0.00010267614,0.00025609144,0.00012458245,0.0348246],"genre_scores_gemma":[0.9940739,0.0011689946,0.00085901236,0.0022958315,0.0003686517,0.000043987573,0.00014471565,0.00006251883,0.000982389],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9519732,0.031520076,0.003194018,0.002251435,0.008869456,0.0021918183],"domain_scores_gemma":[0.7621595,0.16586418,0.040959783,0.010499325,0.013811536,0.0067056795],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.06067248,0.0005442317,0.0008300142,0.004426045,0.0047400277,0.019239359,0.0014361413,0.0035484233,0.0036984996],"category_scores_gemma":[0.17980845,0.00065266533,0.0007105011,0.0025635213,0.010298469,0.01820854,0.010281763,0.006471389,0.00073441333],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00032408827,0.00008816924,0.063423686,0.00052542577,0.00009874778,0.0008197326,0.8747269,0.00019414545,0.0018428398,0.017216284,0.005301705,0.035438277],"study_design_scores_gemma":[0.00004600179,0.00020352493,0.05420301,0.0012539967,0.00013698384,0.00065729674,0.8527386,0.00096801005,0.0011569873,0.021295318,0.06718475,0.00015553612],"about_ca_topic_score_codex":0.0035494484,"about_ca_topic_score_gemma":0.0017895006,"teacher_disagreement_score":0.98076063,"about_ca_system_score_codex":0.004847829,"about_ca_system_score_gemma":0.0031504447,"threshold_uncertainty_score":0.32087058},"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":["scholarly_communication"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4403337641","doi":"10.1002/asi.24957","title":"When data sharing is an answer and when (often) it is not: Acknowledging data‐driven, non‐data, and data‐decentered cultures","year":2024,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"Research Data Management Practices","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"H2020 European Research Council; Social Sciences and Humanities Research Council of Canada; Horizon 2020 Framework Programme","keywords":"Computer science; Data sharing; Information retrieval; Data science","score_opus":0.13412300046449738,"score_gpt":0.407161431979203,"score_spread":0.27303843151470564,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403337641","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.37286055,0.0058175097,0.092527695,0.47139838,0.0019559406,0.0002706952,0.00011496463,0.00015206478,0.054902185],"genre_scores_gemma":[0.97643524,0.0013500832,0.010762965,0.009818885,0.0002176085,0.00014619342,0.000035786008,0.000089025554,0.0011442016],"study_design_codex":"qualitative","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.6979061,0.23622279,0.016765753,0.013911739,0.026723124,0.008470461],"domain_scores_gemma":[0.5891358,0.26933527,0.039421316,0.04908465,0.03913127,0.013891675],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.2334614,0.000616301,0.0011826009,0.0038711189,0.020085232,0.036748454,0.004796403,0.009738794,0.0019879823],"category_scores_gemma":[0.3295517,0.0014362166,0.0009909156,0.005258866,0.08684885,0.05925569,0.024243264,0.014141208,0.00038683743],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006945522,0.000049700877,0.0121335145,0.00032295135,0.00007176563,0.00042607996,0.7646068,0.0001745711,0.0008837958,0.19783509,0.0031527444,0.020273479],"study_design_scores_gemma":[0.000022660959,0.000048736707,0.0023801718,0.0013296438,0.00004810932,0.00044166585,0.7163007,0.00087443134,0.0012873857,0.22657211,0.050591256,0.00010312524],"about_ca_topic_score_codex":0.006972794,"about_ca_topic_score_gemma":0.0049542766,"teacher_disagreement_score":0.9952036,"about_ca_system_score_codex":0.011318913,"about_ca_system_score_gemma":0.02548998,"threshold_uncertainty_score":0.9452786},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4403681703","doi":"10.1515/jcim-2024-0364","title":"A new chapter in traditional, complementary, and integrative medicine: Advancing the <i>Journal of Complementary and Integrative Medicine</i> through open science, meta-research, artificial intelligence, and early career researcher involvement","year":2024,"lang":"en","type":"editorial","venue":"Journal of Complementary and Integrative Medicine","topic":"Complementary and Alternative Medicine Studies","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital","funders":"","keywords":"Integrative medicine; Alternative medicine; Complementary medicine; Psychology; Medicine; Open science; Engineering ethics; Medical education; Traditional medicine; Engineering; Pathology","score_opus":0.342887259939807,"score_gpt":0.468570996232936,"score_spread":0.125683736293129,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403681703","genre_codex":"editorial","genre_gemma":"editorial","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000095641,0.0033294922,0.00006463668,0.022084653,0.9741271,0.000008509391,0.000012317822,0.00001378944,0.00034978503],"genre_scores_gemma":[0.00012483144,0.003726874,0.00010580997,0.020591645,0.97268265,0.000018257033,0.000011126401,0.000019615389,0.0027192528],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9880121,0.0030391596,0.001656286,0.001172731,0.005551502,0.00056827697],"domain_scores_gemma":[0.921187,0.04639573,0.002964524,0.0014273466,0.020641686,0.007383754],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.019196806,0.004312476,0.0056570107,0.006621688,0.0059542013,0.015406908,0.0047138706,0.019279322,0.013374741],"category_scores_gemma":[0.05521457,0.0015905086,0.004027117,0.0032706312,0.0050478852,0.009030502,0.0028570734,0.037697878,0.007445835],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00004083497,0.00001877172,0.000027633512,0.00043920224,0.000024716554,0.00006692411,0.000026630274,0.000016407404,0.000054572403,0.00063342636,0.9931161,0.0055347756],"study_design_scores_gemma":[0.0001063406,0.000059740596,0.0003731883,0.0016159555,0.00012038551,0.0002612422,0.00017105305,0.00018511689,0.00011771169,0.0030378834,0.9939032,0.00004811457],"about_ca_topic_score_codex":0.002428354,"about_ca_topic_score_gemma":0.010747298,"teacher_disagreement_score":0.9952861,"about_ca_system_score_codex":0.0050235847,"about_ca_system_score_gemma":0.009459142,"threshold_uncertainty_score":0.10152364},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"evaluation","study_design":"not_applicable","genre":"editorial","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"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":"W4403865951","doi":"10.1007/s11948-024-00516-x","title":"Authorship and Citizen Science: Seven Heuristic Rules","year":2024,"lang":"en","type":"article","venue":"Science and Engineering Ethics","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Sveriges Lantbruksuniversitet; Svenska Forskningsrådet Formas","keywords":"Transparency (behavior); Philosophy of science; Subject (documents); Heuristic; Research integrity; Political science; Citizen science; Term (time); Scientific misconduct; Public relations; Medical research; Engineering ethics; Sociology; Epistemology; Law; Computer science; Library science; Medicine; Alternative medicine; Engineering","score_opus":0.03919945871547186,"score_gpt":0.29018179500242125,"score_spread":0.2509823362869494,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4403865951","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07077343,0.0058698547,0.3335048,0.26561508,0.0024914972,0.0035525714,0.0010250323,0.0005255259,0.31664214],"genre_scores_gemma":[0.7494274,0.0018280414,0.21347609,0.021266319,0.0013337587,0.0030440975,0.00046232622,0.00020451198,0.0089575155],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.85045207,0.09512671,0.012313613,0.010031554,0.025932768,0.006143331],"domain_scores_gemma":[0.5783059,0.36156204,0.015142408,0.019301852,0.01939653,0.0062912223],"candidate_categories":["metaresearch","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.111011535,0.0012220162,0.0027441464,0.01104893,0.01099055,0.02244994,0.005885051,0.019283826,0.010703236],"category_scores_gemma":[0.26636478,0.0015052567,0.0020262287,0.008625312,0.04716356,0.025075557,0.009941218,0.012772402,0.0014618718],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000065714434,0.00005895313,0.0021317545,0.0002129964,0.000043231103,0.00020095427,0.0019281207,0.00072535203,0.000036419653,0.97468907,0.0072601684,0.01264732],"study_design_scores_gemma":[0.000085089705,0.000016827153,0.00050118373,0.00022840653,0.00003384544,0.00014584596,0.0011246633,0.0027035638,0.00010439427,0.9844915,0.010520366,0.000044248358],"about_ca_topic_score_codex":0.0054923054,"about_ca_topic_score_gemma":0.006964389,"teacher_disagreement_score":0.98071617,"about_ca_system_score_codex":0.009908668,"about_ca_system_score_gemma":0.010400857,"threshold_uncertainty_score":0.58709216},"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":["metaresearch","research_integrity","open_science"],"domain":"incentives","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4404033710","doi":"10.1097/md.0000000000040259","title":"Open science practices among authors published in complementary, alternative, and integrative medicine journals: An international, cross-sectional survey","year":2024,"lang":"en","type":"article","venue":"Medicine","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; Ottawa Hospital","funders":"","keywords":"Medicine; Cross-sectional study; Alternative medicine; Family medicine; MEDLINE; Survey research; Medical education; Applied psychology; Pathology","score_opus":0.3971064348246936,"score_gpt":0.6008773255398717,"score_spread":0.20377089071517812,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404033710","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99805105,0.00029431973,0.00019219906,0.00023979128,0.000007876697,0.00010150971,0.00046625716,0.0000055456017,0.0006415021],"genre_scores_gemma":[0.99719244,0.00073179306,0.0006901021,0.00031141035,0.000027047192,0.000305176,0.0004539287,0.0000079011,0.00028020234],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9910926,0.0030107494,0.002288731,0.00078939914,0.0021629238,0.00065557245],"domain_scores_gemma":[0.93199825,0.026664022,0.027578237,0.00203084,0.007550265,0.004178326],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.013256225,0.00023985213,0.0007625089,0.0038980502,0.0010914089,0.0023848938,0.0005986437,0.0011089208,0.00224817],"category_scores_gemma":[0.038181093,0.00077781576,0.00058640924,0.0043619145,0.0011213485,0.0041701593,0.0019072291,0.0012612753,0.0008008555],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008197753,0.00025939924,0.97650063,0.00025759128,0.00004965564,0.00012372999,0.016852425,0.000027377686,0.0003528819,0.00007283713,0.00037620115,0.005045372],"study_design_scores_gemma":[0.000028411421,0.0007576762,0.9375454,0.00025087618,0.00005026173,0.0006977924,0.05738261,0.00027602725,0.00035562622,0.00012721901,0.002485032,0.000042966298],"about_ca_topic_score_codex":0.0011977957,"about_ca_topic_score_gemma":0.001782702,"teacher_disagreement_score":0.99940133,"about_ca_system_score_codex":0.0008294019,"about_ca_system_score_gemma":0.0011315595,"threshold_uncertainty_score":0.07010645},"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","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W4404566630","doi":"10.1038/s41467-024-54300-3","title":"Data sharing ethics toolkit: The Human Cell Atlas","year":2024,"lang":"en","type":"review","venue":"Nature Communications","topic":"Single-cell and spatial transcriptomics","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University; McGill University Health Centre","funders":"Fonds de Recherche du Québec - Santé; Chan Zuckerberg Initiative; McGill University; Leona M. and Harry B. Helmsley Charitable Trust; Klarman Family Foundation","keywords":"Interoperability; Data sharing; Scope (computer science); Data science; Computer science; Atlas (anatomy); Engineering ethics; Diversity (politics); Knowledge management; World Wide Web; Political science; Biology; Medicine; Engineering","score_opus":0.22913393585906836,"score_gpt":0.4358943295859236,"score_spread":0.20676039372685526,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404566630","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009396602,0.40001374,0.2412845,0.12265846,0.012457582,0.002970868,0.04078747,0.017759647,0.16112815],"genre_scores_gemma":[0.014078742,0.4543012,0.33745712,0.045280263,0.0049053314,0.011112058,0.08538248,0.0044599175,0.043022953],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99013937,0.0048262114,0.0012694672,0.0005240027,0.0028665315,0.00037447998],"domain_scores_gemma":[0.97146726,0.014460013,0.0015095157,0.004750584,0.0054984186,0.0023141305],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.02238965,0.0009132191,0.0014996257,0.0064490307,0.0011403641,0.005179116,0.006889328,0.0032678298,0.019300574],"category_scores_gemma":[0.032931797,0.00072263426,0.0015621017,0.007488756,0.0024848601,0.006116911,0.009789361,0.0068066353,0.016254144],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000048178168,0.000031057138,0.00020206664,0.0057361214,0.000095630065,0.0001363953,0.000553139,0.00063729816,0.00046547226,0.09518425,0.5134854,0.38342497],"study_design_scores_gemma":[0.000008172359,0.0000053553285,0.00019200632,0.001400721,0.000011604426,0.00013115714,0.000067610046,0.00007124718,0.00014242325,0.013975615,0.98398143,0.0000127199655],"about_ca_topic_score_codex":0.004105905,"about_ca_topic_score_gemma":0.003839867,"teacher_disagreement_score":0.99311066,"about_ca_system_score_codex":0.0033547143,"about_ca_system_score_gemma":0.017862586,"threshold_uncertainty_score":0.11840922},"labels":[{"model":"gemma","categories":["research_integrity","metaresearch"],"domain":"methods","study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4404840467","doi":"10.32920/27926781.v1","title":"A cross-sectional description of open access publication costs, policies and impact in emergency medicine and critical care journals","year":2024,"lang":"en","type":"preprint","venue":"","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan; McMaster University","funders":"","keywords":"Cross-sectional study; Medicine; Family medicine; Medical emergency","score_opus":0.8802126189463708,"score_gpt":0.6923778208821509,"score_spread":0.18783479806421988,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404840467","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.91946393,0.0057463883,0.0012662127,0.0008647045,0.00007548854,0.000619443,0.06556376,0.00010753127,0.0062924763],"genre_scores_gemma":[0.96627796,0.0027972816,0.0020528026,0.00043914132,0.00017755963,0.0017565654,0.024806354,0.00008610429,0.0016063246],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98590755,0.0023234766,0.0053765094,0.0016249617,0.004048068,0.00071947917],"domain_scores_gemma":[0.87011987,0.041256256,0.06207077,0.004202499,0.017981589,0.0043689846],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.0114487205,0.00041417909,0.0009338724,0.015440296,0.00096243346,0.00311465,0.0008744181,0.00075950875,0.007983718],"category_scores_gemma":[0.04660382,0.0006143194,0.0011796464,0.026462592,0.00080736546,0.0049912417,0.001676134,0.0011427989,0.0021932255],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015211845,0.00010859477,0.98998404,0.00044061913,0.00024056986,0.00006876219,0.00039240654,0.00010024839,0.00009068554,0.00022452978,0.0021167898,0.006080619],"study_design_scores_gemma":[0.000010304528,0.00012286962,0.9953343,0.00014688706,0.00008474572,0.00053055387,0.0006912686,0.00015793195,0.00010436047,0.00011980081,0.0026720522,0.000024818157],"about_ca_topic_score_codex":0.003033164,"about_ca_topic_score_gemma":0.0028348027,"teacher_disagreement_score":0.9991256,"about_ca_system_score_codex":0.0018289611,"about_ca_system_score_gemma":0.0019208685,"threshold_uncertainty_score":0.06054735},"labels":[{"model":"gemma","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4404849222","doi":"10.1097/qad.0000000000004076","title":"Assessing coronavirus disease 2019 pandemic impacts on the health of people who inject drugs using a novel data sharing model","year":2024,"lang":"en","type":"review","venue":"AIDS","topic":"HIV, Drug Use, Sexual Risk","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"British Columbia Centre on Substance Use; Simon Fraser University","funders":"National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention; National Institute on Drug Abuse; National Institute of Allergy and Infectious Diseases; Centers for Disease Control and Prevention","keywords":"Pandemic; Environmental health; Medicine; Survey data collection; Multilevel model; Moderation; Coronavirus disease 2019 (COVID-19); Mental health; Demography; Psychology; Social psychology; Statistics; Psychiatry; Sociology","score_opus":0.4342052924227519,"score_gpt":0.526325326235946,"score_spread":0.09212003381319406,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4404849222","genre_codex":"methods","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13719578,0.03384798,0.7832194,0.013551877,0.0012099403,0.006493469,0.016362976,0.0015266298,0.0065919873],"genre_scores_gemma":[0.7652553,0.005906152,0.20884937,0.0027835923,0.00038625792,0.008933567,0.0052816174,0.00016982396,0.002434367],"study_design_codex":"simulation_or_modeling","study_design_gemma":"observational","domain_scores_codex":[0.8638175,0.11756059,0.004849309,0.009023749,0.0038154924,0.00093329116],"domain_scores_gemma":[0.766563,0.2034751,0.012921873,0.010799471,0.0050865794,0.0011539913],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.1469286,0.0031134954,0.0046636593,0.0058299806,0.00092437805,0.0046977727,0.0057370043,0.003476468,0.0040144282],"category_scores_gemma":[0.16992879,0.0016254808,0.021568572,0.0050263503,0.002194336,0.0045425007,0.005686384,0.0029751488,0.00046907846],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0040131947,0.0007042667,0.066559575,0.0125832185,0.19161966,0.00044749395,0.0014597087,0.5854395,0.0013857099,0.05076822,0.0049705077,0.080048986],"study_design_scores_gemma":[0.0032197952,0.0031959072,0.019124517,0.0032094042,0.09225796,0.0004978282,0.0005669274,0.74032843,0.001724612,0.116709195,0.01877653,0.00038884798],"about_ca_topic_score_codex":0.012719774,"about_ca_topic_score_gemma":0.0061241854,"teacher_disagreement_score":0.1469286,"about_ca_system_score_codex":0.003928488,"about_ca_system_score_gemma":0.0067659398,"threshold_uncertainty_score":0.777042},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"observational","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4405031048","doi":"10.2139/ssrn.5029611","title":"Legislating Open Access: Making the Case for a Secondary Publishing Right in Canada","year":2024,"lang":"en","type":"preprint","venue":"SSRN Electronic Journal","topic":"Publishing and Scholarly Communication","field":"Arts and Humanities","cited_by":1,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Publishing; Political science; Business; Law","score_opus":0.06856929381170981,"score_gpt":0.32033571091179425,"score_spread":0.2517664171000844,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405031048","genre_codex":"other","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.16227852,0.0027786205,0.004100774,0.39144382,0.0013027182,0.0003155583,0.0011947312,0.00015120769,0.43643403],"genre_scores_gemma":[0.7905191,0.0013278538,0.0026392723,0.06355203,0.00047949757,0.00015148355,0.00028661173,0.00017701305,0.14086716],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9691808,0.0032175637,0.00086332834,0.0028180487,0.010402393,0.013517974],"domain_scores_gemma":[0.9219515,0.024670463,0.0022289893,0.0030873846,0.028711852,0.019349854],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.015519245,0.00041083756,0.0011245995,0.0022709104,0.03001541,0.019563202,0.0039312374,0.013239145,0.01971527],"category_scores_gemma":[0.057752267,0.0007686507,0.00095910334,0.004045496,0.014192994,0.0061636,0.0072560487,0.016418433,0.0010680477],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011620139,0.00008384211,0.0080163125,0.0001498801,0.000051950832,0.00058965187,0.016002377,0.00073622214,0.0007062864,0.8429833,0.10583643,0.024727602],"study_design_scores_gemma":[0.0002475721,0.00006889708,0.025700742,0.0007760959,0.00016066266,0.00019869447,0.02001313,0.0029387132,0.0015001228,0.17383331,0.77428967,0.0002723979],"about_ca_topic_score_codex":0.9883887,"about_ca_topic_score_gemma":0.99291706,"teacher_disagreement_score":0.9960688,"about_ca_system_score_codex":0.123657554,"about_ca_system_score_gemma":0.40783206,"threshold_uncertainty_score":0.89720255},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":true,"about_ca_topic":true,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4405540378","doi":"10.18438/eblip30665","title":"Evidence Summary Theme: Open Access","year":2024,"lang":"en","type":"article","venue":"Evidence Based Library and Information Practice","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"Carleton University","funders":"","keywords":"Theme (computing); Computer science; World Wide Web; Data science; Library science","score_opus":0.5818519236961829,"score_gpt":0.5919576860470508,"score_spread":0.010105762350867908,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4405540378","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0009898983,0.5764951,0.005158613,0.29297838,0.064253464,0.0029117824,0.010319209,0.00016650172,0.046727125],"genre_scores_gemma":[0.036573153,0.6284389,0.01721373,0.24214064,0.044216618,0.011685377,0.008008305,0.00035256305,0.011370753],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.8454535,0.055972897,0.036751304,0.008229336,0.050763886,0.0028291498],"domain_scores_gemma":[0.60134387,0.27932736,0.04019249,0.011243492,0.06195806,0.005934674],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.081970155,0.0014290382,0.0040705823,0.014045304,0.0029757838,0.013840864,0.004979016,0.012825433,0.05116108],"category_scores_gemma":[0.3882108,0.0011283702,0.0055934116,0.017438775,0.004522531,0.012633934,0.009790086,0.013534276,0.008417227],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004901799,0.00006374044,0.0015511699,0.25456133,0.0013471755,0.00013016647,0.0009913425,0.00010843916,0.00018780206,0.03642628,0.39264187,0.31150052],"study_design_scores_gemma":[0.00021912492,0.00012033915,0.0022281967,0.33553964,0.00149963,0.00021792225,0.0012271128,0.00008706749,0.00030271296,0.013168088,0.64534783,0.00004236453],"about_ca_topic_score_codex":0.0041895662,"about_ca_topic_score_gemma":0.005307951,"teacher_disagreement_score":0.995021,"about_ca_system_score_codex":0.011895283,"about_ca_system_score_gemma":0.027578173,"threshold_uncertainty_score":0.43350482},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4406287319","doi":"10.21083/partnership.v19i2.7847","title":"Open Science at the University of Toronto","year":2025,"lang":"en","type":"article","venue":"Partnership The Canadian Journal of Library and Information Practice and Research","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"University of Toronto","funders":"University of Toronto","keywords":"Media studies; Library science; Sociology; Computer science","score_opus":0.09830281079042283,"score_gpt":0.38331487349499827,"score_spread":0.28501206270457546,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406287319","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.045478545,0.024597576,0.010131305,0.05862739,0.0035570066,0.00045440145,0.0125771845,0.0024178184,0.8421587],"genre_scores_gemma":[0.26778552,0.017816348,0.014579131,0.0028525002,0.00059263187,0.00033049702,0.003456585,0.0007740338,0.69181275],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9947754,0.0010214904,0.00023969475,0.0006972206,0.0024540739,0.0008121771],"domain_scores_gemma":[0.9793203,0.004917174,0.00086559355,0.0019633519,0.0058517847,0.007081852],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0037570163,0.00060249504,0.00059077994,0.0019340437,0.01442244,0.010265878,0.0010383778,0.0017944579,0.10913029],"category_scores_gemma":[0.009378142,0.00060863316,0.0005003156,0.0040906062,0.005157739,0.002985296,0.003988088,0.0017427946,0.011317035],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038405985,0.00009064768,0.010698105,0.0023569278,0.000037184,0.0020350723,0.025260475,0.0010162477,0.0024719678,0.109572336,0.6702134,0.17586365],"study_design_scores_gemma":[0.000019504143,0.000031643587,0.0077952137,0.00056725333,0.000011255818,0.00012154291,0.005896972,0.00046725248,0.0008417711,0.0043659117,0.9798313,0.000050317944],"about_ca_topic_score_codex":0.78051203,"about_ca_topic_score_gemma":0.89354515,"teacher_disagreement_score":0.9989616,"about_ca_system_score_codex":0.045111995,"about_ca_system_score_gemma":0.08910361,"threshold_uncertainty_score":0.4415611},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"agree"},{"id":"W4406357471","doi":"10.1086/733018","title":"Open Access in the Field of Contemporary Art","year":2024,"lang":"en","type":"article","venue":"Art Documentation Journal of the Art Libraries Society of North America","topic":"Digital and Traditional Archives Management","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Concordia University","funders":"","keywords":"Field (mathematics); Visual arts; Art; Mathematics","score_opus":0.04756444124792316,"score_gpt":0.2996367965685496,"score_spread":0.25207235532062644,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406357471","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.024659766,0.07025485,0.016164083,0.09499248,0.0020882385,0.000111124944,0.0000799704,0.00021132844,0.79143804],"genre_scores_gemma":[0.77524245,0.06270786,0.01564695,0.015234352,0.006025376,0.00034025311,0.00012381827,0.00036147324,0.1243174],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.982532,0.009011543,0.0010292147,0.0014568061,0.0049119447,0.001058456],"domain_scores_gemma":[0.9568574,0.028245648,0.0026690944,0.006864899,0.002478267,0.002884566],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.01759195,0.0002966164,0.00059153134,0.0044179214,0.014879584,0.031238932,0.0015824218,0.0051573035,0.02058097],"category_scores_gemma":[0.03006582,0.00034786493,0.0004083702,0.0074562626,0.05602376,0.026897062,0.01944838,0.0046722884,0.0024642176],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012890642,0.000028054194,0.00077479146,0.00029254452,0.000006268196,0.00018254439,0.029325329,0.000086073815,0.00013983223,0.8608032,0.013312063,0.09503647],"study_design_scores_gemma":[0.000006602226,0.000024750145,0.00139863,0.0014168391,0.000008084751,0.000506302,0.020308517,0.00012152407,0.00019056183,0.36956868,0.60642475,0.00002474165],"about_ca_topic_score_codex":0.00557034,"about_ca_topic_score_gemma":0.006480206,"teacher_disagreement_score":0.99841756,"about_ca_system_score_codex":0.007666251,"about_ca_system_score_gemma":0.015083016,"threshold_uncertainty_score":0.093036234},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"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":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4406707019","doi":"10.1017/eds.2024.35.pr3","title":"Recommendation: Improving the reproducibility in geoscientific papers: lessons learned from a Hackathon in climate science — R0/PR3","year":2024,"lang":"en","type":"peer-review","venue":"","topic":"Biomedical and Engineering Education","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"Engineering and Physical Sciences Research Council; National Oceanic and Atmospheric Administration; Natural Sciences and Engineering Research Council of Canada; McMaster University; Alan Turing Institute; European Commission; Canada Research Chairs; University of Cambridge","keywords":"Reproducibility; Computer science; Data science; Environmental science; Mathematics; Statistics","score_opus":0.06532426947686354,"score_gpt":0.3250317109235223,"score_spread":0.25970744144665875,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4406707019","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0039003992,0.01103024,0.048620965,0.7488603,0.08530568,0.0031688756,0.015131084,0.0075395913,0.076442875],"genre_scores_gemma":[0.06801359,0.03768899,0.21432175,0.34068713,0.08393966,0.0058371713,0.046214607,0.009346913,0.19395024],"study_design_codex":"not_applicable","study_design_gemma":"qualitative","domain_scores_codex":[0.9127491,0.028960278,0.011354925,0.0043836273,0.040092964,0.0024590495],"domain_scores_gemma":[0.40369353,0.18337387,0.031029288,0.059745368,0.28890827,0.03324962],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0791863,0.0010465786,0.0015581925,0.0067497706,0.0046836026,0.017776834,0.0037082585,0.008148043,0.07914741],"category_scores_gemma":[0.46204317,0.0007384578,0.002043922,0.012403985,0.0026841864,0.016178673,0.0053603034,0.00578259,0.05869014],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000067819296,0.000040790397,0.0016506191,0.0013462084,0.00007957665,0.000070250666,0.00021022756,0.00015511928,0.00030282096,0.0028181905,0.90006185,0.093196556],"study_design_scores_gemma":[0.0002418439,0.00009212949,0.002777429,0.0028801244,0.00013132133,0.00012728549,0.0005510248,0.0012144948,0.0009984126,0.015750604,0.97510105,0.00013436963],"about_ca_topic_score_codex":0.01035952,"about_ca_topic_score_gemma":0.02283738,"teacher_disagreement_score":0.9208137,"about_ca_system_score_codex":0.004354457,"about_ca_system_score_gemma":0.029247733,"threshold_uncertainty_score":0.41878217},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch"],"domain":"reproducibility","study_design":"design_other","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4407160473","doi":"10.31223/x5xb07","title":"A review of open data for studying global groundwater in social-ecological systems","year":2025,"lang":"en","type":"review","venue":"","topic":"Hydrology and Watershed Management Studies","field":"Environmental Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Canadian Space Agency; National Science and Technology Council; Bundesministerium für Bildung und Forschung; Strategic Research Council; Canadian Institute for Advanced Research; Natural Sciences and Engineering Research Council of Canada; Centrum för naturkatastrofslära, Uppsala Universitet; European Commission; Alexander von Humboldt-Stiftung; National Science Foundation","keywords":"Groundwater; Environmental science; Ecology; Geography; Environmental resource management; Water resource management; Geology; Biology; Geotechnical engineering","score_opus":0.18363491261129963,"score_gpt":0.4193952034792956,"score_spread":0.23576029086799596,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407160473","genre_codex":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0024252518,0.8737284,0.035265043,0.023479821,0.003573562,0.00059447606,0.04543914,0.00081085024,0.014683434],"genre_scores_gemma":[0.011402883,0.8928159,0.042493407,0.006130682,0.0023409592,0.0015143376,0.041718777,0.0004033455,0.0011796941],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9816923,0.006095712,0.0045315064,0.0022608675,0.004892109,0.0005275474],"domain_scores_gemma":[0.84147334,0.12456011,0.007821283,0.010039337,0.014066352,0.0020395818],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.020846069,0.0014270677,0.00280882,0.036819827,0.0020930714,0.008050495,0.0047082403,0.00319292,0.0069003706],"category_scores_gemma":[0.09820239,0.0013043855,0.0037242237,0.067219675,0.0039118123,0.017350923,0.0068378267,0.0038281912,0.001873999],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012815864,0.00010332247,0.008046628,0.0499043,0.0006298421,0.0002273102,0.0012892531,0.002599431,0.0005792619,0.100434124,0.13543369,0.7006247],"study_design_scores_gemma":[0.000017255481,0.000030213116,0.0060884426,0.030168617,0.00020477881,0.0001841761,0.0007031684,0.00075393607,0.00023487367,0.030870209,0.93064564,0.00009869063],"about_ca_topic_score_codex":0.01629906,"about_ca_topic_score_gemma":0.017440211,"teacher_disagreement_score":0.99529177,"about_ca_system_score_codex":0.0054987813,"about_ca_system_score_gemma":0.0119324485,"threshold_uncertainty_score":0.11024588},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"review","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4407209761","doi":"10.1016/j.respol.2025.105195","title":"The governance of open science: A comparative analysis of two open science consortia","year":2025,"lang":"en","type":"article","venue":"Research Policy","topic":"Research Data Management Practices","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"McGill University","funders":"Social Sciences and Humanities Research Council of Canada; Fundação para a Ciência e a Tecnologia; HORIZON EUROPE Framework Programme","keywords":"Open science; Open source; Open innovation; Corporate governance; Business; Computer science; Mathematics; Marketing; Statistics; Operating system","score_opus":0.3802670063318805,"score_gpt":0.6304127036853657,"score_spread":0.25014569735348524,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407209761","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9523364,0.0014274563,0.005251972,0.0035599577,0.00010958709,0.00054559344,0.00018654304,0.00003740137,0.036545165],"genre_scores_gemma":[0.99656147,0.00031302872,0.001592246,0.0002469057,0.000019697027,0.0002519395,0.000119611526,0.000018543533,0.0008764585],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.93554693,0.047598243,0.0021141856,0.0021949967,0.007856954,0.004688594],"domain_scores_gemma":[0.8343414,0.116784364,0.019971618,0.007110066,0.0131279975,0.008664527],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.06107879,0.00026336132,0.00083523476,0.007194404,0.007481052,0.0110483505,0.001924918,0.0019065457,0.0033865825],"category_scores_gemma":[0.13669659,0.0003241474,0.00074551196,0.007926539,0.010783142,0.0090397755,0.011676419,0.0017326486,0.00023801306],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00085027365,0.00055556506,0.17122611,0.0011298907,0.00024517122,0.0022481706,0.51143277,0.0015122131,0.0017261618,0.2306082,0.003317736,0.07514786],"study_design_scores_gemma":[0.00026846485,0.00062330783,0.18913618,0.0011355182,0.00022297245,0.0006538848,0.7137471,0.004303462,0.0009609806,0.041476846,0.047310192,0.00016109343],"about_ca_topic_score_codex":0.013981232,"about_ca_topic_score_gemma":0.0168604,"teacher_disagreement_score":0.99807507,"about_ca_system_score_codex":0.014704,"about_ca_system_score_gemma":0.019053413,"threshold_uncertainty_score":0.3230194},"labels":[{"model":"gemma","categories":["open_science","metaresearch"],"domain":"evaluation","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4407935713","doi":"10.19173/irrodl.v26i1.7953","title":"Bibliometric Insights Into the Open Education Landscape","year":2025,"lang":"en","type":"article","venue":"The International Review of Research in Open and Distributed Learning","topic":"Open Education and E-Learning","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Open education; Distance education; Higher education; Data science; Geography; Computer science; Sociology; World Wide Web; Pedagogy; Political science","score_opus":0.0671515032365844,"score_gpt":0.46146022919374263,"score_spread":0.3943087259571582,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4407935713","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.20757574,0.24670559,0.076577134,0.034018178,0.0024782808,0.001189126,0.037335746,0.0012972495,0.392823],"genre_scores_gemma":[0.8375803,0.09557825,0.037423573,0.0012728911,0.002794787,0.0012543858,0.015121299,0.0003317762,0.00864265],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.95636123,0.011889014,0.0060078492,0.0026540807,0.0219724,0.0011154863],"domain_scores_gemma":[0.8536919,0.10276791,0.016164558,0.006769371,0.019500796,0.0011055229],"candidate_categories":["bibliometrics"],"consensus_categories":[],"category_scores_codex":[0.02080007,0.00085368403,0.0018907973,0.23956713,0.003173977,0.019381013,0.0014141,0.0013709416,0.006958232],"category_scores_gemma":[0.12070329,0.00048104845,0.0011213848,0.3114942,0.0038622112,0.013940397,0.005768656,0.0009984543,0.0014053389],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010664965,0.00008776811,0.061019324,0.020153433,0.0008850226,0.00087748276,0.024629889,0.003340907,0.0018364495,0.32017845,0.034525733,0.5323589],"study_design_scores_gemma":[0.00003846672,0.00009128011,0.12007031,0.008910289,0.0006204622,0.0021122396,0.044061966,0.0055913404,0.0016856052,0.2298346,0.58676744,0.0002160523],"about_ca_topic_score_codex":0.005223807,"about_ca_topic_score_gemma":0.0061098128,"teacher_disagreement_score":0.76043284,"about_ca_system_score_codex":0.005137285,"about_ca_system_score_gemma":0.0076331566,"threshold_uncertainty_score":0.11000258},"labels":[{"model":"gemma","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408566433","doi":"10.18438/eblip30630","title":"Identifying Socio-Technical Risks in Open-Source Software for Scholarly Communications: Tools, Metrics, and Opportunities for Libraries to Support Sustainable Development","year":2025,"lang":"en","type":"article","venue":"Evidence Based Library and Information Practice","topic":"Open Source Software Innovations","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"Concordia University; Université Laval","funders":"","keywords":"Computer science; Open source; Data science; Software; Open source software; World Wide Web","score_opus":0.1419888677672576,"score_gpt":0.35946849456636937,"score_spread":0.21747962679911176,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408566433","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7070344,0.010594885,0.19350955,0.04701384,0.00034012296,0.0021843596,0.0024413127,0.0037799517,0.033101678],"genre_scores_gemma":[0.824685,0.0027942583,0.16495557,0.0007223289,0.00026040085,0.001506823,0.0016588115,0.00037962061,0.003037207],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9135323,0.041799042,0.0081617255,0.0029971127,0.031521797,0.00198802],"domain_scores_gemma":[0.54966855,0.22520323,0.08866901,0.024433648,0.102750175,0.009275452],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.089541584,0.0014956831,0.0012014518,0.03572865,0.004398856,0.026303573,0.0033826851,0.0027216931,0.0024868243],"category_scores_gemma":[0.2585331,0.0009918677,0.0012606329,0.026166394,0.005170774,0.030530808,0.017837875,0.0029572658,0.001250684],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0004937057,0.00092054595,0.4062268,0.0032277505,0.00035269998,0.00062658486,0.032992452,0.007808989,0.0026623013,0.0329428,0.0122423535,0.49950305],"study_design_scores_gemma":[0.00016653701,0.0018829722,0.3379126,0.011105208,0.00082817633,0.0012323256,0.14996384,0.10673437,0.021195922,0.20933807,0.15851642,0.0011235353],"about_ca_topic_score_codex":0.0048903413,"about_ca_topic_score_gemma":0.00632634,"teacher_disagreement_score":0.9966173,"about_ca_system_score_codex":0.008421303,"about_ca_system_score_gemma":0.012938717,"threshold_uncertainty_score":0.4735468},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"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":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408661429","doi":"10.63485/qz2g7-n9996","title":"Canadian cities moving on open data","year":2009,"lang":"en","type":"preprint","venue":"","topic":"Human Mobility and Location-Based Analysis","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Open data; Regional science; Geography; Computer science; World Wide Web","score_opus":0.136667586182565,"score_gpt":0.3961454062808494,"score_spread":0.25947782009828435,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408661429","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012513328,0.0110426,0.00261112,0.1714436,0.0066794166,0.0001510487,0.14006807,0.0011182403,0.65437263],"genre_scores_gemma":[0.09334318,0.014267685,0.0074083167,0.017641425,0.0011216309,0.00018382145,0.077663384,0.0011769639,0.7871937],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9942059,0.00036313056,0.00014643493,0.0005729998,0.003492148,0.0012193392],"domain_scores_gemma":[0.98663425,0.00084315386,0.00028863095,0.0009737436,0.008974607,0.002285723],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0026970434,0.00078571116,0.00054762646,0.0052064033,0.013940522,0.0119179245,0.0013390726,0.0019758155,0.065918826],"category_scores_gemma":[0.011736645,0.00070532254,0.0007918261,0.016135316,0.0020898627,0.0037056752,0.004333069,0.002791195,0.010573297],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017038503,0.000004645185,0.0018464148,0.000046439956,0.0000070044616,0.000039333627,0.00039014078,0.000100220524,0.000037817616,0.02373963,0.9565483,0.017223172],"study_design_scores_gemma":[0.0000047646404,0.000001934685,0.0070572197,0.00006239046,0.000005097634,0.000011699512,0.00066861924,0.00008764104,0.000049458667,0.0009463171,0.99108434,0.000020593858],"about_ca_topic_score_codex":0.993266,"about_ca_topic_score_gemma":0.9967604,"teacher_disagreement_score":0.9986609,"about_ca_system_score_codex":0.06531218,"about_ca_system_score_gemma":0.13522126,"threshold_uncertainty_score":0.47387522},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W4408663123","doi":"10.63485/bm1hb-08q72","title":"An OA pledge from the Gustavus Adolphus library faculty","year":2009,"lang":"en","type":"preprint","venue":"","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Pledge; Political science; Library science; Sociology; Computer science; Law","score_opus":0.03841528436749876,"score_gpt":0.26379525370285045,"score_spread":0.2253799693353517,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408663123","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006999397,0.002119552,0.0017733884,0.2054393,0.034547407,0.0006416331,0.0012850395,0.0010104568,0.7461839],"genre_scores_gemma":[0.0137316175,0.00039879297,0.000533586,0.02735243,0.0019572936,0.00018426808,0.00023598204,0.00025089775,0.95535517],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99551696,0.0004646823,0.0002132317,0.0005476136,0.0022400955,0.0010174346],"domain_scores_gemma":[0.9880922,0.0023887644,0.00036977924,0.00089126016,0.003940522,0.0043175044],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.005831767,0.0006427541,0.0005881994,0.001432388,0.010342458,0.008834586,0.0009306756,0.009302372,0.24424966],"category_scores_gemma":[0.018248823,0.0005489988,0.00066225516,0.0012032723,0.0021892271,0.0045608347,0.004531623,0.008036315,0.09959377],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000325713,0.000032718464,0.00024871426,0.00001879451,0.0000022175957,0.00010304325,0.00011037512,0.000017932034,0.00022873879,0.0078246705,0.98334885,0.008031295],"study_design_scores_gemma":[0.000006834929,0.000015555626,0.00060503726,0.000027102547,0.000001885559,0.00001707,0.00019876672,0.000025332823,0.00009908029,0.0005240018,0.9984674,0.000011863091],"about_ca_topic_score_codex":0.029020004,"about_ca_topic_score_gemma":0.06717666,"teacher_disagreement_score":0.99906933,"about_ca_system_score_codex":0.0064633884,"about_ca_system_score_gemma":0.009429624,"threshold_uncertainty_score":0.8170965},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408785920","doi":"10.63485/16sp6-n1d55","title":"New OA journal on the history and philosophy of science","year":2007,"lang":"en","type":"preprint","venue":"","topic":"Interdisciplinary Research and Collaboration","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"History and philosophy of science; Epistemology; Philosophy; Engineering ethics; Engineering","score_opus":0.21288014716398568,"score_gpt":0.44096712164607527,"score_spread":0.2280869744820896,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408785920","genre_codex":"other","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0049720025,0.09504213,0.01055497,0.25277287,0.2276585,0.00013562094,0.0033286854,0.00084018323,0.40469503],"genre_scores_gemma":[0.09531562,0.065450095,0.011746628,0.016662128,0.11888096,0.00017868148,0.0027205576,0.0020522042,0.6869932],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9967956,0.0005200329,0.0002232027,0.0002961355,0.0019466537,0.00021844577],"domain_scores_gemma":[0.9882308,0.0030111321,0.0007153665,0.0014276077,0.003355101,0.0032600355],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.0042915405,0.0006601099,0.0009130503,0.003141759,0.0027671629,0.014132848,0.00078318233,0.0025200418,0.069073215],"category_scores_gemma":[0.01608787,0.0003633832,0.00047553194,0.0040320046,0.0052212942,0.00717955,0.0032615408,0.0036346673,0.010065573],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002085949,0.000023585964,0.00034297927,0.00030414865,0.000023241677,0.00008075363,0.00048182107,0.00006804309,0.00026912458,0.11623416,0.85331815,0.028833138],"study_design_scores_gemma":[0.000007744053,0.000006369489,0.00097238366,0.00020540923,0.000008093685,0.000061146195,0.00028057297,0.00007648648,0.00007148639,0.02529923,0.97300255,0.000008548798],"about_ca_topic_score_codex":0.007334396,"about_ca_topic_score_gemma":0.017330138,"teacher_disagreement_score":0.9992168,"about_ca_system_score_codex":0.0066862353,"about_ca_system_score_gemma":0.011677079,"threshold_uncertainty_score":0.2310729},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4408818599","doi":"10.63485/hxgx8-ftp77","title":"Background on the Canadian Library Association&amp;#39;s OA policy","year":2007,"lang":"en","type":"preprint","venue":"","topic":"Historical and Literary Studies","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Association (psychology); Political science; Medicine; Psychology; Psychotherapist","score_opus":0.06665027434287953,"score_gpt":0.31767672165318755,"score_spread":0.25102644731030804,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4408818599","genre_codex":"other","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0019232653,0.02973426,0.00071393227,0.3813336,0.006301575,0.00014741004,0.0034300936,0.00021125595,0.5762046],"genre_scores_gemma":[0.0783571,0.04034611,0.004336771,0.20816731,0.0051434613,0.0004671107,0.0024007426,0.00040482255,0.6603766],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9773926,0.0017277572,0.000727235,0.0016960247,0.012663829,0.005792463],"domain_scores_gemma":[0.9795577,0.0037843508,0.00052625476,0.000686657,0.010552846,0.0048920996],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.011042502,0.0012251047,0.0010910528,0.010112568,0.030165387,0.028423714,0.005313289,0.016777094,0.089331634],"category_scores_gemma":[0.023589045,0.0011694975,0.0012205773,0.015045788,0.013419502,0.007703684,0.0045887446,0.013029384,0.009794623],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000011330216,0.000026174828,0.00026029666,0.00015821448,0.0000055241817,0.00006502726,0.0011550649,0.0001350691,0.00008858909,0.2598162,0.7213903,0.016888125],"study_design_scores_gemma":[0.0000037832438,0.0000028078239,0.0008988885,0.0003576089,0.0000037918694,0.000013527388,0.00064266304,0.000036105248,0.000030442749,0.003362829,0.99461997,0.00002755172],"about_ca_topic_score_codex":0.9856089,"about_ca_topic_score_gemma":0.9867357,"teacher_disagreement_score":0.9946867,"about_ca_system_score_codex":0.18974169,"about_ca_system_score_gemma":0.31463292,"threshold_uncertainty_score":0.93978477},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":true,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W4409516233","doi":"10.1186/s12920-025-02140-5","title":"Underrepresented populations in genomic research: a qualitative study of researchers’ perspectives","year":2025,"lang":"en","type":"article","venue":"BMC Medical Genomics","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":7,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université Laval; Hôpital du Saint-Sacrement","funders":"Fonds de Recherche du Québec - Santé; Fonds de recherche du Québec","keywords":"Diversity (politics); Harmonization; Thematic analysis; Population; Qualitative research; Inclusion (mineral); Human genetics; Public relations; Data science; Biology; Sociology; Political science; Genetics; Social science; Medicine; Environmental health","score_opus":0.9129197085446948,"score_gpt":0.7410685683667902,"score_spread":0.1718511401779046,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409516233","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96438384,0.0032716515,0.0059439093,0.016824234,0.00042968622,0.00070655334,0.00018678294,0.000049660113,0.008203737],"genre_scores_gemma":[0.9893433,0.0023068492,0.0021301128,0.0033625895,0.000096705924,0.0006607309,0.00006395127,0.000056011067,0.001979783],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.8929193,0.09152216,0.0023036592,0.002845998,0.0044024535,0.0060064783],"domain_scores_gemma":[0.8649348,0.11002395,0.005441236,0.0021582143,0.007858011,0.009583884],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0806438,0.0012867796,0.0024806373,0.0042469073,0.026211608,0.012316706,0.0051699867,0.006022717,0.0031724852],"category_scores_gemma":[0.08685502,0.0014229639,0.0011393005,0.0040816646,0.029555576,0.011791426,0.01666794,0.010799529,0.00056370517],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010648735,0.000012773012,0.0005033257,0.00006501578,0.0000022660122,0.00031114082,0.9973838,0.0000126042805,0.00012437796,0.0006212668,0.00017605479,0.0007766872],"study_design_scores_gemma":[0.0000024771384,0.000014817134,0.00010451631,0.000096291085,0.0000019372949,0.0000975363,0.9969271,0.000030957424,0.000048055103,0.0002403695,0.0024298579,0.000006043251],"about_ca_topic_score_codex":0.013179899,"about_ca_topic_score_gemma":0.013057758,"teacher_disagreement_score":0.9193562,"about_ca_system_score_codex":0.0125495335,"about_ca_system_score_gemma":0.016482841,"threshold_uncertainty_score":0.4264903},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"methods","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4409529490","doi":"10.63485/d4z8b-vdc56","title":"Two library groups support draft CIHR policy","year":2006,"lang":"en","type":"preprint","venue":"","topic":"Biomedical Ethics and Regulation","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Political science; Library science; Computer science","score_opus":0.01904235322547221,"score_gpt":0.31934671318518815,"score_spread":0.30030435995971594,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409529490","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"incentives","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":"incentives","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005090643,0.0023714136,0.008437847,0.6353433,0.011771951,0.00077567605,0.0013770261,0.001466961,0.33336508],"genre_scores_gemma":[0.061310586,0.0013417292,0.018391116,0.38465825,0.006361903,0.0008207718,0.0015233507,0.0013494788,0.5242428],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.82213676,0.02814076,0.008243404,0.008568931,0.1085634,0.024346743],"domain_scores_gemma":[0.56348896,0.19426449,0.009559475,0.03239732,0.17008649,0.030203326],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.13288869,0.0012455472,0.0025645993,0.008982911,0.018341912,0.03832501,0.00747233,0.052295573,0.060738847],"category_scores_gemma":[0.317121,0.0015968347,0.0022914,0.009419799,0.011011893,0.012578053,0.012715235,0.023353316,0.02014538],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016947357,0.00017167334,0.001152668,0.00018947186,0.00003123659,0.00027494438,0.0030305872,0.00018943536,0.00059669145,0.2550945,0.7165692,0.02253013],"study_design_scores_gemma":[0.00009588252,0.000030280946,0.0014448726,0.00018517459,0.000041518062,0.00006774533,0.0012533676,0.0003584948,0.0010306071,0.0125874,0.9828166,0.00008806204],"about_ca_topic_score_codex":0.26963693,"about_ca_topic_score_gemma":0.27134883,"teacher_disagreement_score":0.94789565,"about_ca_system_score_codex":0.05210433,"about_ca_system_score_gemma":0.19571006,"threshold_uncertainty_score":0.702791},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":true,"about_ca_topic":true,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4409629257","doi":"10.63485/djw3v-6t729","title":"Canadian Library Association endorses OA","year":2005,"lang":"en","type":"preprint","venue":"","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Association (psychology); Library science; Psychology; Computer science; Psychotherapist","score_opus":0.01054748751395643,"score_gpt":0.1968457446434901,"score_spread":0.18629825712953368,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409629257","genre_codex":"other","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00023250852,0.003406548,0.0014040364,0.037191737,0.015334985,0.00032502346,0.013733509,0.0025279368,0.9258436],"genre_scores_gemma":[0.0038317873,0.0032386193,0.0031855504,0.012207406,0.0012280296,0.00029243212,0.0086616315,0.0014020915,0.9659525],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9396376,0.0026591758,0.0034775592,0.004358173,0.042908024,0.0069594975],"domain_scores_gemma":[0.83570844,0.0062494674,0.002662648,0.01295365,0.13111207,0.011313772],"candidate_categories":["scholarly_communication","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.016838843,0.001882649,0.0032709015,0.010325696,0.014793091,0.029321533,0.006218262,0.01205642,0.47400436],"category_scores_gemma":[0.074084826,0.0017405396,0.0019776507,0.022159414,0.0052518095,0.008129191,0.0071249255,0.0090309195,0.35813832],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021567428,0.000014659103,0.00012001672,0.0001671612,0.000007866622,0.00004409883,0.000116683295,0.00001905451,0.00007287752,0.0119266305,0.9695294,0.017959995],"study_design_scores_gemma":[0.0000042888437,0.000002071581,0.0003131791,0.00009214486,0.0000048882043,0.000016110913,0.00009878706,0.00001626568,0.00002661649,0.0005607813,0.9988543,0.000010496345],"about_ca_topic_score_codex":0.81401026,"about_ca_topic_score_gemma":0.86960244,"teacher_disagreement_score":0.99378175,"about_ca_system_score_codex":0.035372354,"about_ca_system_score_gemma":0.22891964,"threshold_uncertainty_score":0.75026894},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W4409728141","doi":"10.5206/cjils-rcsib.v48i1.22390","title":"Towards Sustainable and Coordinated Methods for Estimating Open Access Costs at Canadian Higher Education Institutions","year":2025,"lang":"en","type":"article","venue":"Canadian Journal of Information and Library Science","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":true,"ca_institutions":"University of British Columbia; University of Ottawa","funders":"","keywords":"Business; Environmental economics; Higher education; Economic growth; Economics","score_opus":0.354533325933165,"score_gpt":0.5916812547562789,"score_spread":0.23714792882311386,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409728141","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"incentives","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"incentives","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.055011228,0.0024853358,0.9018214,0.009883357,0.00017213586,0.001496038,0.003417935,0.0008138947,0.024898725],"genre_scores_gemma":[0.25261045,0.0014274197,0.7379248,0.0005870543,0.0000868944,0.0011708156,0.0024744482,0.00027220603,0.0034459787],"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.95341474,0.022344442,0.0022830726,0.0029419558,0.016307129,0.0027086353],"domain_scores_gemma":[0.89875704,0.037101675,0.009522378,0.007855239,0.044987768,0.0017759674],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.04022233,0.0018546758,0.0024058588,0.01566106,0.004060434,0.014139007,0.005184216,0.0020398991,0.0023364194],"category_scores_gemma":[0.15284787,0.0014733571,0.002175085,0.023175163,0.0038360513,0.005491154,0.0079807695,0.0034221136,0.0005324475],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010410527,0.00018331756,0.094386466,0.00094857026,0.00092303567,0.00020028854,0.003853268,0.3614076,0.00068989256,0.18863729,0.01900266,0.3296635],"study_design_scores_gemma":[0.000098800214,0.000074149044,0.087283865,0.0010921851,0.0002840391,0.00010644694,0.00600778,0.6834066,0.0016928834,0.17274621,0.046901785,0.00030523224],"about_ca_topic_score_codex":0.93024796,"about_ca_topic_score_gemma":0.922458,"teacher_disagreement_score":0.99481577,"about_ca_system_score_codex":0.068950176,"about_ca_system_score_gemma":0.11152254,"threshold_uncertainty_score":0.50027084},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"incentives","study_design":"not_applicable","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W4409912679","doi":"10.1098/rsos.240857","title":"Towards an inclusive Open Science: examining EDI and public participation in policy documents across Europe and the Americas","year":2025,"lang":"en","type":"article","venue":"Royal Society Open Science","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":5,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal; Simon Fraser University","funders":"Arts and Humanities Research Council; Fundação de Amparo à Pesquisa do Estado de São Paulo; Deutsche Forschungsgemeinschaft","keywords":"Openness to experience; Incentive; Transparency (behavior); Public relations; Ethos; Equity (law); Political science; Public policy; Science policy; Open science; Inclusion (mineral); Diversity (politics); Sociology; Public administration; Economics; Social science; Psychology","score_opus":0.3359765003308867,"score_gpt":0.6180077266880395,"score_spread":0.2820312263571528,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4409912679","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7774813,0.015895542,0.007344299,0.10853348,0.0005970665,0.00066207326,0.0005705377,0.000069797454,0.088845946],"genre_scores_gemma":[0.9882587,0.0023762286,0.0023544752,0.0045687417,0.00008160223,0.00034307447,0.00014029934,0.00003998752,0.0018369007],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.84474796,0.09671168,0.013629964,0.0066674883,0.029292418,0.008950431],"domain_scores_gemma":[0.40435034,0.5074502,0.03777361,0.014685904,0.02780356,0.007936353],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.20476945,0.00037287845,0.0012615547,0.015527823,0.01664358,0.03163588,0.0025075676,0.007758392,0.0027454777],"category_scores_gemma":[0.30818978,0.0007791147,0.0005675299,0.026062561,0.035340324,0.019908533,0.019727381,0.007979882,0.00030092717],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015359395,0.00017738136,0.019104768,0.0019496122,0.000055525645,0.0009238226,0.6358451,0.000603462,0.00115268,0.2821359,0.0045948783,0.05330321],"study_design_scores_gemma":[0.00004681587,0.000089067245,0.023930976,0.0033907031,0.0000395948,0.0002979525,0.75656176,0.00043458585,0.0009853905,0.03172818,0.1823864,0.000108627406],"about_ca_topic_score_codex":0.021008905,"about_ca_topic_score_gemma":0.014734649,"teacher_disagreement_score":0.99749243,"about_ca_system_score_codex":0.041195836,"about_ca_system_score_gemma":0.048379574,"threshold_uncertainty_score":0},"labels":[{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"grok","categories":["open_science","scholarly_communication","metaresearch"],"domain":"incentives","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"opus","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4410877489","doi":"10.1515/iph-2025-2005","title":"“Yes, We’re Open.” International Public History Goes Open Access","year":2025,"lang":"en","type":"article","venue":"International Public History","topic":"Historical Studies and Socio-cultural Analysis","field":"Arts and Humanities","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Public history; Political science; Media studies; History; Public relations; Sociology","score_opus":0.16987548733497923,"score_gpt":0.3281490789075899,"score_spread":0.15827359157261064,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4410877489","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008772493,0.012007049,0.0011104889,0.80728,0.07896584,0.000023626442,0.0002916283,0.00025893384,0.099185206],"genre_scores_gemma":[0.09888612,0.03194881,0.0026635844,0.4697755,0.08003538,0.00018814237,0.0010084716,0.0015013607,0.31399268],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.989794,0.0034625337,0.00047015114,0.0007710963,0.0037068417,0.0017953328],"domain_scores_gemma":[0.9507218,0.018152732,0.0035700854,0.0062101264,0.012401741,0.008943575],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.013647641,0.00049402413,0.00087702903,0.002534123,0.008145008,0.034637544,0.0019156893,0.009663291,0.07093666],"category_scores_gemma":[0.060461935,0.00058859674,0.00059493457,0.0047217966,0.01692466,0.032333426,0.008415862,0.023412868,0.02126188],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000017878132,0.000018071185,0.0002565685,0.00016112694,0.0000072006665,0.000060514005,0.0016046893,0.000016950795,0.00006287956,0.093060784,0.8802001,0.024533309],"study_design_scores_gemma":[0.0000018504063,0.0000046496853,0.00027035075,0.00031347238,0.000002858906,0.000023733182,0.0011643926,0.0000133833255,0.000041889973,0.007474579,0.9906791,0.000009629045],"about_ca_topic_score_codex":0.004434797,"about_ca_topic_score_gemma":0.005555944,"teacher_disagreement_score":0.9980843,"about_ca_system_score_codex":0.005674978,"about_ca_system_score_gemma":0.014545787,"threshold_uncertainty_score":0.23730671},"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":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4412582784","doi":"10.1136/bmjopen-2024-097148","title":"Data sharing statements for clinical trials: a cross-sectional survey of cardiology journals","year":2025,"lang":"en","type":"article","venue":"BMJ Open","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"St. Joseph’s Healthcare Hamilton","funders":"National Natural Science Foundation of China","keywords":"Medicine; Clinical trial; Publication; Data sharing; Odds; Odds ratio; MEDLINE; Family medicine; Logistic regression; Alternative medicine; Internal medicine; Pathology","score_opus":0.9414220175727007,"score_gpt":0.759484280899632,"score_spread":0.18193773667306867,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412582784","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9933832,0.0019859704,0.00030138853,0.0015631061,0.000030555642,0.00012387423,0.0012895601,0.000022423868,0.0012998341],"genre_scores_gemma":[0.99745125,0.0008403053,0.00047677418,0.00035150113,0.000064775144,0.000118107775,0.00053459074,0.000009083433,0.0001535564],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.95495504,0.0175498,0.015681233,0.0026193864,0.007936804,0.0012576483],"domain_scores_gemma":[0.55697423,0.15157631,0.25338078,0.010171101,0.019503873,0.008393717],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.031023854,0.00027062645,0.00070915057,0.006974768,0.0009948986,0.0024792608,0.00096955843,0.0011789347,0.0027147557],"category_scores_gemma":[0.15640461,0.0006081492,0.00064761983,0.010115131,0.0011619925,0.0042289062,0.0026819408,0.0010111477,0.0008027228],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010526132,0.00007315804,0.99291635,0.00023723698,0.00007181183,0.000056694993,0.0016399251,0.00003347979,0.00016295108,0.000054917564,0.0006262058,0.00402207],"study_design_scores_gemma":[0.000021132635,0.00025665856,0.9914124,0.00020692048,0.00004802636,0.0005587298,0.005150985,0.00020459933,0.00016188719,0.00009179992,0.0018566852,0.000030140665],"about_ca_topic_score_codex":0.0014663173,"about_ca_topic_score_gemma":0.0014071991,"teacher_disagreement_score":0.9990305,"about_ca_system_score_codex":0.0010567261,"about_ca_system_score_gemma":0.0018777184,"threshold_uncertainty_score":0},"labels":[{"model":"gpt","categories":["metaresearch","scholarly_communication","open_science"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"grok","categories":["metaresearch","metaepi_broad","open_science","scholarly_communication","research_integrity"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"opus","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4412738309","doi":"10.7202/1118906ar","title":"Les tests génétiques en libre accès et prédisposition au cancer : cadre légal Français et enjeux éthiques","year":2025,"lang":"fr","type":"article","venue":"Canadian Journal of Bioethics","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":2,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":true,"route_about_ca":false,"ca_institutions":"","funders":"","keywords":"Disposition; Psychology; Social psychology","score_opus":0.3675388634619959,"score_gpt":0.5800619874376418,"score_spread":0.21252312397564593,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412738309","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13371594,0.03772939,0.013128737,0.69806737,0.0021397974,0.00019607636,0.0005441634,0.00010353065,0.11437499],"genre_scores_gemma":[0.8643554,0.014207366,0.006522312,0.08534091,0.0014364374,0.00026566692,0.00026130382,0.000081839404,0.027528772],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.94632745,0.03357703,0.0025277943,0.0029256053,0.010733299,0.00390889],"domain_scores_gemma":[0.8683378,0.07737192,0.011902876,0.0074821236,0.02380909,0.011096108],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.047712725,0.00038576155,0.00072773657,0.0025459765,0.006946239,0.0076053506,0.001465505,0.0063718013,0.0064306096],"category_scores_gemma":[0.09031813,0.00042735698,0.00047532367,0.001704118,0.017993985,0.003978967,0.0046211933,0.005416412,0.0008398548],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002942249,0.00030373412,0.11676935,0.00085993233,0.00013006633,0.0028493349,0.061798535,0.0006628469,0.0021217382,0.53653705,0.087713085,0.18996012],"study_design_scores_gemma":[0.00010064965,0.0004979823,0.12042159,0.0052604694,0.00012945036,0.0069114137,0.039153554,0.00097055506,0.0021260187,0.11521931,0.7088035,0.0004055272],"about_ca_topic_score_codex":0.12038671,"about_ca_topic_score_gemma":0.09397525,"teacher_disagreement_score":0.12038671,"about_ca_system_score_codex":0.013325877,"about_ca_system_score_gemma":0.03426162,"threshold_uncertainty_score":0.25233203},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4412777194","doi":"10.56289/ijcsrp.190","title":"Comparative Analysis of Policy and Practice of Kazakhstan's Open Data","year":2023,"lang":"en","type":"article","venue":"International Journal of Civil Service Reform and Practice","topic":"Economic Growth and Development","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Political science; Regional science; Geography","score_opus":0.09079842203810558,"score_gpt":0.40532930790168786,"score_spread":0.3145308858635823,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4412777194","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9283688,0.0026330282,0.0015872158,0.008781073,0.000117135394,0.00016443623,0.00026017503,0.00003536955,0.05805271],"genre_scores_gemma":[0.9975586,0.00045702158,0.00041963602,0.00025064987,0.000012562372,0.000055519307,0.000050299208,0.00000651703,0.0011890498],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.9766867,0.013681598,0.0014347679,0.0012122861,0.0036001564,0.0033844605],"domain_scores_gemma":[0.9657872,0.021018496,0.0043821754,0.0022097188,0.0041926834,0.0024096712],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.019917928,0.00019801986,0.00036707358,0.0033887203,0.0063932077,0.00797067,0.00086219155,0.0013604985,0.004088938],"category_scores_gemma":[0.03733073,0.00027776868,0.00022848487,0.009219849,0.008289532,0.0045154863,0.0061527533,0.001452517,0.0002840726],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00053727964,0.0004920073,0.13073039,0.001507502,0.000096316166,0.0013606839,0.29142925,0.001422736,0.0019072193,0.38186413,0.006980378,0.18167208],"study_design_scores_gemma":[0.000062966035,0.00037346312,0.3068141,0.0011433944,0.00005775074,0.00041043546,0.5235544,0.0008533261,0.0013765334,0.013324441,0.15194246,0.000086659995],"about_ca_topic_score_codex":0.033008534,"about_ca_topic_score_gemma":0.033460658,"teacher_disagreement_score":0.033008534,"about_ca_system_score_codex":0.020042172,"about_ca_system_score_gemma":0.029272923,"threshold_uncertainty_score":0.14541686},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4413883169","doi":"10.1186/s12916-025-04328-z","title":"Practice of data sharing plans in clinical trial registrations and concordance between registered and published data sharing plans: a cross-sectional study","year":2025,"lang":"en","type":"article","venue":"BMC Medicine","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital; University of Ottawa; St. Joseph’s Healthcare Hamilton; Impact; McMaster University; Population Health Research Institute","funders":"Natural Science Foundation of Guangdong Province; National Natural Science Foundation of China","keywords":"Medicine; Clinical trial; Concordance; Data sharing; Family medicine; Trial registration; Pooling; MEDLINE; Alternative medicine; Psychological intervention; Internal medicine; Pathology","score_opus":0.86517481160786,"score_gpt":0.6895049982589103,"score_spread":0.1756698133489497,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4413883169","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99029106,0.0022352033,0.002802464,0.00091476034,0.000045074583,0.0008885418,0.001059657,0.000039555405,0.0017236408],"genre_scores_gemma":[0.9961283,0.00033194234,0.0020833318,0.00024961517,0.000027443335,0.00061396835,0.00039768175,0.000019790441,0.00014795754],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.7227369,0.16036262,0.07067745,0.015193947,0.027621832,0.0034072516],"domain_scores_gemma":[0.31552014,0.3083847,0.30147955,0.044266984,0.025547272,0.004801353],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.20991418,0.00035520334,0.0010309431,0.0040341155,0.001602632,0.0036677958,0.0022010608,0.0015609293,0.0020710425],"category_scores_gemma":[0.36778235,0.0012874563,0.0017061416,0.0063131293,0.0028365378,0.0061152964,0.004305052,0.002492385,0.0005495431],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003218192,0.00016236096,0.9879562,0.00033517464,0.0002427843,0.0000421624,0.004154469,0.00010871642,0.000084177715,0.00025356156,0.0003882449,0.005950222],"study_design_scores_gemma":[0.00011019627,0.0012005445,0.98243815,0.0009082435,0.00028000536,0.0008177222,0.008082887,0.0016412887,0.0004963327,0.00043236878,0.0034887097,0.00010353985],"about_ca_topic_score_codex":0.0021428678,"about_ca_topic_score_gemma":0.0015925014,"teacher_disagreement_score":0.7900858,"about_ca_system_score_codex":0.0017965313,"about_ca_system_score_gemma":0.0038849947,"threshold_uncertainty_score":0.9743165},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4414576662","doi":"10.7910/dvn/w5ai5a","title":"Replication Package for \"From Online Job Postings to Economic Insights: A Machine Learning Approach to Structuring Naturally Occurring Data\"","year":2025,"lang":"en","type":"dataset","venue":"Harvard Dataverse","topic":"AI and HR Technologies","field":"Business, Management and Accounting","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Bank of Canada","funders":"","keywords":"Replication (statistics); Replicate; Transparency (behavior); Confidentiality; Raw data; Code (set theory); Pipeline (software)","score_opus":0.02730397854844898,"score_gpt":0.2575495038037025,"score_spread":0.23024552525525352,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414576662","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016486179,0.00009887043,0.16545804,0.0009265851,0.0010377399,0.0010510711,0.6194251,0.20169747,0.008656439],"genre_scores_gemma":[0.0136963045,0.0001785181,0.30065808,0.0006148044,0.0002247609,0.0077036098,0.5774538,0.088046536,0.011423568],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9936103,0.0016833631,0.00083438767,0.0018216864,0.0017124244,0.00033784675],"domain_scores_gemma":[0.9420559,0.022018569,0.0015029857,0.02378006,0.009813994,0.0008286837],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.012625585,0.0018707955,0.0014563196,0.003194206,0.0014095622,0.0035942069,0.004458843,0.0012887788,0.21436083],"category_scores_gemma":[0.09879031,0.0020774484,0.0037212048,0.0056603644,0.001216319,0.0033648692,0.004682607,0.003412518,0.1352108],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003738555,0.00008266798,0.0018179822,0.0007816602,0.00021948862,0.0000949867,0.00047704615,0.0018799684,0.00075386686,0.008205024,0.9452333,0.04008017],"study_design_scores_gemma":[0.0007853275,0.00009921019,0.006833036,0.00047678215,0.00012910963,0.0001790859,0.0003568518,0.01713842,0.005194306,0.03216575,0.9363984,0.00024376756],"about_ca_topic_score_codex":0.030312989,"about_ca_topic_score_gemma":0.02813768,"teacher_disagreement_score":0.21436083,"about_ca_system_score_codex":0.0019095324,"about_ca_system_score_gemma":0.00798117,"threshold_uncertainty_score":0.7171084},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4414618895","doi":"10.59348/bcw6r-nmn94","title":"Dissemination and Assessment: Open Access, Reputation and Economics","year":2015,"lang":"en","type":"preprint","venue":"","topic":"Library Science and Information Systems","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Reputation; Sharing economy; Information economy; Digital economy; Reputation management; Information economics","score_opus":0.09226499298459453,"score_gpt":0.39962071956343487,"score_spread":0.30735572657884036,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414618895","genre_codex":"commentary","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005604636,0.066156715,0.053972617,0.5024644,0.016210243,0.0008173013,0.0029274293,0.0014050803,0.35044158],"genre_scores_gemma":[0.5733746,0.08586521,0.04139576,0.033163644,0.03359032,0.0017202188,0.00309834,0.0015675307,0.2262244],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.8962952,0.048572805,0.0056489976,0.0052002682,0.041578922,0.0027036977],"domain_scores_gemma":[0.5672113,0.2261266,0.028877227,0.05700689,0.103690006,0.017088007],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.097391136,0.0007956419,0.002201734,0.009864777,0.005801869,0.044122998,0.0031123203,0.01135788,0.071329996],"category_scores_gemma":[0.48331752,0.0009476777,0.00083189714,0.01954328,0.014631948,0.03708261,0.012668539,0.0078065237,0.01312434],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001335689,0.00003744663,0.002216702,0.0013622133,0.000105409585,0.00011422172,0.00093454495,0.0010439815,0.0001950347,0.5665041,0.23959194,0.18776087],"study_design_scores_gemma":[0.0000750249,0.000047111836,0.0024785316,0.0018802262,0.000051805728,0.0001547151,0.00064690097,0.0021376275,0.00033024716,0.48613244,0.50595814,0.00010721944],"about_ca_topic_score_codex":0.01657156,"about_ca_topic_score_gemma":0.009833537,"teacher_disagreement_score":0.9968877,"about_ca_system_score_codex":0.016403763,"about_ca_system_score_gemma":0.031789538,"threshold_uncertainty_score":0.5150597},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"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":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4414831486","doi":"10.1093/jas/skaf300.131","title":"335 The Dog Aging Project: An open science study of aging in companion dogs.","year":2025,"lang":"en","type":"article","venue":"Journal of Animal Science","topic":"Human-Animal Interaction Studies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institute of Aging","funders":"","keywords":"Healthy aging; Successful aging; Companion animal; Disease; Longitudinal study; Longevity; Translational research","score_opus":0.04958880957145081,"score_gpt":0.4466227322329184,"score_spread":0.3970339226614676,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4414831486","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98790866,0.0011180904,0.0012004136,0.0014964741,0.00012301157,0.0003914497,0.0040871766,0.000052097657,0.00362263],"genre_scores_gemma":[0.97456723,0.001278808,0.005609511,0.00090444204,0.00016322809,0.0013550578,0.005736948,0.000036553585,0.0103482045],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9989889,0.00059787004,0.000032712855,0.00013606137,0.00015037456,0.00009400574],"domain_scores_gemma":[0.99780315,0.00027955108,0.0003302686,0.00020420802,0.00047624516,0.00090646534],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0032176538,0.00027516024,0.0003483644,0.0009839587,0.0011647386,0.00054662675,0.00042995502,0.0005561453,0.0026589632],"category_scores_gemma":[0.0020372053,0.0002088779,0.00026603998,0.0006946384,0.0006203484,0.0006539925,0.0012916372,0.00063973083,0.0005989068],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0028219619,0.0052603516,0.87295973,0.00038149636,0.00037282094,0.0016614283,0.005975623,0.0003372179,0.011277104,0.0022411225,0.028180273,0.06853089],"study_design_scores_gemma":[0.000110403045,0.0044119544,0.9613689,0.00006985427,0.00013472508,0.000862797,0.0033169498,0.00029736164,0.0013311987,0.0006284463,0.027443543,0.000023812298],"about_ca_topic_score_codex":0.0069176294,"about_ca_topic_score_gemma":0.010738227,"teacher_disagreement_score":0.9995701,"about_ca_system_score_codex":0.0007347668,"about_ca_system_score_gemma":0.0017024525,"threshold_uncertainty_score":0.017016768},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4415063374","doi":"10.1177/15562646251384573","title":"Weighing Open Science Against Research Participation Burden in Informed Consent: A Randomized Pilot Study","year":2025,"lang":"en","type":"article","venue":"Journal of Empirical Research on Human Research Ethics","topic":"Patient-Provider Communication in Healthcare","field":"Health Professions","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"Réseau de cancérologie Rossy","keywords":"Randomized controlled trial; Open science; Clinical trial; Protocol (science); Open data; Research ethics; Patient participation; Open label","score_opus":0.950138280848278,"score_gpt":0.78406629686054,"score_spread":0.16607198398773804,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415063374","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9648936,0.000448725,0.0040400526,0.0006109111,0.00032134814,0.02821602,0.00013311682,0.00006477543,0.0012714876],"genre_scores_gemma":[0.9655439,0.00024443623,0.008469125,0.00044600456,0.0002725009,0.024598802,0.00006209739,0.000011939047,0.00035120358],"study_design_codex":"randomized_trial","study_design_gemma":"randomized_trial","domain_scores_codex":[0.8888704,0.10114854,0.0037821142,0.0018903278,0.0026367619,0.0016718844],"domain_scores_gemma":[0.7419865,0.21798652,0.019865554,0.011995118,0.0032791202,0.004887236],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.06800449,0.001695922,0.0017567873,0.0013291719,0.0011883469,0.0020251868,0.00132463,0.0027470225,0.0077118534],"category_scores_gemma":[0.11090345,0.00096209836,0.0028173903,0.001399511,0.005564935,0.003247386,0.0018400239,0.0037424008,0.00070286065],"study_design_candidate":"randomized_trial","study_design_consensus":"randomized_trial","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.8338169,0.116236314,0.0058422545,0.0009630402,0.0011680136,0.000087401655,0.001354867,0.0011774777,0.0012010838,0.0015837365,0.000588175,0.035980817],"study_design_scores_gemma":[0.3733256,0.6162073,0.0041027293,0.00011610593,0.0007139275,0.000049969407,0.0003749018,0.0027135739,0.0007649495,0.001053458,0.0005257699,0.00005180465],"about_ca_topic_score_codex":0.00065340754,"about_ca_topic_score_gemma":0.0008944418,"teacher_disagreement_score":0.99867535,"about_ca_system_score_codex":0.002042405,"about_ca_system_score_gemma":0.0043259882,"threshold_uncertainty_score":0.35964644},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"methods","study_design":"randomized_trial","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"methods","study_design":"randomized_trial","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4415360119","doi":"10.12688/f1000research.155832.2","title":"Fostering open science and responsible research practices: A pre-post study","year":2025,"lang":"en","type":"preprint","venue":"F1000Research","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital","funders":"","keywords":"Intervention (counseling); Perception; Open science; Protocol (science); Open peer review; Massive open online course; Graduate students; Open data; Preprint","score_opus":0.9479692848096105,"score_gpt":0.7782260764395977,"score_spread":0.16974320837001278,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4415360119","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99683446,0.00003068023,0.00020250613,0.00020775948,0.000033572054,0.0013894655,0.00010238394,0.000025991454,0.0011730974],"genre_scores_gemma":[0.98554647,0.00013839254,0.0021922605,0.0004587057,0.00005426326,0.007393886,0.00014901656,0.000016066888,0.004050881],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.9966209,0.0011160681,0.0002461213,0.00045136723,0.00056763133,0.0009978152],"domain_scores_gemma":[0.9797717,0.008745777,0.0021336148,0.0014570409,0.003211302,0.004680482],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.009218694,0.0007046157,0.0016093135,0.0013470094,0.003377341,0.0017484972,0.0008391208,0.0015566954,0.011043314],"category_scores_gemma":[0.01980369,0.0011250087,0.0011197974,0.0007568875,0.0020266888,0.00164421,0.0016696039,0.0033491547,0.0019263192],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.011126577,0.2562801,0.13481764,0.002860149,0.00019980429,0.001436797,0.3033753,0.00034705258,0.017232608,0.0010064667,0.005235899,0.26608166],"study_design_scores_gemma":[0.0033250523,0.18801282,0.5906716,0.0007349728,0.00032304405,0.00028339442,0.18716222,0.000742183,0.008726355,0.0012705607,0.018488405,0.00025935262],"about_ca_topic_score_codex":0.0025358314,"about_ca_topic_score_gemma":0.00396594,"teacher_disagreement_score":0.9991609,"about_ca_system_score_codex":0.0015039882,"about_ca_system_score_gemma":0.004066577,"threshold_uncertainty_score":0.04875368},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W4416633866","doi":"10.5256/f1000research.151380.r219458","title":"Referee report. For: A cross-sectional audit and survey of Open Science and Data Sharing practices at The Montreal Neurological Institute-Hospital [version 1; peer review: 1 approved with reservations, 1 not approved]","year":2023,"lang":"en","type":"article","venue":"Faculty of 1000 Research Ltd","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"Montreal Neurological Institute and Hospital","keywords":"Audit; Open science; Data sharing; Open data; Peer review; Peer-to-peer","score_opus":0.8339304603727095,"score_gpt":0.6108585959485638,"score_spread":0.22307186442414562,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4416633866","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006696652,0.0025294868,0.0033193477,0.532233,0.31888396,0.006865744,0.10585631,0.004115975,0.025526626],"genre_scores_gemma":[0.011254739,0.003934611,0.01284149,0.6039519,0.08939799,0.03532032,0.03623827,0.003383663,0.20367704],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.95671135,0.011964885,0.012150291,0.0028419187,0.014225404,0.002106116],"domain_scores_gemma":[0.40525734,0.20104533,0.017612858,0.0220562,0.34496906,0.009059265],"candidate_categories":["metaresearch","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.042616062,0.0019849373,0.003737229,0.008747191,0.004678959,0.0054372456,0.0053917407,0.014100643,0.3324118],"category_scores_gemma":[0.53507805,0.001445662,0.0035186722,0.010380734,0.001803078,0.004020448,0.005464927,0.008188386,0.1493241],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002041638,0.0000058552964,0.00008562327,0.00029130283,0.000008712376,0.00003217951,0.000027059,0.0000075711837,0.00001958407,0.00010937678,0.9976943,0.0016979161],"study_design_scores_gemma":[0.00062750006,0.00006157024,0.006990073,0.004285963,0.00013883274,0.00034708006,0.0007105884,0.000370411,0.00031306347,0.0027931773,0.983117,0.00024480428],"about_ca_topic_score_codex":0.04233881,"about_ca_topic_score_gemma":0.05878028,"teacher_disagreement_score":0.9946083,"about_ca_system_score_codex":0.007778357,"about_ca_system_score_gemma":0.0159254,"threshold_uncertainty_score":0.95223355},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"commentary","about_ca_system":true,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"methods","study_design":"observational","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4417082230","doi":"10.59350/186wa-bjr97","title":"Coalition Publica Receives Support from SSHRC for the Development of Open Access","year":2020,"lang":"","type":"article","venue":"","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"General partnership; Public access; Public policy; Research council","score_opus":0.4634377207147514,"score_gpt":0.518041368829564,"score_spread":0.05460364811481255,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417082230","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002101835,0.0012029568,0.012988156,0.114560656,0.006177056,0.0016515077,0.05266697,0.008937459,0.7997134],"genre_scores_gemma":[0.019502806,0.00077847205,0.013281741,0.006047495,0.0008436561,0.00048709445,0.030423345,0.0051298346,0.92350554],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9646578,0.0035375857,0.0009835346,0.0016663618,0.025599495,0.0035552115],"domain_scores_gemma":[0.8085377,0.028979888,0.003501524,0.018367037,0.08963249,0.050981376],"candidate_categories":["scholarly_communication","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.025988573,0.0008132507,0.00129392,0.0048611136,0.00494858,0.011836655,0.0032454866,0.0037045134,0.48202312],"category_scores_gemma":[0.10246343,0.0006436559,0.0010690392,0.0047742818,0.002791159,0.006973349,0.009855727,0.004607656,0.17974],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012218932,0.000057690526,0.00035878932,0.00010527533,0.0000073906926,0.000071317,0.00023306179,0.000056442383,0.000372323,0.02990848,0.9229363,0.045770686],"study_design_scores_gemma":[0.00002019225,0.00001535564,0.00048495916,0.00004603896,0.0000043699965,0.000032561875,0.0001365899,0.00012310014,0.00023236097,0.0015339913,0.9973557,0.000014791193],"about_ca_topic_score_codex":0.18407933,"about_ca_topic_score_gemma":0.23796347,"teacher_disagreement_score":0.9967545,"about_ca_system_score_codex":0.018286932,"about_ca_system_score_gemma":0.0900934,"threshold_uncertainty_score":0.73883116},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W4417102456","doi":"10.1177/16094069251406154","title":"Reflections From Participants Who Participated in a Photovoice Study: The Importance of Flexibility and Individualization","year":2025,"lang":"en","type":"article","venue":"International Journal of Qualitative Methods","topic":"Participatory Visual Research Methods","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Sinai Health System; Trillium Health Centre; University of Toronto","funders":"Heart and Stroke Foundation of Canada","keywords":"Photovoice; Flexibility (engineering); Qualitative research; Focus group; Participant observation; Perception; Theme (computing); Photo elicitation","score_opus":0.9532401214793264,"score_gpt":0.8217988344980192,"score_spread":0.1314412869813072,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417102456","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9326974,0.0021824879,0.020004345,0.017183324,0.0014368317,0.0009945949,0.0003772382,0.0002962399,0.024827583],"genre_scores_gemma":[0.97773314,0.001465808,0.007658065,0.0038085582,0.00023234192,0.0010943377,0.00010257582,0.00022113333,0.0076841568],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.98397696,0.012276097,0.00047894203,0.0007845415,0.0012969215,0.0011865653],"domain_scores_gemma":[0.9515797,0.036393046,0.0017299367,0.0024939445,0.005158966,0.0026444402],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.022742242,0.0010017486,0.000942268,0.001014125,0.009708366,0.004611268,0.0017472077,0.0031726027,0.004564311],"category_scores_gemma":[0.054178193,0.00072472007,0.0010187508,0.0007973922,0.0072797467,0.004358457,0.004984769,0.0062817587,0.0011025281],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011796851,0.000053148848,0.0020601514,0.00027150864,0.00001247874,0.0017632367,0.9771992,0.00007462736,0.0035382507,0.0006230125,0.0027076625,0.011578758],"study_design_scores_gemma":[0.000015653812,0.00018486267,0.0022077784,0.00031688425,0.000017331868,0.0010298224,0.9485311,0.00018150751,0.0018988856,0.0011040228,0.044441245,0.00007093797],"about_ca_topic_score_codex":0.0032059695,"about_ca_topic_score_gemma":0.0065968614,"teacher_disagreement_score":0.022742242,"about_ca_system_score_codex":0.0018355821,"about_ca_system_score_gemma":0.002414487,"threshold_uncertainty_score":0},"labels":[{"model":"gpt","categories":["metaresearch"],"domain":"methods","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"grok","categories":["metaresearch","open_science"],"domain":"methods","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"opus","categories":["metaresearch"],"domain":"methods","study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W4417113431","doi":"10.59350/etnvz-37q13","title":"Policy for Culture Change: Making data sharing the default","year":2017,"lang":"","type":"article","venue":"","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Data sharing; Open data; Data governance; Key (lock); Data collection; Audit; Data access","score_opus":0.6233634959951544,"score_gpt":0.5401370905487094,"score_spread":0.083226405446445,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417113431","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0023450113,0.0005799065,0.0070239967,0.95045376,0.005089353,0.00010352005,0.0001934164,0.00029369656,0.033917382],"genre_scores_gemma":[0.13851458,0.0020149546,0.019487275,0.7908596,0.0038761632,0.0010604423,0.0007381494,0.0009674102,0.042481393],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.7952431,0.12024362,0.010860637,0.015826821,0.038213193,0.019612692],"domain_scores_gemma":[0.5824492,0.14379872,0.014614116,0.1041781,0.06413574,0.09082415],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.1847223,0.00084706355,0.0014129229,0.0028809852,0.020455794,0.039967712,0.010099073,0.027943116,0.046053734],"category_scores_gemma":[0.37504625,0.0012997636,0.002517842,0.004722175,0.03543966,0.0875865,0.043461364,0.043916028,0.013670222],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009238961,0.00016930577,0.0023244063,0.00034894122,0.00008575787,0.00016482155,0.015067973,0.00025147593,0.000386689,0.38213935,0.53802276,0.060946018],"study_design_scores_gemma":[0.00007687924,0.000061925086,0.0014990453,0.0013333503,0.000034571152,0.00013962969,0.022024129,0.00039166067,0.00058521685,0.26720852,0.7065179,0.00012731516],"about_ca_topic_score_codex":0.019261142,"about_ca_topic_score_gemma":0.011633269,"teacher_disagreement_score":0.98990095,"about_ca_system_score_codex":0.014575796,"about_ca_system_score_gemma":0.08569127,"threshold_uncertainty_score":0.9769166},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","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":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W4417252685","doi":"10.1037/cep0000386","title":"Open science: My insights into data sharing, preregistration, and replication.","year":2025,"lang":"en","type":"article","venue":"Canadian Journal of Experimental Psychology/Revue canadienne de psychologie expérimentale","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"University of Waterloo","funders":"Eunice Kennedy Shriver National Institute of Child Health and Human Development; Canada Excellence Research Chairs, Government of Canada","keywords":"Open science; Transparency (behavior); Open data; Work (physics); Equity (law); Best practice; Key (lock); Open research","score_opus":0.17170995186462434,"score_gpt":0.44908838402634504,"score_spread":0.2773784321617207,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W4417252685","genre_codex":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00067162595,0.009397674,0.025846269,0.95090604,0.005404361,0.00014104579,0.0000968596,0.00007885219,0.0074571935],"genre_scores_gemma":[0.21430983,0.037644982,0.15960541,0.53422135,0.029997608,0.0059340233,0.00039064806,0.00094631413,0.01694978],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.50465465,0.42583305,0.016523069,0.014055706,0.034434542,0.004498982],"domain_scores_gemma":[0.19504668,0.7353259,0.009651674,0.026618792,0.027816564,0.005540392],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.39136514,0.0007644508,0.001638821,0.005118071,0.019238913,0.029274868,0.009008331,0.025243364,0.0044961125],"category_scores_gemma":[0.6180138,0.0013353392,0.0019151561,0.008935932,0.08491272,0.037160408,0.013771143,0.029085241,0.0012869328],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005365467,0.000024036342,0.00061805604,0.0008414938,0.000053146017,0.00030859114,0.068492025,0.00042004025,0.00010334487,0.7250068,0.16292746,0.041151304],"study_design_scores_gemma":[0.000036067417,0.000025257568,0.00039791208,0.0024972975,0.000021166898,0.00017054427,0.013756033,0.000768542,0.00020635365,0.5012349,0.48082766,0.000058352507],"about_ca_topic_score_codex":0.020452851,"about_ca_topic_score_gemma":0.027356423,"teacher_disagreement_score":0.99099165,"about_ca_system_score_codex":0.019673545,"about_ca_system_score_gemma":0.0767792,"threshold_uncertainty_score":0.75055516},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W612909509","doi":"","title":"Open Data and Open Tools: The Global Impact Study Inventory and Web Application","year":2012,"lang":"en","type":"article","venue":"ResearchWorks at the University of Washington (University of Washington)","topic":"ICT Impact and Policies","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"International Development Research Centre; University of Washington; Bill and Melinda Gates Foundation","keywords":"Computer science; World Wide Web; Raw data; Data access; Timeline; Open data; Information and Communications Technology; Data science; Database; Geography","score_opus":0.05992267877842417,"score_gpt":0.30746161430817914,"score_spread":0.24753893552975498,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W612909509","genre_codex":"empirical","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.48214704,0.009918176,0.039016493,0.021745449,0.0009315588,0.0019441226,0.10032115,0.004152927,0.33982307],"genre_scores_gemma":[0.8167488,0.0070951483,0.047705032,0.0019453958,0.00056001864,0.0012722866,0.051248465,0.0018711179,0.07155373],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99259686,0.0015407904,0.00052334485,0.0006391785,0.004169961,0.00052979065],"domain_scores_gemma":[0.9684019,0.011703172,0.0026171408,0.0045758854,0.010321879,0.0023800572],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.014132575,0.00077766157,0.0006780832,0.023855917,0.001969401,0.00914363,0.0010205937,0.0008575158,0.013321174],"category_scores_gemma":[0.01997562,0.0007504216,0.0010717926,0.033332728,0.0016389434,0.0130710425,0.0044056433,0.00166837,0.0037696536],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006019426,0.0008083104,0.24301253,0.0005221988,0.00022596089,0.0002473038,0.0021834197,0.0034420057,0.0029200895,0.029198049,0.09691656,0.6199217],"study_design_scores_gemma":[0.00008076529,0.00065205275,0.5411483,0.0014853857,0.0005390557,0.00055283034,0.01980171,0.012973371,0.015275364,0.042976756,0.3642381,0.00027624244],"about_ca_topic_score_codex":0.028987024,"about_ca_topic_score_gemma":0.056394015,"teacher_disagreement_score":0.9989794,"about_ca_system_score_codex":0.0042718654,"about_ca_system_score_gemma":0.005440888,"threshold_uncertainty_score":0.074741066},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W6889244635","doi":"10.25547/kp0a-9630","title":"Recommandations politiques pour le libre accès aux données de recherche en Europe (RECODE)","year":2024,"lang":"fr","type":"article","venue":"Electronic Textual Cultures Lab","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Context (archaeology); Work (physics); Order (exchange); Field (mathematics)","score_opus":0.6224377247633078,"score_gpt":0.5920105115187804,"score_spread":0.030427213244527462,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6889244635","genre_codex":"commentary","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0044824295,0.022640752,0.055318285,0.545486,0.035064735,0.004918537,0.014480393,0.0036463172,0.31396255],"genre_scores_gemma":[0.048868265,0.02594133,0.31216928,0.33769193,0.018651532,0.04586763,0.030794382,0.0040924093,0.17592321],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.5330858,0.1900506,0.103620775,0.014526538,0.14066873,0.018047547],"domain_scores_gemma":[0.28867117,0.4182881,0.04109564,0.06676867,0.17384228,0.0113342125],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.34103128,0.0020965652,0.0038835744,0.025828967,0.010232852,0.04912866,0.009567538,0.049418718,0.023647891],"category_scores_gemma":[0.5601731,0.006155121,0.00714608,0.020090897,0.016692873,0.021742268,0.020475963,0.042422302,0.032460254],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00041001022,0.00041059617,0.002332151,0.00380883,0.00024609294,0.0005842556,0.0044086766,0.0005746126,0.0019645088,0.28216863,0.6365064,0.06658521],"study_design_scores_gemma":[0.00020433604,0.00011802876,0.00468096,0.008143814,0.0002884644,0.00040217987,0.0014290746,0.00049776817,0.0018080784,0.03580352,0.94639385,0.00022989877],"about_ca_topic_score_codex":0.03646962,"about_ca_topic_score_gemma":0.027968295,"teacher_disagreement_score":0.99043244,"about_ca_system_score_codex":0.014748597,"about_ca_system_score_gemma":0.07213076,"threshold_uncertainty_score":0.81262577},"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":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6891520934","doi":"10.4224/40002774","title":"Plan d'action pour la science ouverte : réponse à la feuille de route pour la science ouverte du gouvernement du Canada","year":2021,"lang":"fr","type":"report","venue":"NRC Digital Repository","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Plan (archaeology); Context (archaeology); Work (physics); Presentation (obstetrics)","score_opus":0.01570691002330522,"score_gpt":0.24062682044082084,"score_spread":0.22491991041751563,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6891520934","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06908545,0.014456817,0.018352238,0.13962422,0.00534235,0.004291493,0.06787508,0.0037749708,0.67719734],"genre_scores_gemma":[0.13952166,0.007186145,0.031449035,0.009359097,0.00021444004,0.0015907378,0.0148408525,0.00068928587,0.7951487],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99014217,0.00065614376,0.00029150306,0.00032802304,0.006344882,0.0022372461],"domain_scores_gemma":[0.98816097,0.0008095317,0.0002879168,0.0003673606,0.007052254,0.0033220565],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.006514662,0.00090310775,0.00060258724,0.003165335,0.0061978134,0.007377776,0.0021403336,0.005655305,0.02693866],"category_scores_gemma":[0.010717191,0.0009125451,0.0013139892,0.00317548,0.0017352344,0.0017901626,0.002862752,0.0043549365,0.0046630995],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006569316,0.00037102992,0.019773081,0.00088162255,0.00012619409,0.00082647195,0.0035914618,0.006631873,0.0042071235,0.083260484,0.77732515,0.102348536],"study_design_scores_gemma":[0.00011870423,0.000109898385,0.030729415,0.0002416853,0.000038662303,0.00007067931,0.002451284,0.0015719736,0.0015354658,0.0020098996,0.9610397,0.00008270138],"about_ca_topic_score_codex":0.98915863,"about_ca_topic_score_gemma":0.9848897,"teacher_disagreement_score":0.99785966,"about_ca_system_score_codex":0.07355652,"about_ca_system_score_gemma":0.4533577,"threshold_uncertainty_score":0.53369236},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6893064476","doi":"10.5281/zenodo.14164212","title":"Mapping Open Science Scholarly Literature","year":2024,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec à Montréal; Dalhousie University","funders":"Deutsche Forschungsgemeinschaft","keywords":"Presentation (obstetrics); Scholarly communication; Open science; Key (lock); Knowledge production","score_opus":0.09546984991447147,"score_gpt":0.3302817465086922,"score_spread":0.23481189659422072,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6893064476","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01799043,0.016413892,0.0272447,0.015363541,0.009359607,0.0012679807,0.1093155,0.009964262,0.7930801],"genre_scores_gemma":[0.07764419,0.028406035,0.06873272,0.001943055,0.005072928,0.0016783433,0.15750094,0.0077606356,0.6512611],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99866056,0.00017178396,0.00009038232,0.0002094559,0.0007480894,0.00011969623],"domain_scores_gemma":[0.99311537,0.0018118772,0.00068858353,0.0007146455,0.0025099039,0.0011596511],"candidate_categories":["metaresearch","bibliometrics","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0019484134,0.00058263366,0.0006013789,0.049479824,0.0032292735,0.009048926,0.00097187195,0.0007638144,0.38355112],"category_scores_gemma":[0.01319525,0.00037115873,0.0006559717,0.04890909,0.00079443393,0.007249567,0.0047352724,0.0009736101,0.11984684],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000059324993,0.0000407134,0.0017311365,0.0013997902,0.000022792063,0.00013196791,0.001102386,0.0002075916,0.0017515116,0.023099877,0.6182376,0.35221532],"study_design_scores_gemma":[0.000008471821,0.000011230748,0.0029073176,0.0005367326,0.000016632799,0.00008595042,0.001831918,0.00017027407,0.00065761607,0.010609277,0.9831523,0.000012347041],"about_ca_topic_score_codex":0.003392518,"about_ca_topic_score_gemma":0.0075169713,"teacher_disagreement_score":0.99902815,"about_ca_system_score_codex":0.0016010785,"about_ca_system_score_gemma":0.0043884222,"threshold_uncertainty_score":0.8792895},"labels":[{"model":"gemma","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6893182128","doi":"10.5281/zenodo.15176863","title":"UOttawa Open Access Toolkit","year":2025,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa","funders":"","keywords":"Access to information; Data access; Information access; Key (lock); Publishing; Public access","score_opus":0.19528247208475824,"score_gpt":0.4475338067158395,"score_spread":0.25225133463108124,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6893182128","genre_codex":"other","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017316873,0.0067079808,0.11969499,0.0055109425,0.0012377598,0.0016154067,0.08473439,0.1583337,0.6204331],"genre_scores_gemma":[0.008922836,0.00746612,0.19286188,0.0023458877,0.00025178574,0.0030521874,0.066968374,0.046405874,0.67172503],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9932882,0.0013585886,0.0010811054,0.0007211423,0.0027650842,0.00078585255],"domain_scores_gemma":[0.97537893,0.0053764144,0.00087051035,0.0065658554,0.009532027,0.002276238],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.0062240055,0.001565215,0.002056941,0.007152968,0.0049387794,0.013693341,0.005327448,0.0029234553,0.23290613],"category_scores_gemma":[0.023004966,0.003966746,0.0011692073,0.012025211,0.0019896026,0.011463236,0.0065554897,0.003631786,0.27694398],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00016671424,0.000043998793,0.00018849429,0.00048733936,0.00001382933,0.0001292813,0.0006972215,0.00016100139,0.0013269752,0.014730374,0.84247077,0.139584],"study_design_scores_gemma":[0.000007808392,0.0000045018237,0.00021442343,0.0001515176,0.0000050313706,0.0000522793,0.0001376302,0.000079800164,0.0004385574,0.0023896436,0.9964849,0.000033886387],"about_ca_topic_score_codex":0.16382046,"about_ca_topic_score_gemma":0.31512016,"teacher_disagreement_score":0.99467254,"about_ca_system_score_codex":0.006474765,"about_ca_system_score_gemma":0.025637543,"threshold_uncertainty_score":0.77914864},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6893454800","doi":"10.5281/zenodo.3831921","title":"Open Synthesis: Open Science in Evidence Synthesis (second speaker)","year":2020,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Meta-analysis and systematic reviews","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Ottawa Hospital","funders":"","keywords":"Presentation (obstetrics); Session (web analytics); Scholarship; Incentive; Analytics; Evidence-based practice; Statement (logic); Best practice","score_opus":0.7292730324795896,"score_gpt":0.47942345862608565,"score_spread":0.24984957385350398,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6893454800","genre_codex":"editorial","genre_gemma":"commentary","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0003573086,0.031970114,0.09235151,0.3234137,0.49685395,0.009024961,0.0040343734,0.003931131,0.038063005],"genre_scores_gemma":[0.014073599,0.042020146,0.15847287,0.23410639,0.3517518,0.060072757,0.004591798,0.011367508,0.12354315],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9027543,0.06481698,0.009018808,0.0065433476,0.01430957,0.002557008],"domain_scores_gemma":[0.7537946,0.15896764,0.014403532,0.015544551,0.04345383,0.01383579],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.17001458,0.0022404972,0.003710687,0.0067002093,0.0026236498,0.013238841,0.0049533625,0.014342597,0.23729135],"category_scores_gemma":[0.3084821,0.0024857766,0.004766164,0.0035230753,0.005973831,0.009147299,0.016788032,0.022208096,0.08026182],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022278055,0.000025332163,0.000036017795,0.0041705538,0.00008759493,0.00011851028,0.0006344167,0.00018000776,0.000316922,0.00983852,0.91422695,0.070142515],"study_design_scores_gemma":[0.00025226598,0.000060126615,0.00018019251,0.008223147,0.000048402326,0.00017768334,0.00037151063,0.00041994138,0.00035718136,0.039636306,0.95018935,0.000083850384],"about_ca_topic_score_codex":0.00081280404,"about_ca_topic_score_gemma":0.00074650795,"teacher_disagreement_score":0.9950466,"about_ca_system_score_codex":0.005273664,"about_ca_system_score_gemma":0.011527259,"threshold_uncertainty_score":0.89913374},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"methods","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6893746320","doi":"10.5281/zenodo.4675774","title":"Europe and North America on Open Science and the Decolonization of Knowledge: Webinar Report","year":2021,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Digital Education and Society","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Decolonization; Commission; Normative; Open university; Kingdom; Open science","score_opus":0.03332754732048797,"score_gpt":0.2773228982931861,"score_spread":0.24399535097269814,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6893746320","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015708597,0.10231144,0.004133022,0.11072254,0.011832783,0.00047679112,0.012414323,0.00088605785,0.7415145],"genre_scores_gemma":[0.10996747,0.10889045,0.0083593875,0.04812341,0.0026570081,0.0013953429,0.017538186,0.0009628828,0.7021058],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.995952,0.0005930842,0.00025491673,0.00038185454,0.0018772152,0.0009409087],"domain_scores_gemma":[0.9935865,0.0012666787,0.0007043603,0.00032840154,0.0024029887,0.0017110351],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.005980772,0.000557112,0.00036586108,0.0034153587,0.0017124857,0.0072931536,0.0007833966,0.0035804678,0.026337447],"category_scores_gemma":[0.0061941347,0.0004122955,0.00039793606,0.0067773894,0.0009052136,0.0067667407,0.004258249,0.0029128387,0.006831385],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013872652,0.000100677964,0.0019835704,0.00076185947,0.00002117104,0.00029560085,0.00215302,0.0001566288,0.00079118053,0.04386578,0.84036857,0.1093631],"study_design_scores_gemma":[0.000004030726,0.000009085574,0.0064570983,0.00021543732,0.000003864691,0.000037647344,0.00086637953,0.000016979671,0.00014088306,0.00077084475,0.9914689,0.000008841922],"about_ca_topic_score_codex":0.04455778,"about_ca_topic_score_gemma":0.042145796,"teacher_disagreement_score":0.9992166,"about_ca_system_score_codex":0.0035122794,"about_ca_system_score_gemma":0.013658313,"threshold_uncertainty_score":0.08859682},"labels":[{"model":"gemma","categories":["open_science","sts"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6901896296","doi":"10.6084/m9.figshare.1221742.v1","title":"Culture change in academia: Making sharing the new norm - UCalgary, Open Access Week","year":2014,"lang":"en","type":"other","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Norm (philosophy); Culture change; Access to information","score_opus":0.3039607030800888,"score_gpt":0.4430867481752102,"score_spread":0.1391260450951214,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6901896296","genre_codex":"commentary","genre_gemma":"other","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009964522,0.0025030011,0.0032063248,0.4945405,0.014510709,0.00009379735,0.00028648402,0.00092376495,0.4739709],"genre_scores_gemma":[0.2545729,0.005244511,0.006734449,0.08307764,0.0057221586,0.00039585214,0.00051295955,0.0022656939,0.6414739],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98957306,0.005335615,0.00028752614,0.0004830639,0.002414175,0.0019065813],"domain_scores_gemma":[0.97070205,0.0053313076,0.0010422845,0.0032655927,0.0043079234,0.015350943],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.01421458,0.0003815315,0.0006197781,0.0008920366,0.020560993,0.03381821,0.0023206174,0.0069436315,0.12907721],"category_scores_gemma":[0.03378591,0.00044679278,0.0005347762,0.0017826656,0.01036684,0.02158872,0.02242953,0.009325185,0.02290356],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000040773553,0.000081429105,0.0004697362,0.00009371101,0.0000060434913,0.00012691304,0.01608921,0.000052225623,0.00017535461,0.043289993,0.8668738,0.072700754],"study_design_scores_gemma":[0.000009103405,0.000025324403,0.0010929918,0.00018790006,0.0000028459024,0.000033053857,0.028319513,0.000041844083,0.00016061479,0.014104332,0.956003,0.000019535917],"about_ca_topic_score_codex":0.008897295,"about_ca_topic_score_gemma":0.031088565,"teacher_disagreement_score":0.99767935,"about_ca_system_score_codex":0.0066050347,"about_ca_system_score_gemma":0.01699352,"threshold_uncertainty_score":0.43180627},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6907888404","doi":"10.25453/plabs.27276609.v1","title":"The importance of open access to scholarly scientific knowledge, science advice and national science advice mechanisms in building trust in science","year":2024,"lang":"en","type":"article","venue":"Figshare","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Declaration; Context (archaeology); Advice (programming); Open science; Open access publishing; Open research; Access to information; Public access","score_opus":0.1545254070908609,"score_gpt":0.4824524023667184,"score_spread":0.32792699527585745,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6907888404","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005559541,0.011188932,0.03821729,0.56182057,0.0046346765,0.00022463282,0.00024752927,0.00037281378,0.37773392],"genre_scores_gemma":[0.6032487,0.027397625,0.07826778,0.13234825,0.0075581535,0.0010179763,0.00067640486,0.0012067561,0.1482783],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.84691435,0.07292066,0.0074677365,0.009129898,0.055436417,0.008130958],"domain_scores_gemma":[0.63857013,0.21475674,0.020516038,0.04031858,0.039279904,0.046558604],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.1250939,0.0005411116,0.0014513466,0.0061219833,0.013988544,0.04405678,0.004107036,0.016971057,0.03254436],"category_scores_gemma":[0.20786987,0.0012811432,0.0011820231,0.008021031,0.06146966,0.051219396,0.032063052,0.024531042,0.007692629],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003072965,0.00004488911,0.0007355483,0.00029371848,0.000029341178,0.00017694825,0.0073309494,0.0001891166,0.0002337259,0.89673376,0.045582537,0.048618738],"study_design_scores_gemma":[0.000024253326,0.000034924255,0.0009485385,0.0009639932,0.000014580744,0.00017691504,0.0028007075,0.0003027514,0.0002492909,0.46041396,0.5339971,0.0000729022],"about_ca_topic_score_codex":0.019986615,"about_ca_topic_score_gemma":0.022564078,"teacher_disagreement_score":0.99589294,"about_ca_system_score_codex":0.023401396,"about_ca_system_score_gemma":0.05729561,"threshold_uncertainty_score":0},"labels":[{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"design_other","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"grok","categories":["metaresearch","scholarly_communication","open_science"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"opus","categories":["open_science","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":"W6908435990","doi":"10.26190/unsworks/26323","title":"Multi-site research and data sharing","year":2017,"lang":"en","type":"article","venue":"Open MIND","topic":"Ethics in Clinical Research","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Data sharing; Data governance; Information governance; Context (archaeology); Pooling; Data Protection Act 1998; Data access; Health care; Information privacy; Process (computing)","score_opus":0.9658159765779182,"score_gpt":0.7731835553437635,"score_spread":0.1926324212341547,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6908435990","genre_codex":"methods","genre_gemma":"empirical","domain_codex":"methods","domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.13031818,0.014283583,0.418438,0.19563478,0.0040158527,0.021751005,0.001943393,0.0010049614,0.2126102],"genre_scores_gemma":[0.7068319,0.004532634,0.22459425,0.023166813,0.0013447129,0.024411656,0.00070643716,0.0004994164,0.013912219],"study_design_codex":"qualitative","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.3002121,0.63250476,0.014504696,0.02073105,0.026800444,0.0052470295],"domain_scores_gemma":[0.3394819,0.41097733,0.039736297,0.15172538,0.03311145,0.024967598],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.39490992,0.00079501024,0.0022836807,0.0051261047,0.016698407,0.021051208,0.0069185523,0.005435361,0.012512592],"category_scores_gemma":[0.33021984,0.0017107544,0.0018835873,0.007617475,0.036249578,0.01806539,0.03495725,0.009532726,0.0025888374],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005997066,0.0005588457,0.02342497,0.0066547734,0.0008265681,0.0023661302,0.3867036,0.002130799,0.0020298322,0.31604952,0.034391176,0.22426403],"study_design_scores_gemma":[0.00025740653,0.0007377321,0.009924499,0.011701053,0.00024967472,0.0024896348,0.1643451,0.002587021,0.0017437671,0.3023971,0.50332266,0.00024437087],"about_ca_topic_score_codex":0.0064744833,"about_ca_topic_score_gemma":0.0075404476,"teacher_disagreement_score":0.99308145,"about_ca_system_score_codex":0.012211208,"about_ca_system_score_gemma":0.056932256,"threshold_uncertainty_score":0.7461838},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","scholarly_communication","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6912570129","doi":"10.5281/zenodo.4117503","title":"Taking FAIR and open science to the field. The evolution of the PFTC field course","year":2020,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université de Montréal","funders":"","keywords":"Course (navigation); Field (mathematics); Massive open online course; Presentation (obstetrics); Science education; Open science","score_opus":0.08259691000852912,"score_gpt":0.329608354515012,"score_spread":0.2470114445064829,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6912570129","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05982418,0.011524195,0.03286208,0.4400944,0.01845516,0.00063488184,0.00091894495,0.002114175,0.43357205],"genre_scores_gemma":[0.39117563,0.005341212,0.036486443,0.032919966,0.0041091214,0.00068658683,0.001525307,0.0009706964,0.5267851],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9954678,0.00089048455,0.000104648156,0.0010112548,0.0014948734,0.0010309626],"domain_scores_gemma":[0.97563374,0.0019282158,0.0009212318,0.00076845055,0.0029428895,0.01780546],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.009814177,0.00047276184,0.00025116428,0.0015563989,0.004891388,0.008943148,0.0019736127,0.0035468927,0.038415886],"category_scores_gemma":[0.011673992,0.0003696788,0.00027258933,0.0011048212,0.004064584,0.006812537,0.010153998,0.0055835564,0.008423549],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011706654,0.00047956692,0.004056127,0.00023021299,0.0000035983353,0.00029357194,0.0072132926,0.00050393195,0.0013492105,0.063460566,0.44080868,0.48148417],"study_design_scores_gemma":[0.000022853048,0.00016578691,0.0077422652,0.00041442996,0.0000024906606,0.00030411655,0.0048105195,0.00042604617,0.00046832307,0.023786528,0.96182746,0.000029160694],"about_ca_topic_score_codex":0.015433999,"about_ca_topic_score_gemma":0.039113097,"teacher_disagreement_score":0.9980264,"about_ca_system_score_codex":0.011090658,"about_ca_system_score_gemma":0.024910156,"threshold_uncertainty_score":0.12851393},"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"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6912635752","doi":"10.5281/zenodo.4320095","title":"Workshop Report: OCSDNet Imagining a Feminist Open Science","year":2018,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"ICT in Developing Communities","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Technoscience; Variety (cybernetics); Situated; Knowledge production; Publishing; Citizen science; Feminism; Politics","score_opus":0.07459541764920144,"score_gpt":0.3157523900562904,"score_spread":0.24115697240708897,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6912635752","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07074335,0.0075707715,0.028213281,0.368345,0.051010612,0.0023827683,0.0051339963,0.0004936848,0.4661065],"genre_scores_gemma":[0.36133638,0.0054040803,0.030029062,0.029898653,0.005745042,0.0028360202,0.0044162394,0.0009502233,0.5593843],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9962174,0.0015379566,0.00008357097,0.0005498018,0.0009734959,0.0006376967],"domain_scores_gemma":[0.9937423,0.0019487065,0.00018855034,0.000282726,0.0010876742,0.0027499031],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.011645917,0.0005620085,0.00021875394,0.0006849147,0.0057608364,0.0058799232,0.0014202002,0.0035532115,0.027603988],"category_scores_gemma":[0.0071432316,0.00031240666,0.0004925041,0.00087634654,0.0031887963,0.0040548695,0.00733782,0.0051135616,0.004083716],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023717244,0.00018237543,0.0012181093,0.00046322483,0.000010367845,0.0012681041,0.035726096,0.00073387125,0.0033843147,0.10430973,0.7852883,0.06717842],"study_design_scores_gemma":[0.000014344826,0.000039082035,0.0004970368,0.00014653492,0.0000022599443,0.00013930458,0.013283197,0.0001370745,0.00050438585,0.0047518876,0.98047066,0.000014357828],"about_ca_topic_score_codex":0.016659087,"about_ca_topic_score_gemma":0.035539877,"teacher_disagreement_score":0.9985798,"about_ca_system_score_codex":0.0064631803,"about_ca_system_score_gemma":0.012695921,"threshold_uncertainty_score":0.09234452},"labels":[{"model":"gemma","categories":["open_science","sts"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6920462805","doi":"10.6084/m9.figshare.12941738","title":"Community Funding for Open Science Infrastructure: SCOSS 2 years in","year":2020,"lang":"en","type":"other","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Euros; Sustainability; Outreach; Resilience (materials science); Service (business); Publishing; Interim; Christian ministry; Scale (ratio)","score_opus":0.16292839901788964,"score_gpt":0.38175956470146655,"score_spread":0.2188311656835769,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6920462805","genre_codex":"commentary","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.072754025,0.01888123,0.0037671006,0.5315097,0.024751272,0.0015742254,0.0059632254,0.003328498,0.3374708],"genre_scores_gemma":[0.31248882,0.008202455,0.0109985825,0.17296885,0.00526417,0.0023158074,0.013020583,0.0021928728,0.4725478],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98099005,0.00212993,0.0005807017,0.0007931148,0.0077881226,0.0077179633],"domain_scores_gemma":[0.9421655,0.0032582725,0.001731688,0.0017529487,0.0075656935,0.04352603],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.025827264,0.00095161627,0.00055139966,0.0035514864,0.007677258,0.016611885,0.0024352332,0.010852226,0.03568942],"category_scores_gemma":[0.030034108,0.0005831062,0.0011593904,0.004149537,0.0038151154,0.009273689,0.01979795,0.008082999,0.010716689],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002507098,0.00019568285,0.0039483607,0.0006214146,0.000037621765,0.0007168435,0.0020300765,0.00038974473,0.0008175176,0.053120337,0.85254186,0.08532982],"study_design_scores_gemma":[0.000067447196,0.00006070737,0.0050341957,0.00043427537,0.000006894365,0.00009279829,0.0012767928,0.000112729926,0.00019104635,0.0026589467,0.99003416,0.000029998295],"about_ca_topic_score_codex":0.07916403,"about_ca_topic_score_gemma":0.13286267,"teacher_disagreement_score":0.9975648,"about_ca_system_score_codex":0.015094346,"about_ca_system_score_gemma":0.09162374,"threshold_uncertainty_score":0.15740651},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6925515292","doi":"10.17605/osf.io/yk35t","title":"Face Puzzle Study 4 Preregistration","year":2022,"lang":"en","type":"other","venue":"Open Science Framework","topic":"Reproductive Biology and Fertility","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Facial expression; Neurotypical; Emotion classification; Boredom; Everyday life; Set (abstract data type); Affective science; Facial Action Coding System; Stimulus (psychology); Jealousy","score_opus":0.05093198846453724,"score_gpt":0.3936288109182572,"score_spread":0.34269682245371996,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6925515292","genre_codex":"empirical","genre_gemma":"protocol","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"protocol","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.73459256,0.0020725091,0.022008538,0.0038378646,0.003637457,0.034466173,0.028677724,0.0015065823,0.16920057],"genre_scores_gemma":[0.65416217,0.0009001418,0.023161821,0.0071417396,0.001173619,0.08824327,0.022273794,0.0015097788,0.20143364],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99679404,0.00060243264,0.00037530813,0.0009822369,0.0009267066,0.0003192872],"domain_scores_gemma":[0.99305266,0.0012666908,0.00040090093,0.0025494883,0.0020844354,0.0006456775],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0053487606,0.0012356638,0.0015142902,0.0011152119,0.0024390623,0.002557818,0.002431845,0.0030067032,0.08871411],"category_scores_gemma":[0.014167024,0.0006136634,0.0012049834,0.00059713697,0.0009229708,0.0019517025,0.0024145285,0.0023817397,0.040738415],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0453846,0.034125816,0.07521918,0.0026833941,0.00045225158,0.01124045,0.029262021,0.001309622,0.036991365,0.024826145,0.23723346,0.5012717],"study_design_scores_gemma":[0.0056597427,0.018609224,0.22516495,0.0008195606,0.0004928307,0.007017642,0.011732014,0.0020468226,0.016880164,0.0210335,0.6901706,0.0003729288],"about_ca_topic_score_codex":0.0046919235,"about_ca_topic_score_gemma":0.005494124,"teacher_disagreement_score":0.08871411,"about_ca_system_score_codex":0.0014986413,"about_ca_system_score_gemma":0.0025197768,"threshold_uncertainty_score":0.29677826},"labels":[{"model":"gemma","categories":["metaresearch"],"domain":"reproducibility","study_design":"not_applicable","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"design_other","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6925537222","doi":"10.17605/osf.io/hs4aw","title":"Questionable Research Practices, the Replication Crisis and Open Science: A National Survey of Canadian Psychology Students","year":2019,"lang":"en","type":"article","venue":"OSF Preprints (OSF Preprints)","topic":"Gothic Literature and Media Analysis","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Replication (statistics); Psychological research; History of psychology; Survey research","score_opus":0.11030152008289403,"score_gpt":0.479019258158841,"score_spread":0.368717738075947,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6925537222","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"methods","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98887175,0.0006515906,0.0000997997,0.006070958,0.000026337022,0.00013229788,0.0004186022,0.000007568955,0.003721139],"genre_scores_gemma":[0.99659437,0.00065016974,0.00017745874,0.0010859155,0.000013581879,0.000059650552,0.00018592889,0.000011512389,0.0012213637],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9714461,0.0051922086,0.0025822015,0.0017440618,0.0139386775,0.0050966945],"domain_scores_gemma":[0.8183807,0.040677432,0.043871798,0.006945469,0.0514939,0.03863066],"candidate_categories":["metaresearch","open_science","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.025920762,0.00028850738,0.00078549265,0.008883459,0.011648327,0.0076976754,0.003918172,0.0029517398,0.004035988],"category_scores_gemma":[0.10440788,0.0010062165,0.0004758053,0.013805962,0.0076756603,0.0034330494,0.005036386,0.0048671425,0.00039666577],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000112770904,0.00037558662,0.8665202,0.0001467001,0.000044658816,0.00018964733,0.111790076,0.00005945659,0.00033453727,0.0014218526,0.0040702354,0.014934306],"study_design_scores_gemma":[0.000016959017,0.00007294939,0.83685654,0.00020905245,0.00002197571,0.00010981391,0.15428327,0.00019279413,0.000096645745,0.00037245918,0.0076960875,0.00007143874],"about_ca_topic_score_codex":0.9555382,"about_ca_topic_score_gemma":0.96765465,"teacher_disagreement_score":0.99704826,"about_ca_system_score_codex":0.040745515,"about_ca_system_score_gemma":0.13485162,"threshold_uncertainty_score":0.29563075},"labels":[{"model":"gemma","categories":["metaresearch","open_science","research_integrity"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science","research_integrity"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6930459529","doi":"10.5281/zenodo.10855046","title":"Transform to Open Science Mission November 2023 Community Panel Summary Report","year":2024,"lang":"en","type":"report","venue":"Bristol Research (University of Bristol)","topic":"Digital Radiography and Breast Imaging","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of British Columbia","funders":"","keywords":"TOPS; Open science; Suite; Citizen science; Open data; Open source","score_opus":0.21541844540512234,"score_gpt":0.4224116574220172,"score_spread":0.20699321201689486,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6930459529","genre_codex":"other","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004705401,0.0023850007,0.002480859,0.09953848,0.02021549,0.002562553,0.1380689,0.0015877987,0.7284554],"genre_scores_gemma":[0.012653825,0.0032389045,0.0037387868,0.023787392,0.002845901,0.00224353,0.09847494,0.0008841201,0.8521326],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9946768,0.0003387932,0.000110584726,0.00020556178,0.0038852205,0.00078296935],"domain_scores_gemma":[0.9881357,0.0010034576,0.0003462807,0.00037214113,0.007265076,0.002877247],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0069811023,0.00092859,0.00029712077,0.001680521,0.0026355768,0.0045265835,0.0016136587,0.005491097,0.13036007],"category_scores_gemma":[0.007913039,0.00043490776,0.00064985926,0.0019317176,0.00045999183,0.0022597075,0.0030420201,0.003515363,0.06809599],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000014962899,0.000026554251,0.00020483433,0.000030790343,9.829005e-7,0.000026917043,0.000013487939,0.000033570344,0.00006606834,0.00034195636,0.9950323,0.0042075654],"study_design_scores_gemma":[0.000010922528,0.000017896818,0.0018176433,0.000051576935,0.0000015375063,0.000019233139,0.000110248846,0.00003871852,0.00015845447,0.00019984554,0.997566,0.0000079172005],"about_ca_topic_score_codex":0.05494206,"about_ca_topic_score_gemma":0.08361584,"teacher_disagreement_score":0.9983863,"about_ca_system_score_codex":0.0038737971,"about_ca_system_score_gemma":0.026614673,"threshold_uncertainty_score":0.43609786},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W6930580054","doi":"10.5281/zenodo.13135636","title":"Interviews with editors of library science journals on transitioning to open access","year":2024,"lang":"en","type":"dataset","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Bacteriophages and microbial interactions","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Saskatchewan","funders":"","keywords":"Confidentiality; Interview; Information science; Access to information; Codebook","score_opus":0.057315166518145375,"score_gpt":0.336296687712855,"score_spread":0.2789815211947096,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6930580054","genre_codex":"empirical","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9565906,0.0006363621,0.0056514433,0.008465814,0.0005301587,0.003290901,0.004383355,0.00020697122,0.020244403],"genre_scores_gemma":[0.9227501,0.00205994,0.0174787,0.00588199,0.00043547337,0.017619187,0.0022544852,0.00021977146,0.0313003],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9814565,0.012883959,0.0012485543,0.0008940349,0.0017363298,0.0017805282],"domain_scores_gemma":[0.816732,0.15377036,0.010563129,0.0027419464,0.008403588,0.007788945],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.027570732,0.0006916925,0.00081019744,0.0033982287,0.011933331,0.0057904683,0.0019268556,0.0033557443,0.013021448],"category_scores_gemma":[0.09625812,0.0011392073,0.00045459968,0.0058578555,0.006205637,0.005385058,0.0051387944,0.00417828,0.00218279],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001887858,0.00015963703,0.00315515,0.00028379943,0.000003762339,0.0007192229,0.97741026,0.00006879464,0.00085242646,0.0008381846,0.0071643665,0.009155655],"study_design_scores_gemma":[0.00004811518,0.00016757338,0.0044354494,0.00032772656,0.0000051959732,0.00021547877,0.9562973,0.000098925644,0.0006314156,0.00067244103,0.037061434,0.000039046994],"about_ca_topic_score_codex":0.0053448677,"about_ca_topic_score_gemma":0.016469296,"teacher_disagreement_score":0.99807316,"about_ca_system_score_codex":0.008867854,"about_ca_system_score_gemma":0.008308574,"threshold_uncertainty_score":0.14580977},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"qualitative","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6931685785","doi":"10.5281/zenodo.7542374","title":"Diversity and Inclusion at Belle II","year":2022,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Archaeology and Rock Art Studies","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Nucleofection; Gestational period; TSG101; Dysgeusia; Diafiltration; Liquation; Hyporeflexia; Fusible alloy","score_opus":0.03626538391725857,"score_gpt":0.25508148482609677,"score_spread":0.21881610090883818,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6931685785","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6963424,0.0027218673,0.004087624,0.038675126,0.00065696484,0.00023400073,0.00033437595,0.00012121757,0.25682646],"genre_scores_gemma":[0.92308027,0.0004681305,0.0024018805,0.0020154011,0.00015208237,0.00022977467,0.00014652369,0.00007501854,0.07143096],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.98796034,0.005819176,0.00023466926,0.0007982694,0.0018606043,0.003326895],"domain_scores_gemma":[0.9891528,0.0023954087,0.0007452384,0.00065550586,0.0011260955,0.0059250165],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.011958517,0.00034493933,0.00042472588,0.0025885135,0.058781974,0.0121074505,0.0015072167,0.0020464475,0.035212066],"category_scores_gemma":[0.008004342,0.00042635374,0.00019700441,0.002697909,0.008583765,0.0052895723,0.035315543,0.003619066,0.0023434567],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00023924073,0.0003280979,0.028068976,0.00016814655,0.000012573139,0.0013363172,0.68152934,0.00008453778,0.002231936,0.095555685,0.054566003,0.1358792],"study_design_scores_gemma":[0.000015243676,0.00013458074,0.034367036,0.00021665085,0.000006963254,0.00082080916,0.18085708,0.00014576357,0.0007795843,0.009841106,0.7727689,0.000046235575],"about_ca_topic_score_codex":0.016160853,"about_ca_topic_score_gemma":0.031774156,"teacher_disagreement_score":0.058781974,"about_ca_system_score_codex":0.0053215455,"about_ca_system_score_gemma":0.010078267,"threshold_uncertainty_score":0.11779612},"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":["sts"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6939786871","doi":"10.6084/m9.figshare.877095","title":"Open science Make your work known!","year":2013,"lang":"en","type":"other","venue":"Figshare","topic":"Mycorrhizal Fungi and Plant Interactions","field":"Agricultural and Biological Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Work (physics); Open science; Citizen science; Biodiversity","score_opus":0.0523958966899383,"score_gpt":0.2632203408658347,"score_spread":0.21082444417589638,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6939786871","genre_codex":"other","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0002609976,0.00093159283,0.0008055055,0.0063245688,0.0055923113,0.00018970258,0.0051246015,0.00265848,0.9781123],"genre_scores_gemma":[0.0005034674,0.00027972652,0.00027677207,0.00060430757,0.0003070013,0.000040742125,0.0009511899,0.0006105095,0.99642617],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985317,0.00007664627,0.000025691385,0.00009946386,0.0009990688,0.00026745646],"domain_scores_gemma":[0.99106956,0.0005293171,0.00022235735,0.0008733708,0.0042572613,0.0030482197],"candidate_categories":["open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.002194857,0.0010315257,0.0008378126,0.0029381881,0.005351752,0.007831884,0.0014752974,0.0026647572,0.8129485],"category_scores_gemma":[0.0069674137,0.00074777956,0.000940687,0.003160207,0.0017689352,0.0037529184,0.0038674835,0.003608292,0.66915774],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000046461178,0.000005370361,0.000035206038,0.000014507527,4.4175698e-7,0.0000060298357,0.000006919758,0.000008018835,0.000032489137,0.00042989425,0.9883068,0.011149567],"study_design_scores_gemma":[0.0000043277532,0.0000027750289,0.000512537,0.000034765904,6.810047e-7,0.000007889542,0.0000380466,0.000018017006,0.00004139687,0.00042231553,0.9989115,0.000005733314],"about_ca_topic_score_codex":0.16042604,"about_ca_topic_score_gemma":0.52536035,"teacher_disagreement_score":0.9985247,"about_ca_system_score_codex":0.005813361,"about_ca_system_score_gemma":0.011466872,"threshold_uncertainty_score":0.31898445},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6939789161","doi":"10.6084/m9.figshare.28630034.v1","title":"Additional file 2 of Evidence of the impacts of pharmaceuticals on aquatic animal behaviour (EIPAAB): a systematic map and open access database","year":2025,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta; Fisheries and Oceans Canada","funders":"","keywords":"Open water; Data extraction; Public access; Aquatic environment; Access to information","score_opus":0.3279255493646885,"score_gpt":0.4830641839400704,"score_spread":0.1551386345753819,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6939789161","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000018088043,0.00001664057,0.000020998772,0.00003175695,0.0000054912266,0.000020945212,0.9996569,0.000026799224,0.00020236129],"genre_scores_gemma":[0.0012151941,0.00017735813,0.0010588726,0.00021198079,0.000021559732,0.001156786,0.9939711,0.00015133138,0.0020358125],"study_design_codex":"not_applicable","study_design_gemma":"systematic_review","domain_scores_codex":[0.9984346,0.00027062773,0.000403889,0.00038339183,0.00028599583,0.00022153644],"domain_scores_gemma":[0.98092514,0.012443232,0.0015562075,0.0012714072,0.0031910439,0.00061301846],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0026749976,0.0013807998,0.0017086872,0.005893783,0.00082338724,0.0021513524,0.002468543,0.0019688986,0.6350238],"category_scores_gemma":[0.029487517,0.00078162673,0.0018930321,0.0108646145,0.00042818865,0.0024231293,0.0024356872,0.0013199421,0.08798307],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010873092,0.000021489124,0.0009654433,0.01305335,0.00010035408,0.000025654394,0.00005236711,0.00025527668,0.00007757277,0.0007731509,0.9813706,0.0031960646],"study_design_scores_gemma":[0.00237753,0.00007944934,0.014384841,0.00937975,0.0004043106,0.00011937921,0.00030982366,0.00039882737,0.0002898809,0.0040781093,0.96807915,0.0000988997],"about_ca_topic_score_codex":0.031154612,"about_ca_topic_score_gemma":0.041040406,"teacher_disagreement_score":0.6350238,"about_ca_system_score_codex":0.0019590606,"about_ca_system_score_gemma":0.0049795047,"threshold_uncertainty_score":0.52059424},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6944426372","doi":"10.17632/wsy7x9fgxz.1","title":"Replication package for 'A Macroeconomic Perspective on Taxing Multinational Enterprises'","year":2024,"lang":"en","type":"dataset","venue":"Data Archiving and Networked Services (DANS)","topic":"Global trade and economics","field":"Economics, Econometrics and Finance","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Multinational corporation; Perspective (graphical); Replication (statistics); Mechanism (biology)","score_opus":0.045593567221856555,"score_gpt":0.2733814404120441,"score_spread":0.22778787319018753,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6944426372","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000065104236,0.000029313036,0.0002904294,0.00018650625,0.00013054165,0.00006001844,0.9979754,0.0005598937,0.0007027937],"genre_scores_gemma":[0.0010071236,0.000075500284,0.0018979738,0.0003222164,0.00006823843,0.0011491292,0.9929051,0.0005309508,0.0020438512],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.995106,0.0013632675,0.00079416274,0.0011356128,0.0009981262,0.0006028667],"domain_scores_gemma":[0.97019523,0.011079522,0.0021695308,0.008832964,0.0063845464,0.0013382506],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.007623746,0.001676536,0.002165151,0.0044754767,0.0012817413,0.0040742625,0.0047190567,0.0033923748,0.29320887],"category_scores_gemma":[0.055287447,0.0013626951,0.0036713977,0.008913375,0.00087238837,0.0026191256,0.003755291,0.0034022469,0.14495915],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008281258,0.00001840872,0.0004504205,0.0004918566,0.00005741941,0.000012957551,0.00002134085,0.00019886026,0.00003199266,0.0007883861,0.9962962,0.0015493057],"study_design_scores_gemma":[0.0013870236,0.000036404588,0.0049473,0.0006991357,0.00010458183,0.000060211587,0.000081855665,0.0006213402,0.00027149805,0.004677985,0.9870309,0.000081732185],"about_ca_topic_score_codex":0.028356105,"about_ca_topic_score_gemma":0.036805775,"teacher_disagreement_score":0.29320887,"about_ca_system_score_codex":0.0018897995,"about_ca_system_score_gemma":0.006702009,"threshold_uncertainty_score":0.9808814},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6949824102","doi":"10.5281/zenodo.4573320","title":"Modèle de plan de gestion des données : Méthodes mixtes (enquêtes et recherche qualitative)","year":2021,"lang":"fr","type":"other","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Plan (archaeology); Context (archaeology); Pilotage","score_opus":0.4711445725229463,"score_gpt":0.3691671399120419,"score_spread":0.10197743261090436,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6949824102","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0070328615,0.00078307174,0.9543317,0.002599785,0.00010690636,0.0029894249,0.002002454,0.0008294235,0.029324375],"genre_scores_gemma":[0.07681053,0.0008014638,0.89966965,0.00032736536,0.00003074476,0.0060329195,0.0016215424,0.00034238663,0.014363447],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.94392353,0.036526397,0.003820304,0.004858183,0.010088021,0.00078349543],"domain_scores_gemma":[0.92644495,0.050816514,0.003457381,0.00829226,0.010181956,0.00080689706],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.044239018,0.0015355825,0.0009958327,0.005813796,0.0023497897,0.012218561,0.002732214,0.0023094248,0.01949758],"category_scores_gemma":[0.057118203,0.0015763779,0.0021097842,0.006624301,0.005076373,0.010213767,0.0055687153,0.002758004,0.0043671275],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002766243,0.00021765141,0.0046927817,0.0041289115,0.00024234893,0.00029856712,0.044702176,0.01990182,0.0040746327,0.566872,0.013969333,0.34062314],"study_design_scores_gemma":[0.00026245127,0.00031403103,0.0048731863,0.0048021777,0.00027140803,0.0005843553,0.026786376,0.075006425,0.010991544,0.3716644,0.50413585,0.00030780793],"about_ca_topic_score_codex":0.016519157,"about_ca_topic_score_gemma":0.012976194,"teacher_disagreement_score":0.9972678,"about_ca_system_score_codex":0.010079924,"about_ca_system_score_gemma":0.01878325,"threshold_uncertainty_score":0},"labels":[{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"opus","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W6950331135","doi":"10.5683/sp3/soe0nm","title":"Replication Data and Code for: Traditional Indigenous foodways and retail subsidies: Evidence from the Northwest Territories Community Survey and Nutrition North Canada","year":2025,"lang":"en","type":"dataset","venue":"Borealis","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Indigenous; Foodways; Code (set theory); Replication (statistics); Replicate","score_opus":0.15424749526074283,"score_gpt":0.29889211185872483,"score_spread":0.144644616597982,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6950331135","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0001778742,0.000036352056,0.00011981677,0.0001150671,0.00005889497,0.00022309764,0.99789065,0.00010038407,0.0012778688],"genre_scores_gemma":[0.0029974654,0.00009874939,0.0013556649,0.00020721326,0.000026425056,0.0037039449,0.9822597,0.00025072042,0.009100071],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99496645,0.00083134166,0.00070566486,0.0008881067,0.0016483909,0.0009600851],"domain_scores_gemma":[0.96574557,0.004685131,0.0018471853,0.005590575,0.020238154,0.0018932836],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0077953343,0.00192863,0.0019232194,0.005622683,0.004152758,0.0037356138,0.0051155807,0.0017007492,0.17766604],"category_scores_gemma":[0.052174777,0.0017562413,0.00263002,0.013638676,0.0011610499,0.0014740609,0.0022213536,0.0030006787,0.042777773],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007336496,0.000016029966,0.0019364323,0.00033087845,0.000055594304,0.000013039419,0.00007545705,0.00017332914,0.000016613758,0.0005684431,0.99515027,0.0015906449],"study_design_scores_gemma":[0.0019525503,0.0000383228,0.05637448,0.0015259648,0.0003147843,0.00007268485,0.000826437,0.0005132421,0.00029664487,0.002125333,0.9358029,0.00015673389],"about_ca_topic_score_codex":0.9420929,"about_ca_topic_score_gemma":0.95981723,"teacher_disagreement_score":0.17766604,"about_ca_system_score_codex":0.018136498,"about_ca_system_score_gemma":0.057277903,"threshold_uncertainty_score":0.5943521},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"medium"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W6962692740","doi":"10.17026/ls/lrf63d","title":"Replication Data for Perceptions of research integrity and open science practices: a survey of Brazilian dental researchers","year":2024,"lang":"en","type":"dataset","venue":"DANS Data Station Life Sciences","topic":"Academic integrity and plagiarism","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ottawa Hospital","funders":"","keywords":"Perception; Research integrity; Academic integrity; Open science; Dental research; Research ethics; Replication (statistics); Computer-assisted web interviewing","score_opus":0.6140305414550996,"score_gpt":0.6205478929486535,"score_spread":0.0065173514935539245,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6962692740","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0401082,0.0002699842,0.0023490088,0.00084722374,0.000120769495,0.0019741138,0.94553643,0.00033959103,0.008454771],"genre_scores_gemma":[0.07602076,0.00031552784,0.009803141,0.00055830705,0.00006034266,0.032241113,0.87480634,0.00034313963,0.0058513843],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9813301,0.006479521,0.003191486,0.0022007343,0.0055420664,0.0012561154],"domain_scores_gemma":[0.91693395,0.031816475,0.009542517,0.016180221,0.023825316,0.001701475],"candidate_categories":["metaresearch","open_science","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.02418175,0.0010175825,0.0009835454,0.005353339,0.0015945975,0.0019048897,0.0018948033,0.0010877661,0.017232439],"category_scores_gemma":[0.107565895,0.00071652903,0.0011634307,0.01199433,0.00078178686,0.0011085907,0.0025307923,0.0016353163,0.009657498],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007497068,0.000510241,0.16562298,0.004466237,0.00034016278,0.0002165034,0.0051109428,0.0009857051,0.0013707841,0.003466867,0.768318,0.048841845],"study_design_scores_gemma":[0.0008581763,0.00021769042,0.4241086,0.001378604,0.00023779049,0.00018518206,0.005859925,0.0009795369,0.0016241479,0.0018150535,0.5626135,0.000121786696],"about_ca_topic_score_codex":0.059459466,"about_ca_topic_score_gemma":0.064214885,"teacher_disagreement_score":0.9989122,"about_ca_system_score_codex":0.0029555906,"about_ca_system_score_gemma":0.0077552027,"threshold_uncertainty_score":0.12788689},"labels":[{"model":"gemma","categories":["metaresearch","open_science","research_integrity"],"domain":"methods","study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science","research_integrity"],"domain":"methods","study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6967477644","doi":"10.5281/zenodo.10600863","title":"Building an Open Science Monitoring Framework with open technologies","year":2024,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Université de Montréal","funders":"","keywords":"Openness to experience; Work (physics); Open data; Open science; Process (computing); International community; Open innovation; Citizen science","score_opus":0.12308604359975563,"score_gpt":0.3788614662119514,"score_spread":0.25577542261219577,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6967477644","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00091436936,0.0006600876,0.9492894,0.014559614,0.0005880333,0.0011011353,0.0009011348,0.010390844,0.021595422],"genre_scores_gemma":[0.025924657,0.0011775956,0.9549744,0.0021349585,0.0007069277,0.0020298786,0.0042939493,0.0025366878,0.006220896],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.86589956,0.06285462,0.016803125,0.013363627,0.035902172,0.005176938],"domain_scores_gemma":[0.7904227,0.09000603,0.014426719,0.050900843,0.04387461,0.010369116],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.18532804,0.0026602256,0.0028493272,0.02055974,0.0072582443,0.037350606,0.009392652,0.011691533,0.011529192],"category_scores_gemma":[0.16284318,0.0028706533,0.0053143194,0.016009763,0.015139109,0.06395466,0.03255171,0.014904782,0.009958081],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005971086,0.00012392725,0.0015093212,0.000584352,0.00008233414,0.00022553865,0.0014943249,0.009831152,0.0005428599,0.88945186,0.024622325,0.071472265],"study_design_scores_gemma":[0.00005052126,0.000053092917,0.00048605865,0.0015333305,0.00004809311,0.00016114497,0.00063632464,0.021970257,0.0011777163,0.5504446,0.4233086,0.00013023175],"about_ca_topic_score_codex":0.012964296,"about_ca_topic_score_gemma":0.006509855,"teacher_disagreement_score":0.9906073,"about_ca_system_score_codex":0.010712581,"about_ca_system_score_gemma":0.032326892,"threshold_uncertainty_score":0.98012006},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"evaluation","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W6967774443","doi":"10.5281/zenodo.14004728","title":"Open Data The Need of TNR","year":2024,"lang":"en","type":"other","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Bishop's University","funders":"","keywords":"Open data; Key (lock); Context (archaeology); Work (physics)","score_opus":0.11676621238289313,"score_gpt":0.3184301352311905,"score_spread":0.20166392284829737,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6967774443","genre_codex":"other","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016285701,0.0013385902,0.031929903,0.04304719,0.0088749,0.00031643102,0.02286414,0.014790393,0.8752099],"genre_scores_gemma":[0.021985691,0.0015411082,0.030619724,0.019303061,0.003800435,0.00046039806,0.057109665,0.020922277,0.8442577],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99284124,0.0014407175,0.00040167882,0.00087795954,0.0035139285,0.00092436676],"domain_scores_gemma":[0.96695864,0.0060777566,0.001159698,0.010218681,0.011088073,0.0044970494],"candidate_categories":["open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.010404627,0.0010429765,0.0010884202,0.0022997493,0.0017874567,0.009023827,0.003742875,0.005893686,0.33535033],"category_scores_gemma":[0.026859023,0.00062241463,0.0011576158,0.0027885085,0.0019296373,0.010245343,0.0074214535,0.005108116,0.26104838],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00013925723,0.00004502501,0.00016963814,0.00020648612,0.000008451514,0.00009148027,0.00007116851,0.0002391533,0.00088660733,0.08944009,0.85555816,0.05314461],"study_design_scores_gemma":[0.000019211748,0.000011215608,0.00010717016,0.000072728755,0.000003414277,0.000054084237,0.000029122106,0.00047076758,0.00041449873,0.011323605,0.9874835,0.000010826187],"about_ca_topic_score_codex":0.0105434675,"about_ca_topic_score_gemma":0.0072236243,"teacher_disagreement_score":0.9962571,"about_ca_system_score_codex":0.0030997174,"about_ca_system_score_gemma":0.0064006452,"threshold_uncertainty_score":0.9480421},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6968483934","doi":"10.5281/zenodo.15363243","title":"The Open Science approach of the Intergovernmental Panel on Climate Change (IPCC)","year":2025,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Ouranos","funders":"","keywords":"Framing (construction); Climate change; Downscaling; Documentation; Transparency (behavior); Open science; Open data; Citizen science","score_opus":0.14754600228241008,"score_gpt":0.3269886946217394,"score_spread":0.17944269233932933,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6968483934","genre_codex":"methods","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"review","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0022818106,0.010120284,0.54015124,0.15198821,0.013359231,0.002172366,0.005186507,0.0021607971,0.27257958],"genre_scores_gemma":[0.091728516,0.022806896,0.72550833,0.04743576,0.0076343883,0.010905246,0.009729283,0.0031115953,0.08113994],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.8661867,0.06316658,0.008267894,0.010579221,0.04844107,0.0033585588],"domain_scores_gemma":[0.86283857,0.05912352,0.007099547,0.037449144,0.029503038,0.0039860597],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.100938074,0.0018560198,0.0016290954,0.010564381,0.006071563,0.025263965,0.006299859,0.013878918,0.010942668],"category_scores_gemma":[0.096527085,0.0012835182,0.0032074207,0.015355611,0.021900145,0.024666512,0.020225396,0.021551602,0.0054185926],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000022427817,0.000039742143,0.00023697027,0.00056941406,0.00004596595,0.000044075285,0.0020576497,0.0011307688,0.00028434506,0.8926767,0.05699052,0.045901477],"study_design_scores_gemma":[0.000013810009,0.000020740037,0.00034952318,0.00073551113,0.000025483621,0.000059051443,0.00050966005,0.0005514306,0.00036739677,0.36012307,0.6371949,0.00004948088],"about_ca_topic_score_codex":0.018731363,"about_ca_topic_score_gemma":0.00904424,"teacher_disagreement_score":0.99370015,"about_ca_system_score_codex":0.01435116,"about_ca_system_score_gemma":0.047693186,"threshold_uncertainty_score":0.53381795},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W6968933212","doi":"10.5281/zenodo.5655304","title":"Developing a REPO Assessment Workbook: Supporting Open Science Communities in the Transition to Online Learning","year":2021,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Lethbridge","funders":"","keywords":"Open science; Citizen science; Community of practice; Citizen journalism; Workbook; Open education; Best practice; Open learning; Open data; Online community","score_opus":0.21707289174215938,"score_gpt":0.44497745998982546,"score_spread":0.22790456824766608,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6968933212","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18912543,0.000994693,0.65547967,0.024096143,0.0019651225,0.0160554,0.0019646082,0.016060866,0.09425808],"genre_scores_gemma":[0.17861028,0.00089334784,0.75611347,0.0011132195,0.00024026916,0.0058152797,0.0017408178,0.0014283066,0.05404503],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9767958,0.013969525,0.0011852216,0.00203106,0.004792527,0.0012257342],"domain_scores_gemma":[0.92793465,0.041439608,0.0031386574,0.009428449,0.008427341,0.00963124],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.049099345,0.0016984587,0.00094672653,0.0029430967,0.0035515688,0.008662737,0.0053890105,0.0028770962,0.009912944],"category_scores_gemma":[0.05676163,0.0010248687,0.0008928155,0.0016382813,0.0044475584,0.01162936,0.012885138,0.005760891,0.0055998387],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00035850468,0.0053139087,0.0049929596,0.0015161711,0.000045715486,0.0013591946,0.12300728,0.005605005,0.015320646,0.024382723,0.0803291,0.7377689],"study_design_scores_gemma":[0.00039109544,0.0031807662,0.008035717,0.0018476549,0.000057790865,0.0014627925,0.055721905,0.013221362,0.01858218,0.03956764,0.8573328,0.000598354],"about_ca_topic_score_codex":0.00210802,"about_ca_topic_score_gemma":0.0053790137,"teacher_disagreement_score":0.99461097,"about_ca_system_score_codex":0.0028464738,"about_ca_system_score_gemma":0.010918383,"threshold_uncertainty_score":0.25966525},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"agree"},{"id":"W6969203973","doi":"10.5683/sp3/khdjwu","title":"Replication data for the study: \"A novel approach to the assessment of motor-unit identification accuracy and its radical incongruence with conventional methodology\" in the Journal of Neurophysiology","year":2025,"lang":"en","type":"dataset","venue":"Borealis","topic":"Agricultural Economics and Policy","field":"Agricultural and Biological Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Brock University","funders":"","keywords":"Replication (statistics); Identification (biology); Neurophysiology; Clinical neurophysiology","score_opus":0.18597236858280874,"score_gpt":0.3847996771094771,"score_spread":0.19882730852666836,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6969203973","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0064674565,0.0007930304,0.011130713,0.002277322,0.00078608794,0.0012220064,0.9735339,0.001255687,0.0025337094],"genre_scores_gemma":[0.02696581,0.00017093497,0.022836944,0.0014045408,0.00019558282,0.009669447,0.9331444,0.0006866274,0.004925728],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9553669,0.023723954,0.0069172457,0.006593417,0.0058974633,0.0015008994],"domain_scores_gemma":[0.8817576,0.056889374,0.008915798,0.03728625,0.013452634,0.0016983042],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.043610252,0.0020721192,0.0023827178,0.0026477429,0.002326124,0.003991412,0.006999834,0.0049420614,0.034239177],"category_scores_gemma":[0.20341365,0.0011956681,0.0035928497,0.0054386435,0.0022124809,0.0020863446,0.004112612,0.0044912104,0.018373154],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0012346423,0.0002137913,0.0137880575,0.003588776,0.0015239968,0.00023972051,0.00036296653,0.0017731523,0.0006290175,0.005251104,0.955928,0.015466787],"study_design_scores_gemma":[0.007311166,0.00026829372,0.050474282,0.001506755,0.0017497935,0.00049610494,0.00040769734,0.003855665,0.0018398736,0.01622426,0.91551495,0.0003511197],"about_ca_topic_score_codex":0.019807935,"about_ca_topic_score_gemma":0.047917567,"teacher_disagreement_score":0.043610252,"about_ca_system_score_codex":0.0018407807,"about_ca_system_score_gemma":0.0064934813,"threshold_uncertainty_score":0.23063588},"labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W6977025636","doi":"10.6084/m9.figshare.7690283.v4","title":"No Sheriff in Town: Le rôle des dépôts institutionnels dans le Far West du libre accès","year":2019,"lang":"fr","type":"other","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Government (linguistics); Legislation; Context (archaeology); Field (mathematics)","score_opus":0.0322896965409984,"score_gpt":0.2329860105024237,"score_spread":0.20069631396142532,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6977025636","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.074269116,0.014551902,0.0020752442,0.21547677,0.0014578948,0.000031849602,0.0005926009,0.00022638738,0.6913183],"genre_scores_gemma":[0.48644483,0.0057104495,0.0013109106,0.0062383674,0.00034334228,0.000031399937,0.00018359361,0.00034632933,0.49939087],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9968849,0.0012894034,0.000082214014,0.0003045383,0.0007090644,0.0007298078],"domain_scores_gemma":[0.99317306,0.002374492,0.00047974655,0.00041647625,0.0013006402,0.0022555636],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.0036981285,0.00018691759,0.0003338738,0.0013694469,0.012283621,0.018921496,0.0010948475,0.0027608383,0.095899075],"category_scores_gemma":[0.011749661,0.00034467128,0.0001924505,0.003082675,0.0068077086,0.012593182,0.005525095,0.0024379778,0.0083942665],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002208359,0.00006703443,0.011051598,0.00027792444,0.000014571618,0.0015460869,0.069365904,0.00014672354,0.0005759739,0.402119,0.3443878,0.17022653],"study_design_scores_gemma":[0.000010879912,0.000021205251,0.007833526,0.00042262027,0.000011080447,0.0003125351,0.0670836,0.0000825605,0.00026292307,0.009078496,0.9148479,0.000032792574],"about_ca_topic_score_codex":0.1671894,"about_ca_topic_score_gemma":0.34852087,"teacher_disagreement_score":0.9989052,"about_ca_system_score_codex":0.007673198,"about_ca_system_score_gemma":0.015351713,"threshold_uncertainty_score":0.3324325},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W6977158599","doi":"10.6084/m9.figshare.26488030","title":"Wellcome Open Access Spend Report 2021/22","year":2024,"lang":"en","type":"dataset","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Government (linguistics); Block (permutation group theory); Quarter (Canadian coin); Block grant; Payment","score_opus":0.3564229446992313,"score_gpt":0.47365701501492674,"score_spread":0.11723407031569544,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6977158599","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00003312792,0.000038607584,0.000017916718,0.00007916906,0.000017216857,0.0000034410596,0.99880004,0.00010230075,0.0009082125],"genre_scores_gemma":[0.00035311916,0.0000944386,0.00014353974,0.00007994769,0.00001755213,0.00007030728,0.99716526,0.000171967,0.0019038055],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9979177,0.0002873949,0.0003477558,0.00044339985,0.00065160013,0.000352222],"domain_scores_gemma":[0.98973864,0.0030923875,0.0016988361,0.001380924,0.003168687,0.0009204913],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0016279714,0.0014399204,0.0012460788,0.007238816,0.0008591179,0.004723477,0.0018866872,0.0017304989,0.2025137],"category_scores_gemma":[0.021974185,0.00079376844,0.0011091607,0.017710673,0.00036235855,0.0026517957,0.002494342,0.0019908273,0.19131514],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024254385,0.0000032853459,0.00028211818,0.00033048764,0.000011028911,0.0000056425547,0.000010322741,0.00004118046,0.000019949952,0.00036876698,0.99799454,0.00090848916],"study_design_scores_gemma":[0.0001397148,0.000010423234,0.003543942,0.00040603938,0.000022962684,0.000024530545,0.00006437324,0.000087395776,0.000083436265,0.0007779475,0.9948166,0.0000225513],"about_ca_topic_score_codex":0.033945933,"about_ca_topic_score_gemma":0.050765302,"teacher_disagreement_score":0.9952765,"about_ca_system_score_codex":0.0019186895,"about_ca_system_score_gemma":0.0039423984,"threshold_uncertainty_score":0},"labels":[{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"opus","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W6983091719","doi":"","title":"Le scientifique en chef, les fonds de recherche du Québec et la science ouverte","year":2018,"lang":"fr","type":"other","venue":"Bibliothèque et Archives nationales du Québec (Québec government)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Context (archaeology); History of science; Sociology of scientific knowledge; Perspective (graphical)","score_opus":0.028950200151712912,"score_gpt":0.2850818505786706,"score_spread":0.2561316504269577,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6983091719","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0034076525,0.0057670544,0.0010557377,0.013923695,0.0027535297,0.0004106366,0.08592761,0.0015926377,0.8851614],"genre_scores_gemma":[0.008712949,0.0010471814,0.0006142607,0.00083990267,0.00014623436,0.00010650816,0.0070148995,0.0003366225,0.98118156],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99596715,0.00021772472,0.0001456711,0.0003008188,0.0021107488,0.0012578446],"domain_scores_gemma":[0.98624855,0.000928754,0.0003095767,0.0010454115,0.0088689085,0.0025988286],"candidate_categories":["metaresearch","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0030085696,0.0015166649,0.001474143,0.005588139,0.007367604,0.010189651,0.0031007535,0.004867064,0.3666949],"category_scores_gemma":[0.0119768195,0.0008772062,0.0012720035,0.006449303,0.0020514063,0.0020916515,0.0022428723,0.0042755543,0.085501045],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010150312,0.000061508654,0.0019560847,0.00019045873,0.000028211034,0.000069597205,0.00015026447,0.00022046662,0.00013283551,0.020551024,0.9421678,0.034370232],"study_design_scores_gemma":[0.000040117673,0.000013473538,0.007948059,0.00017680466,0.000013810849,0.000026778265,0.00013599047,0.00026262618,0.0002470547,0.0010497487,0.9900608,0.00002487556],"about_ca_topic_score_codex":0.97866666,"about_ca_topic_score_gemma":0.98487306,"teacher_disagreement_score":0.99699146,"about_ca_system_score_codex":0.072260104,"about_ca_system_score_gemma":0.15525703,"threshold_uncertainty_score":0.9033329},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W6995465946","doi":"","title":"Open Access in Canada","year":2017,"lang":"en","type":"other","venue":"CERN Document Server (European Organization for Nuclear Research)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Government (linguistics); Public access; Work (physics); Confidentiality","score_opus":0.06430243737529817,"score_gpt":0.3433033379972194,"score_spread":0.27900090062192123,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6995465946","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00055630424,0.00060371344,0.0009231636,0.0021624507,0.0008972201,0.00016322108,0.030928126,0.0042546904,0.95951104],"genre_scores_gemma":[0.0014803032,0.00030248237,0.0003892722,0.00023748746,0.00004994413,0.000029493887,0.007922926,0.0010306427,0.98855746],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9973073,0.00011250655,0.00008701753,0.00027329536,0.001448224,0.0007717297],"domain_scores_gemma":[0.9891945,0.00051129115,0.00018000597,0.0010012246,0.0067213657,0.002391653],"candidate_categories":["open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0017736829,0.0017710464,0.0015385989,0.005226746,0.007912792,0.013272158,0.0028757271,0.0029240253,0.7480083],"category_scores_gemma":[0.008070301,0.0008162576,0.0013463174,0.006737994,0.0026712664,0.0035951636,0.0041393563,0.002490701,0.64919686],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000031151423,0.000024105215,0.00017242138,0.00007636806,0.0000047602753,0.000050099636,0.00007226901,0.000078088684,0.00014047053,0.0068121315,0.9684739,0.024064306],"study_design_scores_gemma":[0.0000109702305,0.0000055065334,0.00068507507,0.00007824717,0.0000038942594,0.000022416116,0.000117922646,0.00011113987,0.00013146568,0.0013862681,0.99743134,0.000015601197],"about_ca_topic_score_codex":0.8680672,"about_ca_topic_score_gemma":0.8869406,"teacher_disagreement_score":0.99712425,"about_ca_system_score_codex":0.021701895,"about_ca_system_score_gemma":0.08838975,"threshold_uncertainty_score":0.35943556},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"medium"}],"label_agreement":"agree"},{"id":"W6996092637","doi":"","title":"Questionable Research Practices, the Replication Crisis and Open Science: A National Survey of Canadian Psychology Students","year":2021,"lang":"en","type":"other","venue":"OSF Preprints (OSF Preprints)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Replication (statistics); Psychological research; History of psychology; Survey research","score_opus":0.17497047455718232,"score_gpt":0.4896291490376248,"score_spread":0.3146586744804425,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W6996092637","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9804318,0.00073067093,0.00009276539,0.009716703,0.00003724153,0.00017075319,0.00163834,0.000012052978,0.007169722],"genre_scores_gemma":[0.99258643,0.0009283372,0.00021344221,0.002026996,0.000021273838,0.00009643753,0.0007196472,0.000021091097,0.003386259],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98528755,0.0019450961,0.0011027373,0.00089194597,0.0075363987,0.0032361925],"domain_scores_gemma":[0.8905705,0.016715188,0.027696991,0.0037502479,0.030871637,0.030395467],"candidate_categories":["metaresearch","open_science","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.0125629315,0.0002700867,0.00064163434,0.0058470713,0.008955551,0.0051343595,0.003230858,0.0025163312,0.006072299],"category_scores_gemma":[0.06108087,0.00080223114,0.00045732834,0.010559732,0.0043487055,0.0027306553,0.0034877488,0.0042273705,0.00067648536],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007852558,0.0003363703,0.9436795,0.00010379978,0.000033812074,0.00013033795,0.032300126,0.000058960417,0.00017760784,0.000909147,0.010822267,0.011369541],"study_design_scores_gemma":[0.000015129678,0.000042583608,0.9405856,0.0001567081,0.000015649744,0.000059609785,0.050364163,0.00015795963,0.000050688395,0.00022786245,0.0082762055,0.00004780622],"about_ca_topic_score_codex":0.9735613,"about_ca_topic_score_gemma":0.9839706,"teacher_disagreement_score":0.9974837,"about_ca_system_score_codex":0.03405728,"about_ca_system_score_gemma":0.13097872,"threshold_uncertainty_score":0.24710399},"labels":[{"model":"gemma","categories":["metaresearch","open_science","research_integrity"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["metaresearch","research_integrity","open_science"],"domain":"methods","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7005849595","doi":"","title":"Science Forum: Donated chemical probes for open science","year":2018,"lang":"en","type":"article","venue":"Oxford University Research Archive (ORA) (University of Oxford)","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"Novartis Pharma; Ministero dello Sviluppo Economico; Fundação de Amparo à Pesquisa do Estado de São Paulo; Deutsche Forschungsgemeinschaft; Ontario Ministry of Economic Development and Innovation; European Federation of Pharmaceutical Industries and Associations; Merck KGaA; Genome Canada; Bayer; Pfizer","keywords":"Pharmaceutical industry; Function (biology); Control (management); Chemical industry; Chemist; Drug industry","score_opus":0.022868398922985825,"score_gpt":0.30361365029782633,"score_spread":0.2807452513748405,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7005849595","genre_codex":"other","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004227886,0.01593901,0.08772573,0.044502974,0.06824054,0.001997093,0.041567527,0.050100166,0.685699],"genre_scores_gemma":[0.041990943,0.01327008,0.07457938,0.018543573,0.015853168,0.006676689,0.06218533,0.027554046,0.7393468],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98996943,0.0016964813,0.000366394,0.0011365066,0.0052710315,0.0015600885],"domain_scores_gemma":[0.9707101,0.004206081,0.0017785152,0.008172733,0.004444039,0.010688589],"candidate_categories":["open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.020839285,0.001554614,0.0015402348,0.005226032,0.0037969728,0.011184927,0.0040850597,0.009323815,0.49611723],"category_scores_gemma":[0.036890306,0.0015320714,0.0014196673,0.0055921176,0.0026237897,0.012412709,0.014174878,0.0073351115,0.2669263],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003850771,0.000113912516,0.00021083666,0.0005199119,0.000026370788,0.00019095631,0.00022911961,0.00014453701,0.007694843,0.040489987,0.84042156,0.10957303],"study_design_scores_gemma":[0.000038417285,0.00003298957,0.00007809111,0.000045895573,0.0000038157254,0.000059537444,0.000021613558,0.00007655528,0.0010930188,0.005176217,0.99336135,0.000012396646],"about_ca_topic_score_codex":0.00036868927,"about_ca_topic_score_gemma":0.00074996083,"teacher_disagreement_score":0.99591494,"about_ca_system_score_codex":0.0033239347,"about_ca_system_score_gemma":0.0056729172,"threshold_uncertainty_score":0.7187276},"labels":[{"model":"gemma","categories":["open_science"],"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":"commentary","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7005988486","doi":"","title":"The state and evolution of Gold Open Access: A country level analysis","year":2018,"lang":"en","type":"other","venue":"Leiden Repository (Leiden University)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Web of science; Quarter (Canadian coin); China; Bibliometrics; State (computer science); Filter (signal processing)","score_opus":0.02455380905811092,"score_gpt":0.26295163211002615,"score_spread":0.23839782305191523,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7005988486","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92976505,0.009166448,0.0011529952,0.0012423364,0.00006197729,0.000068160465,0.037559066,0.0001871113,0.020796914],"genre_scores_gemma":[0.98153394,0.002467958,0.0009448572,0.000117825395,0.000052985182,0.000045472716,0.013177955,0.000090691196,0.0015683084],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9949326,0.0008031991,0.0010953336,0.0007739292,0.0017395299,0.0006553816],"domain_scores_gemma":[0.9173651,0.024218198,0.035141718,0.004307795,0.016298713,0.0026686352],"candidate_categories":["metaresearch","bibliometrics","open_science"],"consensus_categories":[],"category_scores_codex":[0.0049743615,0.00023130655,0.0005957332,0.029042551,0.0005160863,0.0037908612,0.00064257707,0.0004093116,0.0031210214],"category_scores_gemma":[0.035053626,0.00023064479,0.0009494672,0.039400782,0.0010877941,0.0034418423,0.002378247,0.00065394456,0.001069295],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022394783,0.000039134688,0.95073426,0.0007905512,0.00053115096,0.00034764037,0.0014280813,0.00063116144,0.000937686,0.002935109,0.0048892787,0.03651201],"study_design_scores_gemma":[0.00000664104,0.000038731763,0.9847549,0.00022723057,0.00016290748,0.00027283406,0.0015049307,0.000553129,0.0007456682,0.00028430112,0.0114273215,0.000021468577],"about_ca_topic_score_codex":0.01279959,"about_ca_topic_score_gemma":0.012172819,"teacher_disagreement_score":0.9993574,"about_ca_system_score_codex":0.0011770306,"about_ca_system_score_gemma":0.0011755644,"threshold_uncertainty_score":0.026307285},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7017740980","doi":"","title":"Canada's new Open Access policy: Integrating libraries into open scholarship","year":2016,"lang":"en","type":"other","venue":"Brock University Digital Repository (Brock University)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":false,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Scholarship; Key (lock); Access to information; The Internet","score_opus":0.025666448063440678,"score_gpt":0.2557874426181428,"score_spread":0.2301209945547021,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7017740980","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00435861,0.0026818223,0.0017507927,0.053827595,0.0015374157,0.0003684997,0.007998582,0.0014307539,0.9260458],"genre_scores_gemma":[0.02022434,0.0010871472,0.0025617871,0.012752485,0.00024351032,0.00014313377,0.0012726958,0.00057503197,0.96114004],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9816095,0.00076723343,0.0005055295,0.0006949294,0.012066583,0.00435631],"domain_scores_gemma":[0.9721362,0.0028320088,0.00044103627,0.0021064659,0.013610266,0.0088739535],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.0066543217,0.00077082444,0.0010368205,0.0051404987,0.018506221,0.025792137,0.0046366537,0.018929098,0.09598307],"category_scores_gemma":[0.027770393,0.0015686211,0.0013345471,0.0069544143,0.006323056,0.006502566,0.008252766,0.010029234,0.020183353],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00003380526,0.00007793884,0.0007097038,0.000105954365,0.000008915111,0.000119261036,0.000829035,0.00020081444,0.00017227803,0.1399997,0.8373908,0.020351838],"study_design_scores_gemma":[0.000021605902,0.000008135523,0.0032326994,0.00015662596,0.000013236728,0.000035858142,0.0007551625,0.00021848237,0.00028123957,0.00659397,0.988627,0.00005588491],"about_ca_topic_score_codex":0.9818032,"about_ca_topic_score_gemma":0.9941269,"teacher_disagreement_score":0.99536335,"about_ca_system_score_codex":0.11240882,"about_ca_system_score_gemma":0.38889736,"threshold_uncertainty_score":0.81558686},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7028590992","doi":"","title":"FDRI digital workshop: enabling reproducibility in hydrological research","year":2024,"lang":"en","type":"other","venue":"NERC Open Research Archive (Natural Environment Research Council)","topic":"European history and politics","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"","funders":"British Geological Survey; Cardiff University; University of Bristol; Cranfield University; University of Leeds; University of Oxford; University of Southampton; Trent University; University of Aberdeen; Newcastle University; Consortium of Universities for the Advancement of Hydrologic Science; Natural Resources Wales; Scottish Environment Protection Agency; James Hutton Institute; Nottingham Trent University; King's College London; Natural Environment Research Council; UK Research and Innovation; Queen Mary University of London","keywords":"Key (lock); Reproducibility; Measure (data warehouse); Reliability (semiconductor); Digital data","score_opus":0.3265586735517418,"score_gpt":0.439211242601276,"score_spread":0.11265256904953419,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7028590992","genre_codex":"commentary","genre_gemma":"other","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.009185844,0.02083519,0.1881922,0.34816384,0.051650066,0.0041521387,0.03497247,0.012838027,0.3300102],"genre_scores_gemma":[0.17037126,0.021239433,0.33577585,0.08051661,0.025743892,0.012869195,0.07771597,0.017357409,0.25841045],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.8483497,0.07681568,0.009066113,0.008538471,0.047877435,0.00935264],"domain_scores_gemma":[0.63299215,0.16325574,0.01165202,0.10154721,0.055659782,0.034893125],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.2561181,0.0016329101,0.0016622727,0.006726821,0.005829232,0.02133751,0.011277531,0.008216393,0.07507955],"category_scores_gemma":[0.2716247,0.0014068244,0.002521068,0.007605848,0.007579041,0.02487688,0.04799857,0.013111409,0.02876633],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000281694,0.00015214551,0.0010446915,0.0023611651,0.000113882415,0.0003465181,0.0053474437,0.0011730435,0.001790422,0.114772685,0.6977885,0.17482787],"study_design_scores_gemma":[0.00006311657,0.000043662523,0.0013391596,0.00053946726,0.000020245732,0.00008569432,0.0011082728,0.0002623214,0.0012510853,0.024960902,0.9702517,0.00007443391],"about_ca_topic_score_codex":0.008789463,"about_ca_topic_score_gemma":0.008811863,"teacher_disagreement_score":0.74388194,"about_ca_system_score_codex":0.008009023,"about_ca_system_score_gemma":0.03505109,"threshold_uncertainty_score":0.91733885},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"reproducibility","study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"W7036827764","doi":"","title":"Connecting Readers with Open Access Resources: The CUFTS Free! Open Access Collections Group","year":2009,"lang":"en","type":"other","venue":"Elpub digital library","topic":"Computational Physics and Python Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Metadata; Directory; Suite; Dissemination; Free access; Download; Open access publishing; Digital library","score_opus":0.04666607055086626,"score_gpt":0.3140999652273604,"score_spread":0.26743389467649414,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7036827764","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0074363826,0.007157466,0.032956425,0.35465017,0.037224844,0.0034944366,0.0032568441,0.03308336,0.52074015],"genre_scores_gemma":[0.011475075,0.0054323543,0.04127282,0.040808655,0.017248645,0.002078267,0.010835871,0.00906894,0.8617794],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.97018415,0.003632622,0.0016056251,0.0043156887,0.015977899,0.0042840014],"domain_scores_gemma":[0.84230846,0.012626664,0.0039767288,0.02499065,0.040688027,0.07540939],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.04002837,0.0021291827,0.0016490717,0.011689748,0.019121746,0.034865033,0.010444628,0.016749987,0.17738482],"category_scores_gemma":[0.052147787,0.0018894586,0.00173806,0.009895629,0.007194152,0.031558495,0.037774723,0.009933853,0.12957317],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000026941074,0.00008546453,0.00036269415,0.0000765547,0.0000053205617,0.000104664716,0.0006459208,0.000032376647,0.00040383197,0.005373605,0.9279797,0.06490283],"study_design_scores_gemma":[0.000008678445,0.000018373496,0.00032226768,0.00006278766,0.0000031765132,0.000041420004,0.00030840395,0.00008225097,0.00015993378,0.0007301574,0.9982418,0.000020688849],"about_ca_topic_score_codex":0.05603937,"about_ca_topic_score_gemma":0.06844536,"teacher_disagreement_score":0.98955536,"about_ca_system_score_codex":0.009695155,"about_ca_system_score_gemma":0.0731066,"threshold_uncertainty_score":0.5934113},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"software","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7042331891","doi":"","title":"Openness and data sharing at The Neuro (Montreal Neurological Institute-Hospital)","year":2021,"lang":"en","type":"other","venue":"OSF Preprints (OSF Preprints)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Audit; Openness to experience; Data sharing; Protocol (science); Data collection","score_opus":0.03877506496761851,"score_gpt":0.2866840090623571,"score_spread":0.24790894409473857,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7042331891","genre_codex":"protocol","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01694363,0.0005362303,0.02782421,0.0050207432,0.0009641041,0.5684308,0.32753992,0.0028596667,0.049880803],"genre_scores_gemma":[0.02400949,0.00036592482,0.021173036,0.00085306546,0.00015168673,0.90051943,0.035561133,0.0005353491,0.016830875],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.94986314,0.030015295,0.010331858,0.0023107098,0.0043761027,0.0031029051],"domain_scores_gemma":[0.8299153,0.0495965,0.015162046,0.036788035,0.056082368,0.01245568],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.09009823,0.0011874806,0.0016116803,0.006908711,0.005566532,0.0040171845,0.003060318,0.0015897184,0.17727871],"category_scores_gemma":[0.1483465,0.001798143,0.0014928387,0.0072998675,0.002226801,0.004338475,0.0050279223,0.0031772146,0.027423022],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.009883909,0.0013408557,0.011574502,0.0062224865,0.00014446289,0.00034238416,0.011239971,0.0006741031,0.0008882158,0.014347379,0.823446,0.11989576],"study_design_scores_gemma":[0.009611788,0.0012960696,0.08743652,0.009946588,0.0002061942,0.00022359152,0.008467409,0.0015201033,0.0030631323,0.012885585,0.8648153,0.00052780035],"about_ca_topic_score_codex":0.06402737,"about_ca_topic_score_gemma":0.06774363,"teacher_disagreement_score":0.99693966,"about_ca_system_score_codex":0.012122573,"about_ca_system_score_gemma":0.074046925,"threshold_uncertainty_score":0.5930564},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"not_applicable","genre":"commentary","about_ca_system":true,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7043527902","doi":"","title":"SSEA Preregistration Workshop","year":2019,"lang":"en","type":"other","venue":"OSF Preprints (OSF Preprints)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Information system; Work (physics); The Internet; Agency (philosophy); Attendance","score_opus":0.020676219090083322,"score_gpt":0.27638271318228175,"score_spread":0.2557064940921984,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7043527902","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0071438826,0.0024487176,0.0044986866,0.05500364,0.08912137,0.0025846784,0.018202553,0.0034888897,0.81750757],"genre_scores_gemma":[0.011360185,0.0008076335,0.0014603197,0.0047496944,0.0041684774,0.0010157869,0.006676519,0.0009341884,0.96882725],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99477744,0.0015974769,0.0002750558,0.00042724167,0.001629867,0.0012929535],"domain_scores_gemma":[0.9832108,0.0012593896,0.00031426753,0.0011050492,0.007218113,0.0068924143],"candidate_categories":["metaresearch","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.011381979,0.0008381631,0.00071358046,0.0021022481,0.0054182615,0.0069253207,0.0019286962,0.0028406125,0.5211182],"category_scores_gemma":[0.014505583,0.0004700988,0.00097254023,0.001443639,0.0007344948,0.0026563357,0.0065769893,0.0041255904,0.335688],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000101597805,0.000042733605,0.000109430155,0.000058163663,0.0000015760045,0.000045908197,0.00019309612,0.000011521859,0.00013422569,0.0012068739,0.98289883,0.015196004],"study_design_scores_gemma":[0.000015461295,0.00003798659,0.00078245054,0.00005759039,0.0000013631753,0.00002113462,0.00048169694,0.000015891537,0.00013092138,0.00045710438,0.9979905,0.000007824964],"about_ca_topic_score_codex":0.010923899,"about_ca_topic_score_gemma":0.031689152,"teacher_disagreement_score":0.9980713,"about_ca_system_score_codex":0.0041476497,"about_ca_system_score_gemma":0.015643032,"threshold_uncertainty_score":0.6830667},"labels":[{"model":"gemma","categories":["open_science","metaresearch"],"domain":"reproducibility","study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7066102103","doi":"","title":"Geospatial open data: reshaping citizens and governments, roles and interactions","year":2017,"lang":"en","type":"dissertation","venue":"eScholarship@McGill (McGill)","topic":"Astrophysical Phenomena and Observations","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Open government; Open data; Framing (construction); Crowdsourcing; Geospatial analysis; Government (linguistics); Civic engagement; Outsourcing","score_opus":0.03701863691737942,"score_gpt":0.2923951837310236,"score_spread":0.25537654681364413,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7066102103","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.39818215,0.0117182825,0.048160788,0.1600841,0.0011361481,0.00031744453,0.0011846615,0.0005067964,0.37870958],"genre_scores_gemma":[0.93517053,0.00813322,0.016607312,0.005547186,0.00035158716,0.00022464892,0.0005450071,0.00025082822,0.033169653],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.99369353,0.0039259475,0.00014687517,0.0005540999,0.0010542618,0.0006252361],"domain_scores_gemma":[0.99016154,0.0055118455,0.00056994794,0.0018380906,0.0008252608,0.001093246],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0077589094,0.0003345834,0.00032374437,0.0026366075,0.0105411345,0.018004112,0.0013526736,0.0017037311,0.004693781],"category_scores_gemma":[0.009748976,0.00040128647,0.00040487322,0.007263242,0.025642952,0.017185213,0.014531271,0.002980106,0.0009158838],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000034572822,0.000050256986,0.010971585,0.00025868847,0.000018408196,0.00030527424,0.43470624,0.00044748263,0.0007908762,0.35448977,0.031066716,0.1668602],"study_design_scores_gemma":[0.0000080511245,0.000019394765,0.006508504,0.00060936855,0.000013408722,0.00015224976,0.2539897,0.00050459447,0.0005831201,0.08572939,0.6518537,0.000028483346],"about_ca_topic_score_codex":0.050279025,"about_ca_topic_score_gemma":0.07501228,"teacher_disagreement_score":0.99864733,"about_ca_system_score_codex":0.0068949154,"about_ca_system_score_gemma":0.015243632,"threshold_uncertainty_score":0.099972725},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7092178964","doi":"10.5281/zenodo.17362084","title":"Open science in Canada: Here's the data","year":2025,"lang":"","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Concordia University","funders":"Fonds de recherche du Québec","keywords":"Open science; Presentation (obstetrics); Open source; Citizen science; Open data","score_opus":0.15095694122841136,"score_gpt":0.3819073973124258,"score_spread":0.23095045608401443,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7092178964","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00035159336,0.0024990726,0.00056728197,0.04148347,0.018325904,0.000456617,0.7552696,0.0041098706,0.17693655],"genre_scores_gemma":[0.0135010565,0.01108755,0.0036830807,0.038924046,0.002831541,0.0016796953,0.5755073,0.0064276718,0.3463581],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9718388,0.00082809344,0.0014863679,0.0014236991,0.019232392,0.005190633],"domain_scores_gemma":[0.87195116,0.005207174,0.0026951518,0.007928135,0.09207751,0.020140845],"candidate_categories":["metaresearch","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.009135736,0.0015003284,0.00395661,0.011247387,0.009853736,0.027367577,0.005796923,0.006114259,0.49396735],"category_scores_gemma":[0.059709746,0.0016284753,0.0018887124,0.03169242,0.004861882,0.008699656,0.009243505,0.009841503,0.31373185],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000024338355,0.0000058080486,0.000086960405,0.00016150919,0.000004168367,0.000007915166,0.000019506573,0.000015269647,0.000021629256,0.0007373237,0.99480176,0.004113862],"study_design_scores_gemma":[0.000025487494,0.0000039406837,0.0010627247,0.0003559193,0.000007158053,0.0000068072873,0.00015004797,0.000020460158,0.000056685567,0.00047712075,0.99779797,0.00003576928],"about_ca_topic_score_codex":0.9259138,"about_ca_topic_score_gemma":0.9265727,"teacher_disagreement_score":0.9942031,"about_ca_system_score_codex":0.059206277,"about_ca_system_score_gemma":0.2885547,"threshold_uncertainty_score":0.7217941},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"agree"},{"id":"W7092368277","doi":"10.6084/m9.figshare.30391284","title":"Comparing companion open access journals to their traditional journal counterparts","year":2025,"lang":"","type":"article","venue":"Figshare","topic":"Historical Economic and Legal Thought","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Publishing; Citation; Publication; Demographics; Bibliometrics; Scope (computer science); Quarter (Canadian coin)","score_opus":0.4060789229441768,"score_gpt":0.439557671757479,"score_spread":0.033478748813302206,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7092368277","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92647195,0.028895264,0.0017181716,0.0039204936,0.001296709,0.00025580337,0.009082486,0.000165936,0.028193146],"genre_scores_gemma":[0.98350877,0.0069532576,0.0015868098,0.00067290076,0.0011549027,0.00017639031,0.0032075443,0.00009952005,0.0026399936],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9844059,0.0045154607,0.0031275877,0.0016395063,0.005390231,0.0009212843],"domain_scores_gemma":[0.73877716,0.13655199,0.07061891,0.0099143665,0.031001596,0.01313603],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.013137972,0.00029396164,0.00079135544,0.011242228,0.0007352816,0.0044993237,0.0009039574,0.00066189736,0.016711073],"category_scores_gemma":[0.16496857,0.0002957447,0.0009302877,0.020315494,0.0012651321,0.006089496,0.0019151268,0.0009825049,0.0026839771],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029851485,0.0004533214,0.8437753,0.0037035542,0.0022827457,0.00062438525,0.0021739095,0.00041475368,0.0005334654,0.0082324715,0.017861119,0.116959795],"study_design_scores_gemma":[0.0003333552,0.0015591631,0.93906724,0.001126592,0.0008295152,0.0021969583,0.004595726,0.0011622357,0.00074112846,0.006177069,0.042156376,0.000054738797],"about_ca_topic_score_codex":0.00094943185,"about_ca_topic_score_gemma":0.0015158158,"teacher_disagreement_score":0.9955007,"about_ca_system_score_codex":0.0015970704,"about_ca_system_score_gemma":0.0014758046,"threshold_uncertainty_score":0.069481015},"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":["scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7094999132","doi":"10.5281/zenodo.17444614","title":"What Does Reproducibility Look Like for DH Projects?","year":2025,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Race, Genetics, and Society","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Presentation (obstetrics); Key (lock); Process (computing); Function (biology); Salient; Variety (cybernetics); Component (thermodynamics)","score_opus":0.02170035974714531,"score_gpt":0.2674268714701937,"score_spread":0.2457265117230484,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7094999132","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013955478,0.015064253,0.1953557,0.6530917,0.016879587,0.0006559723,0.0011462645,0.004616156,0.099234894],"genre_scores_gemma":[0.49151045,0.022026267,0.2842438,0.10994956,0.021870786,0.0050125835,0.0032180806,0.0120327175,0.05013573],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.61108816,0.26913637,0.02659332,0.02501407,0.059056,0.009112126],"domain_scores_gemma":[0.32896563,0.30664316,0.0246008,0.24216656,0.0802094,0.017414415],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.3494841,0.0013239445,0.0022334252,0.008760547,0.019005835,0.055020966,0.007878752,0.013176581,0.012951566],"category_scores_gemma":[0.59965926,0.0015538787,0.0034970087,0.014253144,0.04860636,0.07977573,0.028227389,0.012038672,0.007158846],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012992878,0.00009593789,0.0101094805,0.0015302851,0.00022227572,0.0004263111,0.02400761,0.0010953255,0.0006381891,0.7103154,0.08247955,0.16894971],"study_design_scores_gemma":[0.000066633555,0.00011758689,0.0031488268,0.002281961,0.000090757756,0.0005456955,0.008952389,0.00077775685,0.001168952,0.5737816,0.40888056,0.00018728254],"about_ca_topic_score_codex":0.005416959,"about_ca_topic_score_gemma":0.0037260943,"teacher_disagreement_score":0.6505159,"about_ca_system_score_codex":0.013102034,"about_ca_system_score_gemma":0.03608955,"threshold_uncertainty_score":0.8022019},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7095356221","doi":"","title":"STUDY PROTOCOL Open Access The Alberta moving beyon","year":2016,"lang":"en","type":"article","venue":"","topic":"Prenatal Screening and Diagnostics","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Protocol (science); Psychosocial; Identification (biology); Data collection","score_opus":0.08638478296553129,"score_gpt":0.43157603665512223,"score_spread":0.3451912536895909,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7095356221","genre_codex":"protocol","genre_gemma":"protocol","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"protocol","genre_consensus":"protocol","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0079243425,0.0013026887,0.006554661,0.0019437514,0.0024382544,0.8050175,0.1334481,0.00068951905,0.0406812],"genre_scores_gemma":[0.010171692,0.00074303435,0.007772948,0.0013404449,0.00032341835,0.9375497,0.012699123,0.00009978088,0.02929992],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975465,0.000876435,0.00031093438,0.00044736522,0.00044417873,0.00037459767],"domain_scores_gemma":[0.99051136,0.0015175372,0.00072436006,0.0016030619,0.004006088,0.0016375941],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.012527064,0.0019020449,0.0025469153,0.0028112463,0.0030240056,0.0025143763,0.0029957998,0.0028057778,0.31772313],"category_scores_gemma":[0.017993433,0.0015876384,0.0012353096,0.0022908398,0.0010933263,0.001101395,0.001859093,0.0021934123,0.04592569],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.04866013,0.0036023636,0.007508736,0.012676825,0.00027984567,0.00070385676,0.0014732318,0.0010590712,0.0008344663,0.014264925,0.74069965,0.16823693],"study_design_scores_gemma":[0.06586684,0.004105408,0.04172054,0.021225022,0.00040321518,0.0002364097,0.0018715597,0.0017008777,0.00080335187,0.007871071,0.8539214,0.00027429813],"about_ca_topic_score_codex":0.11211234,"about_ca_topic_score_gemma":0.15177432,"teacher_disagreement_score":0.88788766,"about_ca_system_score_codex":0.0062344484,"about_ca_system_score_gemma":0.041654427,"threshold_uncertainty_score":0.9731852},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"protocol","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"design_other","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7098330228","doi":"","title":"STUDY PROTOCOL Open Access Study protocol","year":2013,"lang":"en","type":"article","venue":"","topic":"Marine Invertebrate Physiology and Ecology","field":"Earth and Planetary Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Brainstorming; Protocol (science); Knowledge translation; Best practice; Collaborative network; Social network analysis","score_opus":0.08891238628795363,"score_gpt":0.3961833607384418,"score_spread":0.3072709744504882,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7098330228","genre_codex":"protocol","genre_gemma":"protocol","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"protocol","genre_consensus":"protocol","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0006918867,0.00020165039,0.003193955,0.00037563953,0.0009260613,0.9838657,0.007802756,0.00014277447,0.0027996385],"genre_scores_gemma":[0.0010197862,0.00009974444,0.0027946222,0.00021442617,0.000049623213,0.9945798,0.0004748056,0.000022557415,0.00074462726],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.95594317,0.023368241,0.009696218,0.004389881,0.0042631086,0.0023394085],"domain_scores_gemma":[0.88810545,0.036161296,0.011763745,0.022734275,0.035257056,0.005978092],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.08715929,0.004740147,0.009871374,0.009403959,0.0052297194,0.008857192,0.0048188246,0.0072725653,0.27345723],"category_scores_gemma":[0.13721216,0.004869945,0.0039125355,0.009300096,0.0053715794,0.0055992072,0.0036442617,0.008292045,0.050682716],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.1252003,0.009620921,0.004883167,0.16932796,0.0019516645,0.003364915,0.0085812695,0.0044577946,0.002566091,0.074848935,0.39906272,0.19613431],"study_design_scores_gemma":[0.13162202,0.008533892,0.008706333,0.11524999,0.0009692237,0.00083653204,0.0074716276,0.006200863,0.003496687,0.03663454,0.6795623,0.0007160155],"about_ca_topic_score_codex":0.004109453,"about_ca_topic_score_gemma":0.004771695,"teacher_disagreement_score":0.7265428,"about_ca_system_score_codex":0.0081166895,"about_ca_system_score_gemma":0.037616886,"threshold_uncertainty_score":0.9148056},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["insufficient_payload"],"domain":null,"study_design":"not_applicable","genre":"protocol","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7102415238","doi":"10.6084/m9.figshare.30475016.v7","title":"Additional file of Improving accessibility to radiotherapy services in Cali, Colombia: cross-sectional equity analyses using open data and big data travel times from 2020","year":2025,"lang":"","type":"other","venue":"Figshare","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Ottawa; Queen's University","funders":"","keywords":"Big data; Open data; Equity (law); Open source; Real world data; Data collection","score_opus":0.334455149662149,"score_gpt":0.46266606086333645,"score_spread":0.12821091120118744,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7102415238","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000121177945,0.00000573739,0.000015072708,0.00004475241,0.000010599949,0.000015956752,0.99913836,0.000032395903,0.00061586854],"genre_scores_gemma":[0.005539611,0.00007806327,0.00049690646,0.00014105969,0.000049850172,0.0006780454,0.98547137,0.00014042735,0.0074046897],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9990644,0.0000881476,0.00016349118,0.00015891356,0.0003240276,0.0002009379],"domain_scores_gemma":[0.9904908,0.0032215212,0.0014629444,0.0006348835,0.003479729,0.0007102046],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00092962274,0.0008658684,0.0008419863,0.003592353,0.0007764535,0.0015902285,0.0014960114,0.0008738617,0.4853302],"category_scores_gemma":[0.016211197,0.0005361872,0.00083664857,0.009390603,0.0001709717,0.0017253944,0.0012007297,0.0008595039,0.07597666],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000088523426,0.000036222486,0.00637979,0.000533832,0.000026283124,0.000024757484,0.000039369865,0.0002197358,0.00002405443,0.00026124276,0.9891075,0.0032586518],"study_design_scores_gemma":[0.0018952785,0.00011875793,0.2804157,0.0018827177,0.00018304586,0.00015750644,0.0017562251,0.0014555961,0.0004822051,0.0017016837,0.7097884,0.00016280936],"about_ca_topic_score_codex":0.16431755,"about_ca_topic_score_gemma":0.21488746,"teacher_disagreement_score":0.4853302,"about_ca_system_score_codex":0.002259041,"about_ca_system_score_gemma":0.004244722,"threshold_uncertainty_score":0.734114},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W7106314809","doi":"10.5281/zenodo.17665019","title":"THE GLOBAL SANCTUARY EASEMENT PROTOCOL (GSEP) A Universal Covenant for Land Stewardship, Humanitarian Infrastructure, and Open Science","year":2025,"lang":"","type":"dataset","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Easement; Human rights; Land registration; Stewardship (theology); Covenant; Custodians; International law; International community; Asset (computer security)","score_opus":0.02695370300174516,"score_gpt":0.30821549853847413,"score_spread":0.281261795536729,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7106314809","genre_codex":"other","genre_gemma":"protocol","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"protocol","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013139961,0.0012709787,0.34548113,0.027135115,0.0056950347,0.0059742304,0.0069439956,0.013463039,0.5808966],"genre_scores_gemma":[0.24181212,0.0042952406,0.24469769,0.042470723,0.004904867,0.023027983,0.02333977,0.007591086,0.40786055],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9535873,0.019680299,0.0072359215,0.0034008138,0.012779544,0.0033161133],"domain_scores_gemma":[0.92585313,0.03277379,0.0035316288,0.022268096,0.012554597,0.0030187059],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.045944594,0.0011333374,0.0011894335,0.0035265104,0.0039353194,0.016915048,0.0032689776,0.0123712355,0.028916068],"category_scores_gemma":[0.10647628,0.001735568,0.0011936874,0.003058494,0.010100909,0.014181056,0.014351774,0.013898497,0.025867403],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00010601473,0.0001133915,0.001018748,0.00027413323,0.000016603006,0.00046084094,0.0014070378,0.00227502,0.0018284561,0.7823228,0.15920182,0.05097514],"study_design_scores_gemma":[0.000103839586,0.0001571978,0.001406215,0.0011220394,0.000024904155,0.00044894285,0.0005880277,0.003896295,0.0022662014,0.13467205,0.85505295,0.00026136756],"about_ca_topic_score_codex":0.0068162386,"about_ca_topic_score_gemma":0.0035977662,"teacher_disagreement_score":0.99673104,"about_ca_system_score_codex":0.003557367,"about_ca_system_score_gemma":0.016900748,"threshold_uncertainty_score":0.24298108},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7108215335","doi":"10.25504/fairsharing.e1eb50","title":"FAIRsharing record for: Burnaby's Open Data Portal","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Open data; Check-in; Open government; Range (aeronautics); Open source; Open standard","score_opus":0.16708865644286439,"score_gpt":0.40264336739972045,"score_spread":0.23555471095685607,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108215335","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00013337054,0.00003144324,0.00005475113,0.00027771926,0.000072728035,0.0000195356,0.9945193,0.0006228298,0.0042682705],"genre_scores_gemma":[0.0008150671,0.00006254994,0.00029773684,0.00015405146,0.000029475746,0.00013914019,0.9884979,0.00051740947,0.009486547],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981797,0.00019286636,0.0001409811,0.00030864455,0.0007309718,0.00044690142],"domain_scores_gemma":[0.9912349,0.0013361546,0.00057333393,0.002251027,0.0029755172,0.0016290344],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021677064,0.0012258388,0.0013159792,0.005834205,0.002937702,0.006678076,0.003291191,0.0018565401,0.26614237],"category_scores_gemma":[0.015261043,0.00088241923,0.00076567347,0.011673349,0.00086369796,0.0047597853,0.005227763,0.0027318604,0.21365768],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000011590251,0.0000043744967,0.00014456533,0.000039594543,0.0000015383308,0.0000042303673,0.00001463038,0.000019666273,0.000007944415,0.00029586485,0.99860364,0.0008524252],"study_design_scores_gemma":[0.000043738208,0.000003848942,0.001940428,0.00012429508,0.000002506681,0.000009220214,0.00014281903,0.00010426497,0.00006451623,0.00080134283,0.99674207,0.000020896909],"about_ca_topic_score_codex":0.20312434,"about_ca_topic_score_gemma":0.24446446,"teacher_disagreement_score":0.79687566,"about_ca_system_score_codex":0.0038377538,"about_ca_system_score_gemma":0.0066192797,"threshold_uncertainty_score":0.8903349},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W7108238944","doi":"10.25504/fairsharing.6bacac","title":"FAIRsharing record for: Central Lake Ontario Conservation Open Data Portal","year":2025,"lang":"","type":"dataset","venue":"FAIRsharing.org","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Download; Natural (archaeology); Flood myth; Open data; Public access; Access to information; Natural resource","score_opus":0.13926967350016684,"score_gpt":0.35263213643828717,"score_spread":0.21336246293812033,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108238944","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00010835953,0.00002556974,0.000061797946,0.00034162652,0.000067551926,0.000030503465,0.9896046,0.0005271498,0.009232832],"genre_scores_gemma":[0.0010961657,0.00008496466,0.00042519826,0.00019909872,0.00003399204,0.00022433742,0.97361034,0.00063003984,0.023695828],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983115,0.00012504112,0.000107432155,0.0002515841,0.0008155338,0.0003888029],"domain_scores_gemma":[0.9888815,0.0013394679,0.0005451004,0.002548365,0.0045178775,0.0021675678],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.002020723,0.001074515,0.0011044844,0.0049156416,0.00328886,0.0061315503,0.0029447973,0.001625678,0.40926144],"category_scores_gemma":[0.014521848,0.0008766961,0.000679689,0.011306252,0.0009770697,0.0042892937,0.0045122807,0.001736621,0.22977774],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000009250322,0.000002579796,0.00015092544,0.000033699824,0.0000014586236,0.000003532026,0.000017003322,0.000013718629,0.000007813219,0.00027667408,0.99861586,0.0008673298],"study_design_scores_gemma":[0.000038436676,0.0000020267523,0.0020896255,0.000086762986,0.0000024941799,0.00000707788,0.00010861282,0.00006566269,0.00004354773,0.0005793501,0.9969584,0.000017989532],"about_ca_topic_score_codex":0.600554,"about_ca_topic_score_gemma":0.74281555,"teacher_disagreement_score":0.600554,"about_ca_system_score_codex":0.010971275,"about_ca_system_score_gemma":0.017387228,"threshold_uncertainty_score":0.84261686},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":[],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7108631754","doi":"10.31274/jlsc.20194","title":"Publications Produced and Services Offered by Library Publishing Programs in the United States and Canada: A Data-Driven Analysis","year":2025,"lang":"","type":"article","venue":"Journal of Librarianship and Scholarly Communication","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Staffing; Publishing; Service (business); Table (database); Order (exchange); Focus (optics)","score_opus":0.11000385768366996,"score_gpt":0.33369865505929525,"score_spread":0.22369479737562528,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7108631754","genre_codex":"dataset","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.24479513,0.0008047307,0.0007604113,0.0014201218,0.000038801114,0.0002887016,0.74523413,0.0002451768,0.006412851],"genre_scores_gemma":[0.46508548,0.0012771316,0.0048482404,0.0006559271,0.000058026606,0.0009086255,0.51961875,0.00015801903,0.0073897303],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99566984,0.00023007698,0.00030752606,0.0005523114,0.0021757216,0.0010644974],"domain_scores_gemma":[0.977736,0.002324114,0.0025378973,0.0006834245,0.01454748,0.002171105],"candidate_categories":["bibliometrics","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0026736697,0.00065507327,0.00088951754,0.01421226,0.00260017,0.0043924134,0.002247427,0.0006078071,0.005158862],"category_scores_gemma":[0.014325654,0.0005923454,0.0016919329,0.033607725,0.0011251738,0.001127812,0.0024798387,0.0012997957,0.0016344844],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017455356,0.000105267405,0.8977489,0.0005344689,0.0003001765,0.00018404979,0.0014689348,0.0030842505,0.00017362316,0.002199411,0.07996936,0.014056941],"study_design_scores_gemma":[0.00003495383,0.000038436385,0.92801607,0.0003766883,0.00012240924,0.000101149686,0.0051054973,0.005417544,0.0005488024,0.0004948344,0.059634987,0.00010872744],"about_ca_topic_score_codex":0.99092776,"about_ca_topic_score_gemma":0.9891186,"teacher_disagreement_score":0.9956076,"about_ca_system_score_codex":0.03718078,"about_ca_system_score_gemma":0.07769845,"threshold_uncertainty_score":0.2697667},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W7110525921","doi":"","title":"Surveying the landscape of CIHR-funded research data sharing practices: An analysis of the published literature","year":2024,"lang":"","type":"other","venue":"OSF Preprints (OSF Preprints)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Data sharing; Data management; Key (lock); Health data; Information sharing; Research data; Data access","score_opus":0.13742648934911544,"score_gpt":0.40253804276078853,"score_spread":0.2651115534116731,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7110525921","genre_codex":"review","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15438601,0.5720396,0.0130146425,0.14589833,0.0014708334,0.0010723496,0.019176802,0.00028772934,0.0926537],"genre_scores_gemma":[0.6354454,0.31424692,0.028121999,0.009103367,0.00064407196,0.0006367705,0.008817356,0.0003573819,0.00262677],"study_design_codex":"design_other","study_design_gemma":"systematic_review","domain_scores_codex":[0.81901044,0.050849862,0.024879254,0.01103315,0.08671164,0.007515568],"domain_scores_gemma":[0.3244198,0.31825712,0.05193049,0.021595376,0.26971057,0.014086689],"candidate_categories":["metaresearch","bibliometrics"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.20017853,0.00078060396,0.0017590155,0.07464837,0.01153962,0.023627305,0.0067861225,0.0021146939,0.0033592915],"category_scores_gemma":[0.39124882,0.0013430377,0.0017273888,0.17282978,0.012147839,0.008150766,0.0080043,0.0027247793,0.0004520215],"study_design_candidate":"systematic_review","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026284182,0.00007978059,0.14970765,0.06330451,0.001879754,0.0015121507,0.17337434,0.002460627,0.002143689,0.089397654,0.056092765,0.45978418],"study_design_scores_gemma":[0.000048162216,0.00011931772,0.19227526,0.099631205,0.0020885828,0.0010573873,0.15927504,0.0013918567,0.0019218909,0.009839576,0.5318377,0.00051404245],"about_ca_topic_score_codex":0.8911883,"about_ca_topic_score_gemma":0.93495697,"teacher_disagreement_score":0.9253516,"about_ca_system_score_codex":0.12537746,"about_ca_system_score_gemma":0.3711341,"threshold_uncertainty_score":0.9863223},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"systematic_review","genre":"review","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["metaresearch","open_science","scholarly_communication"],"domain":"reproducibility","study_design":"design_other","genre":"review","about_ca_system":true,"about_ca_topic":true,"confidence":"medium"}],"label_agreement":"split"},{"id":"W7117298754","doi":"10.5281/zenodo.18056545","title":"Barriers and pathways for advancing open science and open scholarship in academic institutions: a Canadian perspective","year":2025,"lang":"en","type":"preprint","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Concordia University","funders":"Fonds de recherche du Québec; Social Sciences and Humanities Research Council; Canada Research Chairs","keywords":"Scholarship; Corporate governance; Equity (law); Open science; Indigenous; Engaged scholarship; Work (physics)","score_opus":0.18662070235144906,"score_gpt":0.3937475157614601,"score_spread":0.20712681341001105,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7117298754","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10243761,0.016246442,0.011218743,0.66900927,0.000943964,0.0003860483,0.001031766,0.0002464327,0.19847976],"genre_scores_gemma":[0.95138556,0.012146491,0.008563131,0.012367976,0.00014153728,0.00016117733,0.00027331559,0.00014243166,0.014818434],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.9222551,0.014961291,0.0035028993,0.0044714534,0.030864846,0.023944395],"domain_scores_gemma":[0.8682653,0.041377485,0.0067691817,0.0057901237,0.043484114,0.034313783],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.046852283,0.00078249496,0.0014737892,0.009681095,0.065986164,0.04332751,0.0056142304,0.007848385,0.0073453533],"category_scores_gemma":[0.074271716,0.0012285484,0.00121387,0.01728475,0.05542504,0.017358908,0.02174348,0.013154026,0.0004791442],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000048666276,0.000047173155,0.0069786725,0.00047121965,0.000041366824,0.0003583158,0.06404922,0.0012199788,0.00045898178,0.8798814,0.018549599,0.02789543],"study_design_scores_gemma":[0.000053582764,0.00006309909,0.025440106,0.0024332008,0.0001198578,0.00025134033,0.24627233,0.0024326416,0.0008924017,0.17070226,0.55088633,0.00045282417],"about_ca_topic_score_codex":0.9892852,"about_ca_topic_score_gemma":0.9896375,"teacher_disagreement_score":0.9943858,"about_ca_system_score_codex":0.36407474,"about_ca_system_score_gemma":0.669642,"threshold_uncertainty_score":0.73758316},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"incentives","study_design":"qualitative","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"commentary","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"W7117312558","doi":"10.5281/zenodo.18056544","title":"Barriers and pathways for advancing open science and open scholarship in academic institutions: a Canadian perspective","year":2025,"lang":"en","type":"preprint","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Research Data Management Practices","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":true,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"Concordia University","funders":"Fonds de recherche du Québec; Social Sciences and Humanities Research Council; Canada Research Chairs","keywords":"Scholarship; Corporate governance; Equity (law); Open science; Indigenous; Engaged scholarship; Work (physics)","score_opus":0.18662070235144906,"score_gpt":0.3937475157614601,"score_spread":0.20712681341001105,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7117312558","genre_codex":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"reproducibility","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":"reproducibility","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10243761,0.016246442,0.011218743,0.66900927,0.000943964,0.0003860483,0.001031766,0.0002464327,0.19847976],"genre_scores_gemma":[0.95138556,0.012146491,0.008563131,0.012367976,0.00014153728,0.00016117733,0.00027331559,0.00014243166,0.014818434],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.9222551,0.014961291,0.0035028993,0.0044714534,0.030864846,0.023944395],"domain_scores_gemma":[0.8682653,0.041377485,0.0067691817,0.0057901237,0.043484114,0.034313783],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.046852283,0.00078249496,0.0014737892,0.009681095,0.065986164,0.04332751,0.0056142304,0.007848385,0.0073453533],"category_scores_gemma":[0.074271716,0.0012285484,0.00121387,0.01728475,0.05542504,0.017358908,0.02174348,0.013154026,0.0004791442],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000048666276,0.000047173155,0.0069786725,0.00047121965,0.000041366824,0.0003583158,0.06404922,0.0012199788,0.00045898178,0.8798814,0.018549599,0.02789543],"study_design_scores_gemma":[0.000053582764,0.00006309909,0.025440106,0.0024332008,0.0001198578,0.00025134033,0.24627233,0.0024326416,0.0008924017,0.17070226,0.55088633,0.00045282417],"about_ca_topic_score_codex":0.9892852,"about_ca_topic_score_gemma":0.9896375,"teacher_disagreement_score":0.9943858,"about_ca_system_score_codex":0.36407474,"about_ca_system_score_gemma":0.669642,"threshold_uncertainty_score":0.73758316},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"design_other","genre":"commentary","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"W7134613536","doi":"","title":"The rise in subscription fees and the gold open access publishing, examined from financial reports of major publishers","year":2024,"lang":"ja","type":"article","venue":"Kyoto University Research Information Repository (Kyoto University)","topic":"scientometrics and bibliometrics research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Payment; Financial plan; Quarter (Canadian coin); Work (physics); Financial services","score_opus":0.30010216013937585,"score_gpt":0.4569942950357946,"score_spread":0.15689213489641873,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7134613536","genre_codex":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":"evaluation","model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"evaluation","domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9020427,0.008046234,0.00075406395,0.014912359,0.00080646283,0.00005317723,0.004792862,0.00016776173,0.06842435],"genre_scores_gemma":[0.99048984,0.0012450537,0.00023483949,0.00043630076,0.0007028692,0.000010248312,0.0009890761,0.00003114043,0.0058605713],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9896157,0.0013628909,0.0014047865,0.0007164544,0.0058708536,0.0010292606],"domain_scores_gemma":[0.74173254,0.1267457,0.066105135,0.0067533967,0.041282546,0.017380651],"candidate_categories":["metaresearch","bibliometrics","open_science"],"consensus_categories":[],"category_scores_codex":[0.010969439,0.00021817049,0.00045460922,0.012692675,0.0014936251,0.010644554,0.0009867807,0.0013132753,0.0062611327],"category_scores_gemma":[0.12799107,0.00026270552,0.0004681076,0.01726871,0.0021682892,0.0058273477,0.0023626674,0.0017543763,0.0010732576],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0010111926,0.0001491101,0.8489381,0.0003423769,0.00016191552,0.00078905415,0.0046039214,0.00033483063,0.0009646659,0.039684605,0.017199995,0.0858202],"study_design_scores_gemma":[0.000030099514,0.00017091446,0.96228665,0.00010229418,0.000114579125,0.00049673603,0.003967538,0.00063154195,0.0012821368,0.0036513407,0.027212195,0.00005396826],"about_ca_topic_score_codex":0.008866602,"about_ca_topic_score_gemma":0.010336837,"teacher_disagreement_score":0.99901325,"about_ca_system_score_codex":0.00372393,"about_ca_system_score_gemma":0.0035074025,"threshold_uncertainty_score":0.058012664},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","scholarly_communication"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W7157585381","doi":"10.5281/zenodo.19897166","title":"Contouren Open Science Infrastructuur","year":2025,"lang":"","type":"report","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Canarie","funders":"","keywords":"Open university; Plan (archaeology); Open science","score_opus":0.06602567376427049,"score_gpt":0.32431980990169296,"score_spread":0.25829413613742247,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W7157585381","genre_codex":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05744018,0.0033986499,0.21290527,0.009104267,0.0013973048,0.0014280488,0.0031081822,0.0049586995,0.7062593],"genre_scores_gemma":[0.47394493,0.0055336943,0.18168202,0.0016420534,0.0005233569,0.0023634487,0.012925886,0.0029969395,0.31838766],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9869724,0.0015674566,0.00050026027,0.0012358091,0.008466931,0.0012571337],"domain_scores_gemma":[0.9807966,0.0039571146,0.0013355383,0.0044796644,0.0063925623,0.003038483],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.010224828,0.0007987071,0.00071082276,0.0039315904,0.003124314,0.010532012,0.002439255,0.0032043105,0.044400793],"category_scores_gemma":[0.0146082,0.00065494276,0.00079573295,0.0049716127,0.003031123,0.009521257,0.012632376,0.0028800387,0.013354777],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00036954792,0.0002396905,0.0051080044,0.0006045248,0.000036288737,0.0004035886,0.0017671774,0.014657453,0.011348041,0.64438975,0.052568782,0.26850718],"study_design_scores_gemma":[0.00006431922,0.0002684444,0.0047034943,0.00030801113,0.000027109723,0.00028536026,0.0019380704,0.011032234,0.010944251,0.15743496,0.81291,0.0000837656],"about_ca_topic_score_codex":0.006266128,"about_ca_topic_score_gemma":0.004942427,"teacher_disagreement_score":0.99756074,"about_ca_system_score_codex":0.0051426794,"about_ca_system_score_gemma":0.010449266,"threshold_uncertainty_score":0.14853543},"labels":[{"model":"gemma","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science","scholarly_communication"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W935588458","doi":"","title":"Open Access: Accessibility for the Ivory Tower","year":2013,"lang":"en","type":"article","venue":"Memorial University Research Repository (Memorial University)","topic":"Community Development and Social Impact","field":"Economics, Econometrics and Finance","cited_by":0,"is_retracted":false,"has_abstract":true,"route_ca_aff":false,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":true,"ca_institutions":"","funders":"","keywords":"Ivory tower; Political science; Public relations; Scholarly communication; Government (linguistics); Institution; Library science; Business; Sociology; Publishing; Computer science","score_opus":0.15168469941009774,"score_gpt":0.33538985923685194,"score_spread":0.1837051598267542,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W935588458","genre_codex":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0030171992,0.005748346,0.034173563,0.050691597,0.005399987,0.00026152917,0.0023084292,0.01955377,0.87884563],"genre_scores_gemma":[0.03958626,0.007939756,0.039493285,0.015273749,0.0027315035,0.00054039457,0.0035262825,0.010717904,0.88019097],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9937623,0.0016371551,0.0004625089,0.0006961623,0.002538509,0.00090344466],"domain_scores_gemma":[0.9776544,0.0049246773,0.0007663684,0.0057800026,0.0037893532,0.007085205],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.007551778,0.00058728684,0.0006050357,0.0024951654,0.006485669,0.017935436,0.0025538139,0.004171276,0.24071622],"category_scores_gemma":[0.034585357,0.00077720155,0.00062341976,0.002334144,0.004678622,0.02490081,0.018953044,0.003594839,0.111550935],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000047456975,0.00002941005,0.00042358492,0.00018215553,0.0000038288695,0.00013477061,0.0012934915,0.00005156311,0.00052668556,0.07272393,0.7817934,0.1427897],"study_design_scores_gemma":[0.0000050995054,0.000010693676,0.00029982647,0.00020588661,0.0000015781658,0.00009542084,0.000357203,0.00006254414,0.00008760585,0.007485758,0.9913743,0.000014020531],"about_ca_topic_score_codex":0.009127099,"about_ca_topic_score_gemma":0.018210389,"teacher_disagreement_score":0.9974462,"about_ca_system_score_codex":0.0033678415,"about_ca_system_score_gemma":0.010044841,"threshold_uncertainty_score":0.805276},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"}]}