{"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002805374,0.00010755216,0.000096106094,0.0010401091,0.0014798145,0.0010331887,0.97730553,0.000028922532,0.016103372],"genre_scores_gemma":[0.003145351,0.00052578055,0.00009865251,0.00079223304,0.0002240322,0.0000028366721,0.9209767,0.000046112455,0.07418828],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99354976,0.0006966318,0.0010238758,0.0014982534,0.0025952284,0.00063625287],"domain_scores_gemma":[0.99136853,0.0029211377,0.0016781067,0.0027278515,0.00093979377,0.00036455737],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.0033877518,0.0005341216,0.0014278793,0.002834717,0.00026836834,0.004738296,0.05463035,0.0009594284,0.003117792],"category_scores_gemma":[0.0042017,0.00047225098,0.0003406408,0.0054737525,0.0003300438,0.010649614,0.016851956,0.0015467983,0.002127799],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00012759681,0.00008318803,0.05585783,0.00008800668,0.00009177523,0.00007270437,0.0000041359963,0.00002205951,1.4008006e-7,0.00017196512,0.9431179,0.00036266318],"study_design_scores_gemma":[0.0008726522,0.000056353856,0.0009566121,0.00019514083,0.00014080909,1.6200595e-8,0.0016164411,0.000032927095,0.000035063586,0.0019640005,0.99356914,0.0005608437],"about_ca_topic_score_codex":0.006697081,"about_ca_topic_score_gemma":0.014141517,"teacher_disagreement_score":0.058084905,"about_ca_system_score_codex":0.0002515573,"about_ca_system_score_gemma":0.0012375333,"threshold_uncertainty_score":0.9999174},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W1495039464","doi":"10.7910/dvn/sikkdv","title":"Dramatic Growth of Open Access","year":2010,"lang":"en","type":"dataset","venue":"uO Research (University of Ottawa)","topic":"Academic Publishing and Open Access","field":"Decision Sciences","cited_by":8,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Simon Fraser University","funders":"","keywords":"Computer science","score_opus":0.3336734854130005,"score_gpt":0.5300456706965908,"score_spread":0.1963721852835903,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W1495039464","genre_codex":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0028061757,0.000025827649,0.00009219905,0.004044221,0.00037869366,0.000812066,0.9565208,0.000013134564,0.035306897],"genre_scores_gemma":[0.0118156625,0.0007828812,0.0021347217,0.0002453661,0.00042598537,0.000005126705,0.9332267,0.00006135182,0.051302217],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9903596,0.0012921139,0.000662358,0.0009698327,0.0060916566,0.00062443456],"domain_scores_gemma":[0.98886585,0.004274911,0.0011388861,0.0022925665,0.0030095198,0.00041828433],"candidate_categories":["metaresearch","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["open_science"],"category_scores_codex":[0.02206549,0.00023248619,0.0010230504,0.0029294125,0.0005936166,0.0029944465,0.06933033,0.00086969463,0.008761232],"category_scores_gemma":[0.02012994,0.00021526398,0.00021952778,0.0044313893,0.0015264483,0.0075706393,0.023502786,0.0033369705,0.00063357974],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00001301601,0.000112688154,0.0029221976,0.00011043335,0.000045024462,0.00003841874,0.00009360436,8.620935e-7,0.000015369105,0.0014587428,0.99317694,0.0020127203],"study_design_scores_gemma":[0.0004523299,0.00007251802,0.005411168,0.00011653565,0.00002484909,0.000003593305,0.0008947082,0.00004078101,0.000060965744,0.013700164,0.9790104,0.00021201302],"about_ca_topic_score_codex":0.015678521,"about_ca_topic_score_gemma":0.007945547,"teacher_disagreement_score":0.04582754,"about_ca_system_score_codex":0.000092800066,"about_ca_system_score_gemma":0.0013680337,"threshold_uncertainty_score":0.99896234},"labels":[{"model":"gemma","categories":["bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"not_applicable","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"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":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00009133098,0.00063143607,0.00058478425,0.011926852,0.0049714264,0.0013215666,0.000002961378,0.00019166018,0.98027796],"genre_scores_gemma":[0.005612032,0.001741099,0.022867631,0.0014804329,0.0017867056,0.0007264458,0.00015513855,0.0000860337,0.96554446],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99207383,0.002442859,0.00066901895,0.0015474216,0.0020205197,0.0012463479],"domain_scores_gemma":[0.9919498,0.00095133303,0.0002223097,0.0026419396,0.003736008,0.000498606],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.013891612,0.00032024924,0.00045311364,0.0023886927,0.0013414152,0.016499802,0.033759702,0.00044564044,0.0063458215],"category_scores_gemma":[0.0014473074,0.0002873155,0.000093551695,0.003957125,0.00029509363,0.0059544123,0.02169647,0.0031476687,0.00695265],"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":[0.000004045331,0.00018992927,0.00007082899,0.00006049193,0.00001483621,0.000002414145,0.0012445317,6.7450935e-7,0.0000026245946,0.19918105,0.76761687,0.031611685],"study_design_scores_gemma":[0.00010346416,0.00011767297,0.0009588075,0.00032096577,0.0000031137834,0.000009818705,0.0049371733,0.00006590308,0.000091488735,0.035570834,0.9574368,0.00038396873],"about_ca_topic_score_codex":0.018824467,"about_ca_topic_score_gemma":0.0003835009,"teacher_disagreement_score":0.1898199,"about_ca_system_score_codex":0.00047254213,"about_ca_system_score_gemma":0.011770475,"threshold_uncertainty_score":0.9999587},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015373293,0.0008069329,0.018538026,0.019759879,0.0019095343,0.0013801444,0.00040220068,0.00015434505,0.94167566],"genre_scores_gemma":[0.82548445,0.0013910114,0.01880445,0.053181157,0.000518479,0.00013413922,0.0027467378,0.00006425293,0.09767529],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99732745,0.00025059652,0.00071906956,0.00035618848,0.00062132283,0.0007254019],"domain_scores_gemma":[0.9967083,0.00016488634,0.00050890126,0.0013902358,0.00045945775,0.0007682596],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.0020576518,0.00033280288,0.0003234392,0.00012188333,0.00022788372,0.07679322,0.013531642,0.00017842508,0.0001770591],"category_scores_gemma":[0.00079214585,0.00027744565,0.00005157828,0.0005781372,0.00012093239,0.280329,0.023677012,0.00027461624,0.0011955234],"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":[0.00006923111,0.00021915912,0.0015517913,0.00008456059,0.000087978944,0.000040529165,0.010943085,0.00085360225,0.000005755375,0.14904448,0.76678455,0.07031529],"study_design_scores_gemma":[0.0016336393,0.000107862244,0.004692687,0.00025331834,0.000026497286,0.00019668622,0.00020526297,0.014069011,0.00020720987,0.020598441,0.9575899,0.00041945686],"about_ca_topic_score_codex":0.025855076,"about_ca_topic_score_gemma":0.0010352167,"teacher_disagreement_score":0.84400034,"about_ca_system_score_codex":0.00038120773,"about_ca_system_score_gemma":0.001948168,"threshold_uncertainty_score":0.99996775},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8561742,0.00003910972,0.00035469185,0.0027628967,0.00007141916,0.00014707341,0.000009597617,0.00013625891,0.14030476],"genre_scores_gemma":[0.9988528,0.0000046658297,0.0005690994,0.0002629407,0.000018818291,0.00004071988,0.000016695492,0.000001325,0.00023292711],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99972546,0.000009458621,0.00010596354,0.00005056012,0.000049852766,0.00005867902],"domain_scores_gemma":[0.9998379,0.000025943691,0.000038287162,0.00005762829,0.00001711566,0.000023158149],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000030112902,0.000050273924,0.00004965189,0.00016939134,0.00030190128,0.00040045628,0.000071463925,0.000009119928,0.00000386916],"category_scores_gemma":[0.000009885008,0.00003067226,0.0000052688947,0.00004129266,0.0010148592,0.0048280517,0.000060409424,0.00007190699,0.0000036242907],"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":[0.000067199406,0.000018012941,0.0016653696,0.0000083975265,0.0000070357332,0.000002052353,0.027210144,0.000007234043,1.5833751e-7,0.9692053,0.000059531045,0.0017495288],"study_design_scores_gemma":[0.002077127,0.00023454314,0.0155456485,0.00006939766,0.000014784052,0.00018207806,0.06216539,0.0022520304,0.00007128052,0.35112074,0.5660381,0.00022887178],"about_ca_topic_score_codex":0.000018388164,"about_ca_topic_score_gemma":0.000021058415,"teacher_disagreement_score":0.6180846,"about_ca_system_score_codex":0.0000042327183,"about_ca_system_score_gemma":0.000025135088,"threshold_uncertainty_score":0.3861609},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97673607,0.0014142409,0.000020798663,0.011104146,0.0014591358,0.0007347016,0.00014969145,0.00007400609,0.008307201],"genre_scores_gemma":[0.9497935,0.00067669264,0.0007257066,0.04703065,0.0010946447,0.000039061095,0.00014420274,0.00002081119,0.00047473394],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9973547,0.00030243213,0.00061134,0.0009277374,0.000113437054,0.0006903628],"domain_scores_gemma":[0.99825627,0.00048582736,0.00013693304,0.00030975696,0.00010485581,0.00070633437],"candidate_categories":["metaepi_narrow","research_integrity"],"consensus_categories":["research_integrity"],"category_scores_codex":[0.0009726431,0.00035238377,0.0006642779,0.000089461566,0.00010079432,0.00018131359,0.0022431097,0.0013897349,0.00044668195],"category_scores_gemma":[0.00048708054,0.00028851736,0.0001467608,0.00028156763,0.00034601564,0.00030647984,0.0019163983,0.002484904,0.00007566068],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0009919782,0.00070463703,0.014167578,0.000015226823,0.0001395384,0.000028577635,0.000073297604,0.0000010186244,0.78736293,0.005056832,0.05536571,0.13609266],"study_design_scores_gemma":[0.0020054644,0.00011503565,0.0027292243,0.000006801678,0.0001322516,0.00003150999,0.000006475084,0.000033438617,0.5688101,0.0071957447,0.41853917,0.00039476462],"about_ca_topic_score_codex":0.000006378462,"about_ca_topic_score_gemma":0.0000019468036,"teacher_disagreement_score":0.36317345,"about_ca_system_score_codex":0.000035423986,"about_ca_system_score_gemma":0.00013338674,"threshold_uncertainty_score":0.9999567},"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":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2400961,0.0012637585,0.015061325,0.71348614,0.004874718,0.00079123094,0.00001596056,0.00014857386,0.024262182],"genre_scores_gemma":[0.871951,0.0012431002,0.006297165,0.108606786,0.005838982,0.000024031224,0.000006672779,0.00008184791,0.005950429],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9971593,0.00050442794,0.0008512347,0.00031030865,0.0007194197,0.00045530454],"domain_scores_gemma":[0.99618477,0.0018384783,0.00081794447,0.0007477337,0.000072434545,0.0003386293],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.0026139577,0.00025608423,0.00034742182,0.00023191737,0.0003559854,0.0017810868,0.036948353,0.00026859136,0.00022259796],"category_scores_gemma":[0.00030203167,0.00014631481,0.00010844913,0.00080833025,0.00014465014,0.013627508,0.0100080995,0.0019180088,0.00014791182],"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":[0.0010906062,0.00008650939,0.029403264,0.000059362646,0.00040822782,0.0000407459,0.041037913,0.0024973666,0.0007600677,0.14457078,0.7591947,0.020850493],"study_design_scores_gemma":[0.00084595836,0.000240045,0.08986497,0.00023220849,0.000042920976,0.0002266094,0.00034922062,0.0023112083,0.002124369,0.048141275,0.85529524,0.0003259621],"about_ca_topic_score_codex":0.000044671127,"about_ca_topic_score_gemma":0.000009422725,"teacher_disagreement_score":0.6318549,"about_ca_system_score_codex":0.00005086541,"about_ca_system_score_gemma":0.0003547965,"threshold_uncertainty_score":0.9992552},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1408229,0.010931687,0.019675395,0.07371772,0.004405637,0.25619555,0.0008109328,0.00011427442,0.49332592],"genre_scores_gemma":[0.9168285,0.00037608636,0.023336282,0.007845412,0.0024279172,0.02198249,0.000028123142,0.00003987004,0.027135294],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9873357,0.0038680837,0.0052013197,0.00033078456,0.0031446675,0.00011946659],"domain_scores_gemma":[0.98239815,0.0011722906,0.0046188356,0.010981448,0.00077638443,0.000052889187],"candidate_categories":["metaresearch","open_science","insufficient_payload"],"consensus_categories":["metaresearch","insufficient_payload"],"category_scores_codex":[0.099032834,0.000121820325,0.0034386406,0.000051350275,0.00004429752,0.00047080815,0.007977618,0.0000359237,0.006925179],"category_scores_gemma":[0.049883712,0.00004036322,0.00063465483,0.0003702229,0.00006140624,0.00016273036,0.00007727091,0.00016378792,0.0011276095],"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":[0.000021068341,0.00020708404,0.031412873,0.14853244,0.000640763,0.00000524328,0.00065640395,0.00001146289,0.0035690714,0.09331319,0.7190724,0.0025580206],"study_design_scores_gemma":[0.001036418,0.00013958963,0.02423659,0.071926475,0.0016515012,0.00018119483,0.00059059693,0.007907512,0.00073598954,0.029988496,0.8609088,0.0006968519],"about_ca_topic_score_codex":0.0000034809455,"about_ca_topic_score_gemma":0.000013835717,"teacher_disagreement_score":0.7760056,"about_ca_system_score_codex":0.0000040452082,"about_ca_system_score_gemma":0.000028541206,"threshold_uncertainty_score":0.9996501},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.961042,0.00032025776,0.000010151574,0.031301416,0.00016770425,0.0007228729,0.000065744556,0.00002189091,0.006347998],"genre_scores_gemma":[0.99540895,0.0018092466,0.00013342751,0.00011976265,0.00036450732,0.0000090876965,0.00001870145,0.000030810257,0.002105526],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9632889,0.020969765,0.0016643001,0.0010999442,0.011331445,0.0016456691],"domain_scores_gemma":[0.9636017,0.020575505,0.0007775157,0.0016739488,0.012368725,0.0010026215],"candidate_categories":["metaresearch","sts","research_integrity"],"consensus_categories":["metaresearch","sts"],"category_scores_codex":[0.16406792,0.0002432965,0.00076927635,0.0008491003,0.0055676964,0.00019012824,0.0027570426,0.00073931395,0.0001220694],"category_scores_gemma":[0.24594414,0.00016300472,0.00007444591,0.0018448173,0.031153504,0.00053586677,0.0011356695,0.013888321,0.000015931182],"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":[0.008317265,0.0058374405,0.51586026,0.00013420121,0.0004570931,0.012991819,0.30827788,0.000042954045,0.00009907316,0.0063561606,0.14137924,0.00024662394],"study_design_scores_gemma":[0.0042835874,0.027428359,0.6227909,0.0009075611,0.00004189668,0.00023753624,0.30023924,0.00004142024,0.0001273608,0.0026070238,0.040721852,0.00057327485],"about_ca_topic_score_codex":0.02905082,"about_ca_topic_score_gemma":0.04352489,"teacher_disagreement_score":0.10693061,"about_ca_system_score_codex":0.00079733675,"about_ca_system_score_gemma":0.0049814214,"threshold_uncertainty_score":0.99572694},"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":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0022183335,0.00085331214,0.32670465,0.5732855,0.0003756198,0.002966242,0.000008034044,0.00040107043,0.09318723],"genre_scores_gemma":[0.2282679,0.0006063685,0.5533985,0.04203348,0.00294981,0.001445963,0.000118819444,0.00016253427,0.17101666],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99706364,0.00022467079,0.0004142456,0.00093155925,0.0005717836,0.0007941103],"domain_scores_gemma":[0.99738955,0.0009936877,0.0002644746,0.0010363741,0.000060751183,0.0002551564],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.003685189,0.0001744673,0.00023049595,0.00071040075,0.00021966598,0.16114776,0.02349629,0.00010150277,0.00006901741],"category_scores_gemma":[0.004915321,0.00017244976,0.000043984885,0.0022909353,0.00002531411,0.6376082,0.014567608,0.0010266496,0.000024299146],"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":[0.000059479215,0.00005232381,0.0015292306,0.000045532735,0.00001837216,0.00001254129,0.0012545236,0.00073660427,0.00013555361,0.65594226,0.09982763,0.24038593],"study_design_scores_gemma":[0.0008941538,0.000020433947,0.000516828,0.00008286473,0.0000035964497,0.0000023312764,0.00020230886,0.03389058,0.00038530922,0.026125636,0.93763983,0.00023610708],"about_ca_topic_score_codex":0.0014785379,"about_ca_topic_score_gemma":0.0014051701,"teacher_disagreement_score":0.83781224,"about_ca_system_score_codex":0.00014774877,"about_ca_system_score_gemma":0.0005718308,"threshold_uncertainty_score":0.99340236},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.993377,0.00083870493,0.0033990587,0.0002726731,0.00057262124,0.00076688366,0.00026424203,0.0000075521525,0.0005013076],"genre_scores_gemma":[0.9958612,0.000036520305,0.0028976956,0.00008767275,0.0006299022,0.000017626771,0.00021414978,0.000009988876,0.00024525283],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.7854382,0.17160764,0.028186068,0.0026604498,0.011494203,0.00061345997],"domain_scores_gemma":[0.80584264,0.13628195,0.024103839,0.031097516,0.0018607187,0.0008133398],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.6082692,0.00019183426,0.004514113,0.000094122835,0.00015266467,0.0011387629,0.0050373226,0.000093966024,0.0008420367],"category_scores_gemma":[0.35196626,0.00008102434,0.00067618885,0.00044891395,0.0000864652,0.00046462542,0.0009303371,0.00023112979,0.00013108685],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006788341,0.0012352641,0.9691786,0.00004135166,0.003236508,0.0000011933126,0.001006524,0.00003224213,0.0003571811,0.00024427593,0.017251922,0.0073470287],"study_design_scores_gemma":[0.000706616,0.00016136376,0.8253631,0.00036351982,0.0041403878,9.706517e-7,0.00078062515,0.16229336,0.0003983456,0.0013104504,0.004227715,0.00025358237],"about_ca_topic_score_codex":0.00031878788,"about_ca_topic_score_gemma":0.00024156421,"teacher_disagreement_score":0.25630295,"about_ca_system_score_codex":0.000009410888,"about_ca_system_score_gemma":0.000068047986,"threshold_uncertainty_score":0.99989814},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5131612,0.00023778145,0.041510828,0.030303929,0.002384463,0.0013653964,0.000071451264,0.00022828752,0.41073668],"genre_scores_gemma":[0.9875864,0.0000012411649,0.0018270394,0.0010342989,0.00044131695,0.000016387563,0.00021208823,0.0000091875345,0.008872085],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99960047,0.000025722851,0.0000759833,0.00012465281,0.0000891569,0.00008401326],"domain_scores_gemma":[0.99964875,0.000025599395,0.000049299444,0.00017117271,0.000068469286,0.000036694473],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00007844525,0.000063472224,0.00007076935,0.000027268947,0.00010632592,0.00037321053,0.000363649,0.00014915757,0.00046936848],"category_scores_gemma":[0.000008705993,0.000052317835,0.00002961037,0.00018242575,0.000011175665,0.0005016145,0.00011625861,0.0003655432,0.00006453383],"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":[0.000009855603,0.000075123564,0.13224064,0.0000055134956,0.00003189562,1.1968248e-7,0.00025291025,0.00045549817,0.0013334955,0.7742662,0.026351918,0.06497684],"study_design_scores_gemma":[0.0003592836,0.00000950007,0.19130683,0.000012885666,0.000007808739,8.965802e-8,0.00008822516,0.0016057139,0.0030533639,0.02158269,0.78184676,0.00012685946],"about_ca_topic_score_codex":0.000024199418,"about_ca_topic_score_gemma":0.000023421322,"teacher_disagreement_score":0.75549483,"about_ca_system_score_codex":0.0000037445575,"about_ca_system_score_gemma":0.000022610413,"threshold_uncertainty_score":0.51392555},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.37682796,0.00059392,2.3698031e-7,0.013153841,0.00016562012,0.0014100125,0.000031823947,0.000055228847,0.6077614],"genre_scores_gemma":[0.90043217,0.00032578764,0.0000031384363,0.0009681346,0.00036554667,0.000028792334,0.00001503532,0.00002903525,0.097832374],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982313,0.0008418934,0.00022730787,0.0002707224,0.00010819551,0.00032060297],"domain_scores_gemma":[0.9985415,0.00079692167,0.00012673088,0.0004366884,0.000023933626,0.00007427172],"candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0009807032,0.00018752553,0.0002939022,0.00006167886,0.0025228283,0.022832468,0.003512637,0.000049113834,0.0004816056],"category_scores_gemma":[0.00008847293,0.00009142421,0.00004335936,0.00013125465,0.0014443116,0.012144339,0.0050776587,0.00022828778,0.000013905133],"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":[0.0006918315,0.00004751086,0.0011447897,0.0000113648275,0.000042738597,0.000008932776,0.027950507,2.5268508e-7,2.5034322e-7,0.92287004,0.045558866,0.0016728953],"study_design_scores_gemma":[0.001671254,0.000048320107,0.0010494732,0.000028292145,0.000015001284,0.0000018555163,0.027447477,0.000019135112,0.000012662026,0.02559663,0.94394326,0.00016665953],"about_ca_topic_score_codex":0.0011945253,"about_ca_topic_score_gemma":0.015095785,"teacher_disagreement_score":0.8983844,"about_ca_system_score_codex":0.000010267104,"about_ca_system_score_gemma":0.00026099835,"threshold_uncertainty_score":0.9987758},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.985036,0.000013393326,0.0020857248,0.005758423,0.0026938855,0.0012601076,0.0014882119,0.00009328441,0.0015709479],"genre_scores_gemma":[0.99565446,6.985337e-7,0.00250692,0.0010097339,0.00011187099,0.000045102133,0.0001865978,0.000012577463,0.0004720448],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.9916981,0.0003175534,0.00032182087,0.006628861,0.00060151477,0.00043216688],"domain_scores_gemma":[0.980438,0.008017742,0.00014960382,0.010957684,0.00025180067,0.00018515682],"candidate_categories":["metaresearch","sts","scholarly_communication"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.045951333,0.0001399891,0.00021128553,0.00021328109,0.0015251766,0.0020607554,0.004274282,0.000037929338,0.000016237267],"category_scores_gemma":[0.31082746,0.00013240917,0.000014034398,0.0013198466,0.0024036733,0.004385597,0.0055266838,0.00017132556,0.000007350997],"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":[0.0009550698,0.0014025193,0.038028285,0.00019480473,0.000006193299,0.0000027102208,0.0014231217,0.0020115268,0.6608348,0.049150564,0.20107481,0.0449156],"study_design_scores_gemma":[0.0020008415,0.0003539762,0.11421809,0.000054879663,0.000017482944,0.000022839946,0.00064557116,0.20929624,0.06193011,0.067114115,0.54360527,0.0007405782],"about_ca_topic_score_codex":0.00012821886,"about_ca_topic_score_gemma":0.00048130722,"teacher_disagreement_score":0.59890467,"about_ca_system_score_codex":0.00011837585,"about_ca_system_score_gemma":0.00032684876,"threshold_uncertainty_score":0.9997747},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.26716554,0.000016336013,0.041380543,0.0005299265,0.00015756978,0.00050147984,0.00001828798,0.00020220102,0.69002813],"genre_scores_gemma":[0.9740959,0.000011167678,0.018367331,0.000018946055,0.000013157467,3.953956e-7,0.000015618813,0.00000768986,0.0074698385],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99890506,0.00018097807,0.00008706462,0.0004019751,0.00022420095,0.00020070263],"domain_scores_gemma":[0.998918,0.00006896253,0.00020394455,0.00044983704,0.00017699222,0.00018228401],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000105926134,0.00013292252,0.00020638779,0.00021436339,0.00026547487,0.00027846973,0.0049719424,0.000055662298,0.00026679973],"category_scores_gemma":[0.0000119438555,0.00014090985,0.000033780838,0.0005657772,0.00012113935,0.007895249,0.0013673767,0.00016826754,0.000007289018],"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":[0.00060238695,0.002861159,0.05043877,0.00007647157,0.00023342916,0.00014932745,0.021623662,0.00071601244,0.000315202,0.9097012,0.010869819,0.0024125508],"study_design_scores_gemma":[0.0062346775,0.0034484926,0.37344152,0.00035419493,0.00029318532,0.000050638435,0.092574224,0.011392922,0.027230695,0.0008511541,0.4821839,0.0019443905],"about_ca_topic_score_codex":0.0032682423,"about_ca_topic_score_gemma":0.0005116962,"teacher_disagreement_score":0.9088501,"about_ca_system_score_codex":0.00007047,"about_ca_system_score_gemma":0.0002480615,"threshold_uncertainty_score":0.9239189},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9289367,0.0022567227,0.00016341195,0.058969308,0.00015693171,0.0021149488,0.00006175107,0.000018940538,0.0073212446],"genre_scores_gemma":[0.9679515,0.025145859,0.0018516013,0.00025294756,0.002655523,0.00007642092,0.000011453603,0.000081128615,0.0019736183],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9594518,0.012907744,0.0021075679,0.0017209656,0.021570379,0.0022415433],"domain_scores_gemma":[0.4546177,0.49278155,0.0008001405,0.0048704324,0.044349115,0.002581043],"candidate_categories":["metaresearch","sts","research_integrity"],"consensus_categories":["metaresearch","sts","research_integrity"],"category_scores_codex":[0.27513498,0.00033558783,0.0012284024,0.002118206,0.0020985939,0.00032767025,0.004078084,0.0018221934,0.00007370402],"category_scores_gemma":[0.63636756,0.00023522708,0.0003380431,0.002377864,0.007802312,0.000267104,0.0034442125,0.046188664,0.000033022174],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.04950777,0.025709907,0.12251766,0.0149318855,0.0011493902,0.001565068,0.029118776,0.000047308156,0.029990254,0.44005305,0.2742024,0.011206509],"study_design_scores_gemma":[0.0148279,0.19971126,0.3230372,0.009689464,0.00012035262,0.00018783788,0.019296756,0.0020490459,0.0054924376,0.33182985,0.09274006,0.001017824],"about_ca_topic_score_codex":0.00011661357,"about_ca_topic_score_gemma":0.00005286781,"teacher_disagreement_score":0.50483406,"about_ca_system_score_codex":0.001040673,"about_ca_system_score_gemma":0.0079002,"threshold_uncertainty_score":0.99947363},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010029504,0.0005079362,0.00096879574,0.970761,0.0015490052,0.0034183634,0.000022985272,0.000110628505,0.012631787],"genre_scores_gemma":[0.7497126,0.0010624442,0.0031529537,0.20080152,0.0011156051,0.000081115904,0.00009341521,0.00005798001,0.043922372],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9951163,0.0014498448,0.0013108103,0.0003863588,0.0006545859,0.0010821279],"domain_scores_gemma":[0.9968941,0.0002610958,0.0011959018,0.0006039603,0.0005668059,0.00047814066],"candidate_categories":["sts","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.005095324,0.00017030616,0.00048450776,0.00028894312,0.0028282043,0.0001447849,0.0021600542,0.00007156246,0.0041501257],"category_scores_gemma":[0.0020310713,0.0001269879,0.000037633374,0.0024628176,0.00009515479,0.00026437527,0.0010597524,0.00062046695,0.00031770658],"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":[0.000026541202,0.00010320265,0.009331852,0.000576366,0.0000036599085,1.4772046e-7,0.0038643784,0.000032646745,0.000023199911,0.004623667,0.9650899,0.01632444],"study_design_scores_gemma":[0.0006341153,0.00021375276,0.39697734,0.0012145856,0.0000032232808,0.0000019329768,0.0016405816,0.00010278381,0.00011695168,0.00096792704,0.5979648,0.0001619908],"about_ca_topic_score_codex":0.015439542,"about_ca_topic_score_gemma":0.0025882372,"teacher_disagreement_score":0.76995945,"about_ca_system_score_codex":0.0010572012,"about_ca_system_score_gemma":0.01025206,"threshold_uncertainty_score":0.99846995},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8275195,0.00031908942,0.007995857,0.0856018,0.002240911,0.011555812,0.0015596983,0.0004496069,0.06275777],"genre_scores_gemma":[0.98991627,0.00002167126,0.008100948,0.00030354512,0.000027710666,0.00021145439,0.0012948695,0.0000049958185,0.00011853515],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988743,0.00004403071,0.00047072227,0.00014553213,0.00028009937,0.00018528076],"domain_scores_gemma":[0.9986237,0.00006665379,0.0003153466,0.00080867775,0.00014501992,0.00004055773],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0036692955,0.000074918884,0.0001575124,0.00013815273,0.00021600118,0.00045967108,0.010902206,0.00012392605,0.000017650034],"category_scores_gemma":[0.00042724225,0.000055558136,0.0000041871317,0.00033378912,0.00021165637,0.007721395,0.0070505883,0.00007670867,0.000019025223],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015936406,0.0001509944,0.024192905,0.00008664742,0.00009598296,6.264702e-7,0.15117617,0.000013146414,0.000004216891,0.32311228,0.030278282,0.47072938],"study_design_scores_gemma":[0.00052525685,0.000038420756,0.0004427772,0.00003527969,0.000004142978,5.6235666e-7,0.14222392,0.0005827509,0.0001878229,0.0028964924,0.85296863,0.00009393564],"about_ca_topic_score_codex":0.0005753241,"about_ca_topic_score_gemma":0.09442689,"teacher_disagreement_score":0.82269037,"about_ca_system_score_codex":0.00012587429,"about_ca_system_score_gemma":0.0009812133,"threshold_uncertainty_score":0.99444926},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5842754,0.0061934763,0.011117103,0.09017685,0.003906605,0.0054930244,0.0012216143,0.00018205338,0.29743382],"genre_scores_gemma":[0.9697581,0.0014378099,0.018688133,0.00063742325,0.0018934183,0.00006372659,0.00051022146,0.00005825837,0.0069528962],"study_design_codex":"bench_or_experimental","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99649996,0.00006679207,0.0007739889,0.00072951074,0.0014706696,0.0004590987],"domain_scores_gemma":[0.9926489,0.0024858785,0.00012971481,0.0020290748,0.0024037927,0.00030261817],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0187038,0.000121291894,0.00047016307,0.00039432096,0.000075094205,0.0000069492557,0.0022833003,0.00021054148,0.002567478],"category_scores_gemma":[0.03852342,0.00009269559,0.000057323992,0.00029524555,0.001080761,0.00007482247,0.0017448837,0.0012973372,0.000053341857],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0093482165,0.0020426947,0.052934237,0.0058231563,0.002514904,0.00036631522,0.021695966,0.0000034350712,0.6035852,0.21800132,0.03506281,0.048621748],"study_design_scores_gemma":[0.024747804,0.0112690255,0.25115284,0.0031646807,0.00057218334,0.00018933324,0.006517313,0.004037544,0.015111689,0.36122626,0.32124138,0.0007699398],"about_ca_topic_score_codex":0.00049303065,"about_ca_topic_score_gemma":0.00012301454,"teacher_disagreement_score":0.5884735,"about_ca_system_score_codex":0.0001419768,"about_ca_system_score_gemma":0.00065123383,"threshold_uncertainty_score":0.9983443},"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":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.34293267,0.0008853955,0.00033446623,0.6232512,0.00020591245,0.0015397695,0.00012152341,0.000048109745,0.030680962],"genre_scores_gemma":[0.97620267,0.01066283,0.004836647,0.0043591307,0.0020868408,0.00003340275,0.000031715583,0.00011580466,0.0016709408],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.70424086,0.2211059,0.004668306,0.003341062,0.05808696,0.008556908],"domain_scores_gemma":[0.60366285,0.35781267,0.0005530997,0.0033593583,0.02920601,0.005406014],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","sts","research_integrity"],"category_scores_codex":[0.827274,0.00040030427,0.0009824377,0.008240171,0.008288706,0.0019190669,0.006793903,0.0034959326,0.000259595],"category_scores_gemma":[0.5299207,0.0003307509,0.0002039295,0.0094131855,0.026613614,0.0016446686,0.0038944816,0.10131714,0.000031132407],"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":[0.00652858,0.0016890175,0.0024297144,0.00030919447,0.0001441757,0.0020659568,0.37308723,0.0000093424305,0.0041769478,0.54155993,0.05648246,0.011517459],"study_design_scores_gemma":[0.0030509485,0.009522624,0.006893024,0.0015918054,0.00001052002,0.000019405445,0.22247592,0.000528169,0.0009159798,0.3387168,0.41561526,0.0006595504],"about_ca_topic_score_codex":0.0103558805,"about_ca_topic_score_gemma":0.052910563,"teacher_disagreement_score":0.63327,"about_ca_system_score_codex":0.0016515506,"about_ca_system_score_gemma":0.007617528,"threshold_uncertainty_score":0.99991447},"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":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000030320657,0.00013406455,0.004623726,0.019683061,0.0015818035,0.0011600726,0.9223376,0.0000930807,0.050383586],"genre_scores_gemma":[0.0000368678,0.00022601201,0.006913696,0.004157555,0.00061745575,0.00011655092,0.9637206,0.000023386432,0.02418786],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9959562,0.00043702972,0.00023565043,0.0008890818,0.0019153012,0.000566772],"domain_scores_gemma":[0.99207807,0.00026010413,0.00027009635,0.00700197,0.00022534028,0.00016441585],"candidate_categories":["scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.017203309,0.00020928858,0.00019890508,0.00031937158,0.000735379,0.019767705,0.07790623,0.000056840876,0.000052871605],"category_scores_gemma":[0.0027271828,0.00013039118,0.000027767466,0.0010377628,0.000411697,0.042261265,0.051273447,0.0009919291,0.0012957741],"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":[0.000003567245,0.000038493945,3.795542e-7,0.000010357,0.000015295682,0.000016955972,0.000042049858,0.000001824305,0.0000039812367,0.019666633,0.9778132,0.0023872715],"study_design_scores_gemma":[0.00010310716,0.00003969307,0.00001131682,0.000025107309,0.00001055165,0.000009465688,0.00002186817,0.00017185195,0.0000040629457,0.0013936879,0.9980237,0.00018556966],"about_ca_topic_score_codex":0.0022642897,"about_ca_topic_score_gemma":0.00008241272,"teacher_disagreement_score":0.041383047,"about_ca_system_score_codex":0.00014677833,"about_ca_system_score_gemma":0.000989287,"threshold_uncertainty_score":0.99948186},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.072859764,0.002949694,0.0016868983,0.85091424,0.0020659245,0.00042960857,0.0000037830816,0.000031112133,0.06905896],"genre_scores_gemma":[0.96770966,0.007696472,0.009860041,0.0017071436,0.0045915185,0.000014960958,0.0000025378367,0.000052151427,0.008365502],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.97255033,0.015850773,0.0014647585,0.00049274,0.008272412,0.0013689637],"domain_scores_gemma":[0.97241205,0.007121974,0.0015820337,0.00047141893,0.017182523,0.0012299826],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","scholarly_communication"],"category_scores_codex":[0.2771217,0.00019569775,0.00045874406,0.0014563714,0.008916861,0.4328944,0.0080743125,0.00035240132,0.0003066757],"category_scores_gemma":[0.22355494,0.00013486325,0.00013865498,0.0022988694,0.0005049698,0.47425357,0.0032032994,0.014180244,0.000018684072],"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":[0.00008234582,0.0002161258,0.012887753,0.000040692707,0.00017750895,0.00010327451,0.031568345,0.0000022390454,0.00051808055,0.11639329,0.6146766,0.22333376],"study_design_scores_gemma":[0.0009821255,0.00010528657,0.0031993145,0.0014956166,0.000025764699,0.00009297644,0.019603174,0.0000014253768,0.000018493978,0.12953655,0.84477043,0.00016881149],"about_ca_topic_score_codex":0.0048148264,"about_ca_topic_score_gemma":0.00046755705,"teacher_disagreement_score":0.8948499,"about_ca_system_score_codex":0.0014601026,"about_ca_system_score_gemma":0.0038146405,"threshold_uncertainty_score":0.99729246},"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":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.034780663,0.00008916568,0.9494325,0.010843257,0.00033208713,0.00046047365,0.0001619624,0.0000042063293,0.0038956853],"genre_scores_gemma":[0.81139463,0.0005919851,0.18241394,0.0043316064,0.00010587615,0.000003874912,0.000014597285,0.00001722995,0.0011262557],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99370617,0.0036672899,0.0016265269,0.000014809037,0.0008349512,0.0001502786],"domain_scores_gemma":[0.97593886,0.018414462,0.0018724523,0.0031597435,0.0005641405,0.000050321672],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.08041298,0.00007744776,0.00029421627,0.00009059844,0.0004548857,0.0003018768,0.018930463,0.00002436675,0.000045717305],"category_scores_gemma":[0.013282735,0.000026849175,0.00007368773,0.00023627849,0.00019624515,0.0024134414,0.0069099786,0.00026588162,0.000011668226],"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":[0.0003294932,0.00015920826,0.00006160804,0.000053955915,0.00028316904,5.1696137e-8,0.09007178,0.00034171165,0.0001714592,0.025902055,0.1515738,0.7310517],"study_design_scores_gemma":[0.0011045375,0.00014954865,0.00085303513,0.00013430396,0.00014691154,0.00001778577,0.21770251,0.0015070703,0.0012662482,0.0972589,0.6797711,0.00008803212],"about_ca_topic_score_codex":0.00008404275,"about_ca_topic_score_gemma":0.000041856,"teacher_disagreement_score":0.776614,"about_ca_system_score_codex":0.000021924321,"about_ca_system_score_gemma":0.00009757075,"threshold_uncertainty_score":0.9950288},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.885183,0.00089740637,0.020759732,0.007848407,0.004008074,0.0009939644,0.0000062020213,0.000003816206,0.080299444],"genre_scores_gemma":[0.9436078,0.00009544123,0.002271018,0.0007516469,0.0007131082,0.000005514238,2.69859e-7,0.000027335707,0.052527882],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.92723936,0.03480406,0.021887675,0.0013261342,0.014135234,0.00060751167],"domain_scores_gemma":[0.9762678,0.0039045294,0.010681974,0.004205234,0.004682285,0.00025817467],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch","insufficient_payload"],"category_scores_codex":[0.40440154,0.00020380678,0.002137567,0.0008276871,0.00015033783,0.00054311584,0.0050230725,0.000054828062,0.0017303659],"category_scores_gemma":[0.054059852,0.000076611184,0.0015996547,0.00065236434,0.00015963458,0.00021817833,0.00035545806,0.00024685098,0.0021841365],"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":[0.00006309015,0.00031602816,0.23812895,0.000005410523,0.00044589254,0.0005244435,0.0003324029,5.937703e-7,0.0011796827,0.00013604987,0.7403037,0.018563753],"study_design_scores_gemma":[0.0008026599,0.00040875457,0.43554375,0.0000723255,0.00007234915,0.00047278733,0.0001887228,2.7864203e-7,0.000008997259,0.0028742133,0.55947936,0.00007581522],"about_ca_topic_score_codex":2.0588092e-7,"about_ca_topic_score_gemma":0.0000017213356,"teacher_disagreement_score":0.3503417,"about_ca_system_score_codex":0.00004914969,"about_ca_system_score_gemma":0.000035140834,"threshold_uncertainty_score":0.99918216},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.203324,0.00037529407,0.11073678,0.006967179,0.00040317385,0.00054661086,0.00007222661,0.00093545136,0.67663926],"genre_scores_gemma":[0.9892714,0.00006923979,0.0047228043,0.0004613332,0.00010628729,4.46228e-8,0.0000231926,0.00031142213,0.0050342823],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99593645,0.0010887656,0.00023600439,0.0007325666,0.0009886674,0.0010175583],"domain_scores_gemma":[0.9964033,0.00007904624,0.00012664712,0.0006655656,0.0015639466,0.0011615432],"candidate_categories":["sts","scholarly_communication","insufficient_payload"],"consensus_categories":["sts"],"category_scores_codex":[0.008171716,0.00020485908,0.00014800082,0.00032309865,0.0023429333,0.013669537,0.0031291125,0.00010866678,0.00050357234],"category_scores_gemma":[0.006154578,0.00019111713,0.000050582978,0.0029006782,0.0032634519,0.005068193,0.003551948,0.00045177626,0.0016101059],"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":[0.000058638714,0.00088385894,0.000036291847,0.00008967469,0.000018359815,0.00022978877,0.04806615,0.0009275303,0.060075752,0.63161415,0.059304647,0.19869515],"study_design_scores_gemma":[0.00076430157,0.00082401553,0.0066595376,0.00016628888,0.000015773376,0.0034265402,0.0009908538,0.030549945,0.014021941,0.0139430985,0.9281932,0.0004444746],"about_ca_topic_score_codex":0.00022920336,"about_ca_topic_score_gemma":0.000002247331,"teacher_disagreement_score":0.86888856,"about_ca_system_score_codex":0.0014676575,"about_ca_system_score_gemma":0.00045025226,"threshold_uncertainty_score":0.9994491},"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":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.1265496,0.0072833914,0.021852689,0.7801019,0.0014235348,0.060380384,0.00041873084,0.00002640459,0.0019633893],"genre_scores_gemma":[0.9875852,0.00096046354,0.0068288436,0.0009842082,0.0013754528,0.0016626315,0.00027499278,0.000012934491,0.00031528602],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9920835,0.00035205987,0.006769392,0.00007924944,0.00061878696,0.000097008524],"domain_scores_gemma":[0.98504776,0.002747204,0.01092347,0.00011566577,0.0011026948,0.00006323726],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.021854997,0.00009234158,0.0006477699,0.0001569119,0.00007934452,0.000012435389,0.00010123438,0.000041741565,0.000121125144],"category_scores_gemma":[0.024575857,0.000049121398,0.00016045534,0.0001236733,0.00017535182,0.0004688003,0.000016420407,0.00017188993,2.2964763e-7],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.11150448,0.0054344716,0.034335922,0.019042315,0.0059749717,0.000026076308,0.0046880348,0.0018381771,0.05104076,0.01204512,0.46514282,0.28892684],"study_design_scores_gemma":[0.04758866,0.012338519,0.06539263,0.040858634,0.0015024585,0.000023360208,0.0011072896,0.0012444215,0.0126375565,0.002436894,0.8147058,0.00016378096],"about_ca_topic_score_codex":0.00074641994,"about_ca_topic_score_gemma":0.00007025313,"teacher_disagreement_score":0.8610356,"about_ca_system_score_codex":0.00008213826,"about_ca_system_score_gemma":0.00020569292,"threshold_uncertainty_score":0.98364055},"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":"codex-gemma-dda1882f352a","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00007560772,0.0009580695,0.0554145,0.008094502,0.91736954,0.00037486744,0.0000034091606,0.00037786027,0.017331632],"genre_scores_gemma":[0.014001202,0.0041807285,0.010880088,0.000051820796,0.95570725,0.00008709607,0.000043669625,0.000080345024,0.01496782],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9965605,0.00044117228,0.00052797754,0.0009058341,0.001139503,0.0004250221],"domain_scores_gemma":[0.99748516,0.00048275274,0.0006372413,0.00038317422,0.0008318137,0.00017985863],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science","research_integrity"],"consensus_categories":["open_science"],"category_scores_codex":[0.0022616067,0.00042635898,0.00038816183,0.00082669477,0.0008633673,0.0137976,0.007989113,0.0003416111,0.0000065290055],"category_scores_gemma":[0.0011952975,0.00030978737,0.00008595573,0.00033897016,0.00011374529,0.0034535355,0.010015551,0.0024167227,0.000020972175],"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":[0.000023366632,0.00006896628,0.00018785294,0.000038063914,0.00020780094,0.0000073441424,0.0008943094,0.0008866851,0.0000010161073,0.0105002,0.5535659,0.4336185],"study_design_scores_gemma":[0.00042933796,0.00015983885,0.00004636887,0.00078096165,0.00006812476,0.000023528677,0.00217046,0.019561056,0.000001599075,0.010670011,0.9656666,0.00042211922],"about_ca_topic_score_codex":0.000026260335,"about_ca_topic_score_gemma":0.000011612202,"teacher_disagreement_score":0.43319637,"about_ca_system_score_codex":0.0002558911,"about_ca_system_score_gemma":0.00038849592,"threshold_uncertainty_score":0.99993545},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.049018078,0.0004449695,0.07548285,0.04728276,0.0023122402,0.003452975,0.00035225006,0.00006885009,0.821585],"genre_scores_gemma":[0.99577206,0.000041575822,0.0005200066,0.0018549389,0.000014653212,0.00002404081,0.000009423962,0.0000023439973,0.001760953],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980541,0.00052547164,0.0003815112,0.00042025326,0.00050257234,0.0001161025],"domain_scores_gemma":[0.996934,0.0007838681,0.00016382562,0.002005373,0.00008131842,0.000031624793],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.008936161,0.0000737896,0.00015044866,0.000045089426,0.00019678038,0.0009941686,0.005585962,0.000024711993,0.00049398694],"category_scores_gemma":[0.0019120818,0.000032030235,0.0000130118115,0.00028469734,0.0003896581,0.0012688313,0.005540699,0.00010494761,0.00011225829],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00015834808,0.000081270824,0.0028534555,0.0000033321833,0.000011419437,0.0000045260945,0.004233819,0.0000010298398,0.000003871146,0.48397094,0.09841421,0.41026378],"study_design_scores_gemma":[0.0013763377,0.00004496444,0.19725382,0.000028024122,0.000016373644,0.000001148406,0.0069712787,0.0056231907,0.000016536242,0.21171728,0.5768039,0.00014710765],"about_ca_topic_score_codex":0.0031348728,"about_ca_topic_score_gemma":0.047113586,"teacher_disagreement_score":0.946754,"about_ca_system_score_codex":0.000007240135,"about_ca_system_score_gemma":0.00006624896,"threshold_uncertainty_score":0.9997943},"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":"commentary","genre_gemma":"commentary","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.115783244,0.032461192,0.000078625315,0.8367786,0.0011956219,0.0013638553,0.00961035,0.000014382698,0.0027141536],"genre_scores_gemma":[0.34056666,0.07953026,0.0010084906,0.5322894,0.006547614,0.000081860904,0.0034680753,0.00013528402,0.036372326],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99704486,0.000060777846,0.00061617437,0.0006784148,0.0013838516,0.00021590487],"domain_scores_gemma":[0.990978,0.0055565457,0.0005358191,0.0018698565,0.0010064507,0.000053322325],"candidate_categories":["research_integrity"],"consensus_categories":["research_integrity"],"category_scores_codex":[0.0015841002,0.00015135744,0.00088266266,0.00022835619,0.00003047919,0.000030185918,0.0025752399,0.0013597137,0.0005599358],"category_scores_gemma":[0.004315619,0.000096785305,0.00016992971,0.00027674425,0.0006335538,0.000023917079,0.002943556,0.0038646967,0.0000023707385],"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":[0.003005635,0.00012429072,0.34897187,0.002625975,0.046359453,0.00029407424,0.00037199046,0.00027435468,0.004444173,0.00015743569,0.58529294,0.008077801],"study_design_scores_gemma":[0.008956296,0.00014421451,0.32754222,0.0017580319,0.011260742,0.000011133001,0.000022137086,0.0070700934,0.0006519015,0.0019424265,0.64013696,0.00050383306],"about_ca_topic_score_codex":0.00006214584,"about_ca_topic_score_gemma":0.0005037387,"teacher_disagreement_score":0.30448917,"about_ca_system_score_codex":0.000097689524,"about_ca_system_score_gemma":0.00032939328,"threshold_uncertainty_score":0.99993676},"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":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000117711046,0.9928448,0.0036785426,0.00079843495,0.00058447954,0.0013208796,0.000091956455,0.000041841882,0.00052135385],"genre_scores_gemma":[0.00023342042,0.99748737,0.0009358799,0.0000067194787,0.001021196,0.000056021407,0.0000883333,0.000018903553,0.00015213835],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9921601,0.0018959494,0.001406646,0.0017364906,0.0021868711,0.0006139798],"domain_scores_gemma":[0.9834166,0.006403439,0.001098513,0.008442394,0.00029190542,0.00034710526],"candidate_categories":["metaresearch","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","open_science"],"category_scores_codex":[0.059169494,0.0002734746,0.0011239648,0.00027468838,0.00019529337,0.001073978,0.018869897,0.0004513591,0.00002050403],"category_scores_gemma":[0.043431103,0.0001362671,0.00023766089,0.0010226361,0.00075535546,0.006347005,0.017664535,0.0024968984,0.000024193285],"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":[0.0001275258,0.0000804025,0.000112893336,0.002162084,0.00021375825,0.000010068491,0.0000122224155,9.380214e-10,8.7824066e-7,0.0043273885,0.011734058,0.9812187],"study_design_scores_gemma":[0.0018840665,0.00018433665,0.00009128244,0.006596797,0.00009508043,0.0000075908442,0.000018559567,0.00022445495,0.0000011049615,0.000218648,0.99051625,0.00016182162],"about_ca_topic_score_codex":0.00028602584,"about_ca_topic_score_gemma":0.000008387251,"teacher_disagreement_score":0.98105687,"about_ca_system_score_codex":0.00028456902,"about_ca_system_score_gemma":0.0028862306,"threshold_uncertainty_score":0.999963},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8815136,0.000017461452,0.047739264,0.0023087726,0.00017095708,0.00050338573,0.0005072235,0.0003036967,0.066935636],"genre_scores_gemma":[0.977531,0.00005952621,0.002297869,0.00015191954,0.000019610727,1.8107539e-8,0.00006840317,0.000009599255,0.019862024],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9982507,0.000063059575,0.0002725838,0.00051180506,0.0004928033,0.00040901473],"domain_scores_gemma":[0.9980189,0.0001460722,0.000565633,0.0008056969,0.0001523191,0.00031133462],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.00010592218,0.00022364628,0.00041591958,0.00049982144,0.00031795382,0.00062146736,0.010458816,0.00007204298,0.00041713886],"category_scores_gemma":[0.000017982466,0.00022301196,0.00019715606,0.0015299753,0.00045314265,0.09009899,0.007643431,0.000111074325,0.00035878186],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005678627,0.00006847301,0.94867206,0.000033193763,0.00004115995,0.000056404675,0.00019919053,0.000034216766,1.08163526e-7,0.0027892946,0.011536304,0.036512833],"study_design_scores_gemma":[0.00085020514,0.0001444872,0.8594463,0.00007104382,0.000005733464,0.0000068708932,0.00012660876,0.0003110051,0.000020516703,0.00023542998,0.13848467,0.00029713166],"about_ca_topic_score_codex":0.00017864942,"about_ca_topic_score_gemma":0.000105106075,"teacher_disagreement_score":0.12694836,"about_ca_system_score_codex":0.000043801934,"about_ca_system_score_gemma":0.0005034556,"threshold_uncertainty_score":0.9948951},"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":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00012892106,0.0028425609,0.000015852042,0.97460127,0.00066743593,0.0014308622,0.0004302191,0.00004747871,0.019835409],"genre_scores_gemma":[0.0004885313,0.023012437,0.0045785992,0.8905301,0.042708706,0.000104638515,0.001170715,0.00018772899,0.03721856],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99528104,0.00042600223,0.0010056907,0.0015281277,0.0012896418,0.00046952526],"domain_scores_gemma":[0.95002186,0.041152336,0.0004574264,0.0077954954,0.00028895863,0.000283908],"candidate_categories":["metaresearch","research_integrity"],"consensus_categories":["metaresearch","research_integrity"],"category_scores_codex":[0.037198674,0.00024832564,0.0018296167,0.00016550583,0.00012934294,0.00019583142,0.0024363047,0.0016142682,0.00030263228],"category_scores_gemma":[0.2312334,0.00019219196,0.00011441616,0.000030735726,0.00023031331,0.000069555026,0.0035016355,0.0060095866,0.0004671973],"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":[0.00017508474,0.00001541809,0.00008978684,0.0014615031,0.00025313158,0.0006915292,0.00007319478,1.6013157e-8,0.00005875848,0.00036183547,0.9885554,0.00826437],"study_design_scores_gemma":[0.0012415134,0.00019778643,0.0008390298,0.0017696321,0.00021305008,0.00002584468,0.0000071731674,0.0000060708066,0.000013525637,0.010797335,0.9847026,0.00018645474],"about_ca_topic_score_codex":0.00009406997,"about_ca_topic_score_gemma":0.00013520912,"teacher_disagreement_score":0.19403473,"about_ca_system_score_codex":0.000040441464,"about_ca_system_score_gemma":0.0002851666,"threshold_uncertainty_score":0.99968183},"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":"commentary","genre_gemma":"methods","domain_codex":null,"domain_gemma":"reproducibility","model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0011298965,0.00048383622,0.00040894802,0.99293935,0.00017204265,0.00010772019,0.00008449482,0.00018367569,0.004490059],"genre_scores_gemma":[0.022858536,0.00038783674,0.9765785,0.000115274284,0.000008220468,0.0000077769655,0.000011731381,0.0000053633457,0.000026774143],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9991809,0.00009074721,0.00018751844,0.0002390647,0.00017335555,0.0001283889],"domain_scores_gemma":[0.08847716,0.00032683686,0.00030460538,0.9107986,0.0000746956,0.000018106795],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.00067964185,0.000069408874,0.000099474,0.00003696942,0.00080732047,0.00018503121,0.99610245,0.000052138985,0.0000023964678],"category_scores_gemma":[0.98348284,0.000052948348,0.000031964057,0.00013387701,0.0006775819,0.0010266346,0.99957126,0.00024320431,0.000022619988],"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":[3.653538e-7,0.000028365965,0.0010216057,0.0000025536453,0.0000100963625,9.1181725e-8,0.000019584577,1.9763849e-7,0.00018622538,0.008927578,0.9889661,0.0008372196],"study_design_scores_gemma":[0.00009670857,0.000004048994,0.006489985,0.00003060067,0.0000057561233,0.0000023238724,0.0000054675315,0.008557901,0.0009605169,0.96095353,0.022834903,0.000058232894],"about_ca_topic_score_codex":0.000143835,"about_ca_topic_score_gemma":0.00008736785,"teacher_disagreement_score":0.9928241,"about_ca_system_score_codex":0.000019058016,"about_ca_system_score_gemma":0.000056864257,"threshold_uncertainty_score":0.62093383},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9441911,0.000078633784,0.000041843214,0.016625863,0.000065751214,0.0005383333,0.00029795518,0.000015932026,0.038144562],"genre_scores_gemma":[0.99557275,0.00053562084,0.000020185922,0.0007234915,0.00018124771,0.00006107263,0.000022449054,0.00001026593,0.0028729003],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9982849,0.00007572777,0.00023739174,0.00032307362,0.0006834826,0.0003954593],"domain_scores_gemma":[0.9989285,0.00020097735,0.00020542568,0.00052583945,0.000025695614,0.000113574446],"candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0012576616,0.00013071539,0.0001785632,0.000014514296,0.0019672762,0.0010454904,0.0017328166,0.00007081149,0.00016078557],"category_scores_gemma":[0.00041460586,0.00010113031,0.000020933197,0.00012129443,0.00029248264,0.0030576081,0.0006549243,0.000102518105,0.0000037765215],"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":[0.00009097217,0.00007834834,0.8222034,0.00006358605,0.000026186528,0.0000012714738,0.10960762,0.0000053954313,0.00003531312,0.050762706,0.0002498838,0.016875312],"study_design_scores_gemma":[0.0020840955,0.000054313892,0.8332725,0.00015624694,0.00006520759,0.0000013140966,0.048237003,0.0006012035,0.0004660671,0.003261034,0.111277774,0.00052321487],"about_ca_topic_score_codex":0.11656328,"about_ca_topic_score_gemma":0.5638405,"teacher_disagreement_score":0.44727725,"about_ca_system_score_codex":0.00022987403,"about_ca_system_score_gemma":0.00021704043,"threshold_uncertainty_score":0.99999154},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9393099,0.0031063217,0.0012210428,0.011602595,0.00032915495,0.0006118539,0.00009117406,0.00014206745,0.04358588],"genre_scores_gemma":[0.980637,0.00063515775,0.010511489,0.00034670814,0.00022727017,0.00006541047,0.000003612721,0.000024943418,0.007548414],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9963179,0.00025970166,0.00083022425,0.0008590729,0.00067973655,0.0010533127],"domain_scores_gemma":[0.96865183,0.0007729817,0.0011852074,0.0003510455,0.028627971,0.00041097807],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.002313647,0.00060162455,0.000777956,0.00026732706,0.0012559254,0.023067795,0.004500397,0.00044378042,0.00004811014],"category_scores_gemma":[0.024846448,0.0004211144,0.00034204233,0.0009518903,0.00851211,0.06249738,0.0019550666,0.0005040913,0.000009014833],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000054513683,0.00055734545,0.07193567,0.0010604695,0.00016185624,0.000008223227,0.19637385,0.00007019412,0.057802696,0.64520836,0.009823082,0.016943734],"study_design_scores_gemma":[0.0010313855,0.0010272168,0.30338633,0.003747918,0.00016645473,0.0003666863,0.06710169,0.0043498306,0.34990376,0.2133575,0.053922717,0.0016385189],"about_ca_topic_score_codex":0.006649732,"about_ca_topic_score_gemma":0.0002960806,"teacher_disagreement_score":0.43185085,"about_ca_system_score_codex":0.00009554508,"about_ca_system_score_gemma":0.0019255506,"threshold_uncertainty_score":0.9999651},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3750546,0.00044967656,0.0000103219445,0.015855357,0.0007910282,0.00010850826,0.000074867596,0.000019066225,0.6076366],"genre_scores_gemma":[0.94428915,0.000029930121,0.00016122236,0.0012801862,0.0015223038,0.0000025306736,0.00000802956,0.000024549105,0.0526821],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982655,0.00003089533,0.00059042865,0.00020170974,0.00063481665,0.0002766336],"domain_scores_gemma":[0.99477214,0.0000782264,0.0011671608,0.00047017125,0.003235696,0.00027662693],"candidate_categories":["sts","scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.0021991918,0.00015933102,0.0002935561,0.00024122665,0.0020179267,0.68961936,0.003734816,0.00007787638,0.0005830227],"category_scores_gemma":[0.0054603433,0.00013051249,0.00009733039,0.00003253282,0.0003247701,0.4705118,0.0011390772,0.0010252149,0.000021283118],"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":[0.000043018696,0.00008111413,0.014482107,0.000023412811,0.000077290715,0.000025480957,0.005106639,1.4494721e-7,0.000011426774,0.88791615,0.05069546,0.041537754],"study_design_scores_gemma":[0.00071205274,0.000110483175,0.025382647,0.00012471684,0.000015152986,0.00007221103,0.0036250737,0.0000017536361,0.000007388017,0.04486211,0.9249064,0.00018002822],"about_ca_topic_score_codex":0.00016307525,"about_ca_topic_score_gemma":0.00015425809,"teacher_disagreement_score":0.8742109,"about_ca_system_score_codex":0.000057486173,"about_ca_system_score_gemma":0.00015516057,"threshold_uncertainty_score":0.9992813},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.93443,0.00013331855,0.0000075875355,0.056252412,0.000056934172,0.0012564035,0.000056452704,0.000002829355,0.0078040734],"genre_scores_gemma":[0.99475104,0.0046403026,0.00009465235,0.000214097,0.000155226,0.000033623965,0.0000042964507,0.00002588611,0.00008089784],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9570784,0.014581433,0.0036305257,0.0006590581,0.022475388,0.0015751938],"domain_scores_gemma":[0.7907996,0.17443582,0.0016293057,0.0028757814,0.028913306,0.001346191],"candidate_categories":["metaresearch","sts","research_integrity"],"consensus_categories":["metaresearch","sts","research_integrity"],"category_scores_codex":[0.24327019,0.00024560414,0.0012463913,0.002841258,0.0015247079,0.0008609921,0.005201896,0.0013678204,0.000091240545],"category_scores_gemma":[0.36117914,0.00016022682,0.00014417392,0.0018894837,0.013336905,0.001809416,0.003246382,0.045343067,5.549629e-7],"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":[0.010155075,0.0014842087,0.9210206,0.00602724,0.0001766135,0.00010528822,0.0057033724,0.00011413785,0.00088051055,0.04799014,0.0012354089,0.005107361],"study_design_scores_gemma":[0.00445761,0.0041470267,0.92412734,0.0040465323,0.0000141223545,0.0000076190045,0.003709452,0.00046816474,0.0036693616,0.055081077,0.00015288345,0.00011877963],"about_ca_topic_score_codex":0.4163566,"about_ca_topic_score_gemma":0.4236261,"teacher_disagreement_score":0.16627881,"about_ca_system_score_codex":0.0014798163,"about_ca_system_score_gemma":0.034463726,"threshold_uncertainty_score":0.9999286},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0027888883,0.000013325257,0.9571408,0.006557424,0.0005656938,0.0003435592,0.0000043541204,0.000016962838,0.032568995],"genre_scores_gemma":[0.8547153,0.000059422553,0.14483507,0.00021676967,0.000056864425,0.0000038699027,0.000002608324,0.0000028024585,0.00010729546],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9973874,0.000039428654,0.00072690984,0.00015922917,0.001468859,0.00021815086],"domain_scores_gemma":[0.99366057,0.000048323527,0.0021243184,0.003628525,0.0004347557,0.000103491926],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.014029949,0.00007537395,0.00012253836,0.00059193146,0.0008094053,0.0065349406,0.024012676,0.000010209557,0.0000013094368],"category_scores_gemma":[0.0011549682,0.000050807907,0.000026367572,0.00052751246,0.00013273911,0.11337294,0.013156379,0.00013552417,0.000033595275],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000019341476,0.000018420767,0.0008884902,0.00029274976,0.0000857898,0.000020300824,0.0006875512,0.0024805686,0.000033122684,0.858923,0.004789569,0.13176109],"study_design_scores_gemma":[0.0021775106,0.0005838192,0.10990025,0.0020780645,0.00014084236,0.00015248454,0.014840038,0.42922646,0.0015165911,0.0030329505,0.43569425,0.00065672817],"about_ca_topic_score_codex":0.000029293355,"about_ca_topic_score_gemma":0.0000035427427,"teacher_disagreement_score":0.85589004,"about_ca_system_score_codex":0.00009117433,"about_ca_system_score_gemma":0.00008405392,"threshold_uncertainty_score":0.994825},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","bibliometrics","open_science"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6555457,0.000052045176,0.33452216,0.008273919,0.0001529257,0.0004822407,0.0003520166,0.00037863723,0.00024031296],"genre_scores_gemma":[0.9880704,0.00009060259,0.010754162,0.00032914715,0.000060861435,0.0000017947126,0.00039147155,0.000023632976,0.0002779173],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9956287,0.00044036444,0.00036263623,0.0021463514,0.0007373182,0.00068460783],"domain_scores_gemma":[0.9728469,0.00018558449,0.00025653664,0.02640521,0.000051702085,0.000254025],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.0019970292,0.00034994064,0.0005227513,0.00075922237,0.00019535741,0.0017204478,0.12930904,0.00023227888,0.0000024242815],"category_scores_gemma":[0.0027001365,0.00038009678,0.000043578824,0.003311642,0.0002969114,0.008286307,0.51675695,0.00062678347,0.000029081013],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021140188,0.00013145251,0.8810581,0.000021772466,0.00015582574,0.00010946993,0.000048526454,0.0000064493124,0.000014162819,0.00034970843,0.000025675407,0.1180577],"study_design_scores_gemma":[0.0017054492,0.00017340314,0.9644012,0.00016041318,0.00023096381,0.000029677021,0.00022124953,0.0046841023,0.0001426446,0.005688576,0.021716889,0.0008454622],"about_ca_topic_score_codex":0.00020260357,"about_ca_topic_score_gemma":0.017653333,"teacher_disagreement_score":0.3874479,"about_ca_system_score_codex":0.0003728406,"about_ca_system_score_gemma":0.00017131289,"threshold_uncertainty_score":0.9998651},"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":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4493245,0.0036907375,0.025529563,0.46696708,0.001567821,0.0022673556,0.000022648257,0.00003183809,0.050598484],"genre_scores_gemma":[0.6962835,0.0023997896,0.29903677,0.0011699974,0.0006182748,0.000012370453,0.0000063973966,0.000030597566,0.00044229033],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.9936638,0.001420403,0.0012016651,0.0008437377,0.002359714,0.0005106999],"domain_scores_gemma":[0.98027885,0.0014530432,0.011662192,0.0043293647,0.0018651681,0.0004113919],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","scholarly_communication","open_science"],"category_scores_codex":[0.13979344,0.00016412642,0.00041954173,0.0007587454,0.0013287866,0.9319236,0.14772896,0.00009163256,0.000022530514],"category_scores_gemma":[0.31693116,0.00013635366,0.000026545982,0.00058195763,0.000059403304,0.980358,0.18247744,0.0029767356,0.0000020074679],"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":[0.00013658719,0.00033285943,0.38457733,0.00013028657,0.00024301025,0.0009460113,0.0011820432,0.00003460334,0.0017555946,0.013146913,0.00900042,0.5885143],"study_design_scores_gemma":[0.0057494333,0.00049152476,0.84895015,0.0031444624,0.000090699614,0.0008939551,0.0014599079,0.041851208,0.0002065552,0.024658673,0.07184941,0.00065401423],"about_ca_topic_score_codex":0.0021001562,"about_ca_topic_score_gemma":0.00039517425,"teacher_disagreement_score":0.58786035,"about_ca_system_score_codex":0.00011687829,"about_ca_system_score_gemma":0.0010361179,"threshold_uncertainty_score":0.99997133},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.72448945,0.0014559218,0.0016463716,0.10656251,0.015171819,0.00034370154,0.00015675106,0.000020223824,0.15015328],"genre_scores_gemma":[0.98833615,0.00028795341,0.004801818,0.00071664277,0.0048119547,0.0000037220946,0.000005366537,0.000041975567,0.0009943914],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9758969,0.0011473504,0.0025036442,0.001025041,0.018329203,0.0010978502],"domain_scores_gemma":[0.91402906,0.006315053,0.009006974,0.0027882613,0.06374548,0.0041151536],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","scholarly_communication","open_science"],"category_scores_codex":[0.1650671,0.00034843606,0.0008807163,0.05186339,0.004778166,0.9670092,0.076787025,0.00053375016,0.0023121403],"category_scores_gemma":[0.52051526,0.00026000585,0.0005895458,0.01954708,0.00038430543,0.8922455,0.008726421,0.008395823,0.00009304696],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0014729294,0.0007944292,0.23279968,0.000004500322,0.00046294605,0.000618578,0.0011203303,0.00025096408,0.00039666027,0.08729751,0.5100968,0.16468467],"study_design_scores_gemma":[0.0076963706,0.0012969048,0.5992395,0.00007369957,0.000051031257,0.0010628721,0.0070621935,0.0006764107,0.00012834236,0.27428055,0.10787681,0.0005553115],"about_ca_topic_score_codex":0.0009909322,"about_ca_topic_score_gemma":0.00017492325,"teacher_disagreement_score":0.40221998,"about_ca_system_score_codex":0.0025866453,"about_ca_system_score_gemma":0.0057737366,"threshold_uncertainty_score":0.9999852},"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":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000032451695,0.96781725,0.0033451826,0.0021444098,0.0019752132,0.0029281697,0.021639926,0.000009228817,0.00010816888],"genre_scores_gemma":[0.000062518026,0.9313646,0.00057813135,0.00068613625,0.01269892,0.00026242793,0.05385471,0.00010280671,0.00038976083],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9971984,0.000057761743,0.00070806756,0.001491385,0.00017727049,0.00036713504],"domain_scores_gemma":[0.9878006,0.0016933657,0.0006737374,0.009626956,0.000087238084,0.00011808895],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0018091088,0.00046722344,0.0009971961,0.000059316764,0.0004977828,0.00011070753,0.00690449,0.00033350658,0.000019906096],"category_scores_gemma":[0.0031994425,0.00024152148,0.00017174012,0.000060561273,0.00034801036,0.000006586545,0.003986319,0.00021403591,0.00001028541],"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":[0.00007590446,0.000028416745,8.6867186e-7,0.0007475359,0.00033571175,0.000017638784,0.00002037726,2.4686614e-7,0.0001203221,0.000056499062,0.3185984,0.6799981],"study_design_scores_gemma":[0.0009840489,0.00045465113,0.0000033935567,0.0037636443,0.0019467927,0.000116188385,0.00004639356,0.00021145708,0.000012647584,0.00024533315,0.9919298,0.00028562386],"about_ca_topic_score_codex":0.000794625,"about_ca_topic_score_gemma":0.00047494337,"teacher_disagreement_score":0.6797125,"about_ca_system_score_codex":0.000045891546,"about_ca_system_score_gemma":0.0003244445,"threshold_uncertainty_score":0.99846864},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.495826,0.00086165895,0.010336102,0.48162034,0.008093616,0.00043200364,0.0000738792,0.0005622214,0.0021941569],"genre_scores_gemma":[0.98522574,0.00007913823,0.01228542,0.00022125585,0.00007135484,0.000009837156,0.0000026554671,0.0000049596824,0.0020996425],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99601644,0.000036501704,0.0005121289,0.0021803044,0.00096958986,0.00028500854],"domain_scores_gemma":[0.986064,0.00054017507,0.00051949534,0.012475417,0.00037780765,0.000023106171],"candidate_categories":["metaresearch","open_science"],"consensus_categories":[],"category_scores_codex":[0.011038692,0.00013188364,0.00024210234,0.0001553209,0.00078288926,0.0004653905,0.008209341,0.00056057354,0.000021706048],"category_scores_gemma":[0.061301094,0.000078296456,0.00012529996,0.0005063119,0.0011671843,0.00016775829,0.0057989866,0.00063797383,0.000096580494],"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":[0.000025947622,0.000046125722,0.0015448928,0.0000029403673,0.000039808416,0.0000020519074,0.00007360833,0.000011483503,0.006108864,0.052732516,0.25137916,0.6880326],"study_design_scores_gemma":[0.000084417596,0.000045140045,0.010080483,0.000012924945,0.000005248515,0.0000027682506,0.00082107546,0.00020552208,0.22876956,0.098686986,0.6611718,0.00011409235],"about_ca_topic_score_codex":0.000074027084,"about_ca_topic_score_gemma":0.00016937395,"teacher_disagreement_score":0.68791854,"about_ca_system_score_codex":0.000057105644,"about_ca_system_score_gemma":0.00004306081,"threshold_uncertainty_score":0.99715674},"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":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00007245887,0.9280457,0.062224567,0.000060584618,0.0033907918,0.0049302,0.00019049586,0.00001614678,0.0010690659],"genre_scores_gemma":[0.000088856905,0.9730482,0.01127193,0.00014102901,0.00024043098,0.0004581346,0.000058073423,0.000086884786,0.0146064665],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9942583,0.00032983877,0.0005769291,0.0030531646,0.00068822404,0.0010935742],"domain_scores_gemma":[0.9978906,0.000033616874,0.0003425628,0.0013142395,0.00008726322,0.00033175125],"candidate_categories":["metaepi_narrow","open_science"],"consensus_categories":[],"category_scores_codex":[0.0015049543,0.00057449844,0.0009871178,0.0004547248,0.00044112428,0.00086582673,0.008528531,0.00020288333,7.4735135e-7],"category_scores_gemma":[0.00022372293,0.00049100997,0.000301457,0.0015504344,0.0016094135,0.00010246616,0.0038499525,0.00033096786,0.000003134981],"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":[0.00019506215,0.00066434144,0.00007214537,0.0063826134,0.000047231544,0.00001571821,0.00019396198,0.0004475932,0.016795263,0.00015428555,0.26611537,0.7089164],"study_design_scores_gemma":[0.00035272946,0.00025147156,0.000002677084,0.00038092674,0.000038447364,0.000008635275,0.000035598572,0.0012002744,0.0035606704,0.000025138504,0.9935765,0.0005669181],"about_ca_topic_score_codex":0.00003179743,"about_ca_topic_score_gemma":0.000018951316,"teacher_disagreement_score":0.72746116,"about_ca_system_score_codex":0.00013940276,"about_ca_system_score_gemma":0.0028936611,"threshold_uncertainty_score":0.99975413},"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":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000042761567,0.000036920417,0.00033775595,0.0015213643,0.0004878107,0.00053137203,0.004480064,0.000009103432,0.9925913],"genre_scores_gemma":[0.0017536137,0.0000074979125,0.00019227425,0.00013944144,0.00023933047,0.000017003409,0.00065383717,0.000024680041,0.9969723],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99587345,0.00025259762,0.0009320966,0.00085214333,0.001780787,0.00030893766],"domain_scores_gemma":[0.98980576,0.0013888744,0.0017995121,0.0067063505,0.00022054961,0.00007894388],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.009052572,0.00024557786,0.00064202637,0.00013385083,0.001117386,0.0023803262,0.035884455,0.00018521912,0.0001404625],"category_scores_gemma":[0.0014436287,0.00015654636,0.00013679975,0.000040372277,0.0012638986,0.00029880283,0.023040615,0.00031275392,0.00022473744],"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":[0.000016499838,0.0000068547174,7.389218e-7,0.00000450185,0.000055335735,0.000004433602,0.00030618985,2.046688e-8,1.1713708e-7,0.19139907,0.4944133,0.31379297],"study_design_scores_gemma":[0.00013266805,0.000025701698,0.000025708086,0.0000209553,0.00006052331,6.845275e-7,0.0010884104,0.000063838226,0.0000010185338,0.20217094,0.79625237,0.00015719504],"about_ca_topic_score_codex":0.00026661798,"about_ca_topic_score_gemma":0.00553647,"teacher_disagreement_score":0.31363577,"about_ca_system_score_codex":0.0001016116,"about_ca_system_score_gemma":0.00089596125,"threshold_uncertainty_score":0.9986553},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96721405,0.0000018435338,0.031666577,0.00014610196,0.000017818847,0.00043485084,0.000005407892,0.000013108581,0.0005002314],"genre_scores_gemma":[0.9992802,7.697605e-7,0.0006444073,0.000006315426,0.00003245316,0.000012994395,0.0000010831084,0.000007875936,0.000013901553],"study_design_codex":"observational","study_design_gemma":"qualitative","domain_scores_codex":[0.9978277,0.00016571718,0.00037132128,0.0003587165,0.00094658334,0.0003299293],"domain_scores_gemma":[0.9984764,0.00019617479,0.00028400868,0.0004961622,0.00045727956,0.00009000579],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002881206,0.000097280354,0.00023178803,0.00074365933,0.0005672697,0.0004536173,0.001341408,0.000009673142,0.00012861924],"category_scores_gemma":[0.00003649474,0.000074935975,0.0000300079,0.0024379669,0.0011704317,0.00085516536,0.0016804052,0.00013520113,3.5891742e-7],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006484748,0.0030802498,0.94370943,0.00003387287,0.00008888314,0.000014531839,0.015173807,0.000026970743,0.011692062,0.02102828,0.00006047397,0.0050265607],"study_design_scores_gemma":[0.0026663686,0.004180921,0.08483757,0.00087415334,0.00010988908,0.000011658533,0.74957347,0.013003196,0.10700311,0.037282575,0.000019638108,0.00043747746],"about_ca_topic_score_codex":0.005020816,"about_ca_topic_score_gemma":0.0012619218,"teacher_disagreement_score":0.8588719,"about_ca_system_score_codex":0.000115033916,"about_ca_system_score_gemma":0.00013726972,"threshold_uncertainty_score":0.7590005},"labels":[{"model":"gemma","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"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":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02358038,0.09117135,0.00032235566,0.79096276,0.0029130022,0.0016407042,0.0012979838,0.00002173031,0.08808971],"genre_scores_gemma":[0.11867617,0.011172104,0.00006394897,0.819078,0.04645118,0.000017150182,0.0013733448,0.00020593309,0.002962157],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9972454,0.00020420272,0.00066372193,0.00063479366,0.00043447668,0.0008174225],"domain_scores_gemma":[0.9966932,0.00007178514,0.00033542686,0.0015179069,0.00060185447,0.00077981915],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0010267384,0.00040125317,0.002111191,0.0006102988,0.00015334258,0.00029600676,0.0028231787,0.0008799186,0.00013711877],"category_scores_gemma":[0.00026381554,0.00035210175,0.0005680998,0.00023407882,0.00036330905,0.0005321246,0.00044422253,0.0021648565,0.00002754969],"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":[0.00006882326,9.510299e-7,0.005747513,0.000115585965,0.006264789,0.005138366,0.00003183054,0.000002944231,0.000005862674,0.000007091283,0.9808345,0.0017817486],"study_design_scores_gemma":[0.0010884794,0.00030765592,0.024348503,0.00013673671,0.0027329775,0.0037301923,0.000013600323,0.000003530823,0.000023795114,0.00025811308,0.96704143,0.00031500068],"about_ca_topic_score_codex":0.0057785055,"about_ca_topic_score_gemma":0.0040153204,"teacher_disagreement_score":0.09509579,"about_ca_system_score_codex":0.00014012566,"about_ca_system_score_gemma":0.0052233804,"threshold_uncertainty_score":0.99989307},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99655616,0.0015700989,0.00085578655,0.00030706852,0.00010741455,0.00026799904,0.000018931829,0.0000013085803,0.00031524527],"genre_scores_gemma":[0.97819865,0.00021788859,0.019951051,0.00074309064,0.0005727378,0.000013608925,0.00026568113,0.000012274173,0.000025025036],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.99883795,0.000045628676,0.00051505514,0.00034425716,0.000088351306,0.0001687372],"domain_scores_gemma":[0.99907994,0.00007415494,0.000101316764,0.00067057414,0.00003924688,0.00003474061],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0018776123,0.00008141044,0.00014794573,0.00003890456,0.00003987033,0.000004362551,0.000699739,0.00006551327,0.000004894948],"category_scores_gemma":[0.00040969538,0.000058932845,0.00002064281,0.0000564743,0.0001848273,5.630089e-7,0.00041749136,0.000058794656,8.692442e-7],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000953178,0.00067361764,0.40765062,0.00042776993,0.00021711653,0.000031547224,0.009075265,0.00017318035,0.41098145,0.00025888978,0.03271056,0.1368468],"study_design_scores_gemma":[0.005489359,0.0010362663,0.26880246,0.00036088907,0.0000867966,0.00003229547,0.0061501223,0.003382004,0.015606348,0.0005809218,0.69801295,0.00045956922],"about_ca_topic_score_codex":0.000007992849,"about_ca_topic_score_gemma":0.00045216148,"teacher_disagreement_score":0.6653024,"about_ca_system_score_codex":0.000007608592,"about_ca_system_score_gemma":0.00016365285,"threshold_uncertainty_score":0.24032117},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"protocol","genre_consensus":"protocol","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.013336882,7.366907e-7,0.000024329796,0.004786267,0.0000055302557,0.97678787,0.000070406066,0.000052216255,0.004935789],"genre_scores_gemma":[0.07444551,3.4920131e-7,0.0026747906,0.00016778204,0.00026272182,0.92136574,0.000014352754,0.00010132452,0.00096741493],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"nonrandomized_trial","domain_scores_codex":[0.9832409,0.003526958,0.0016081447,0.002499397,0.006969893,0.0021546765],"domain_scores_gemma":[0.9528271,0.034966715,0.00025212072,0.0041572982,0.0071562715,0.00064052746],"candidate_categories":["metaresearch","metaepi_narrow","sts","research_integrity"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.101704344,0.00035341282,0.0010816165,0.0015393035,0.0007422533,0.0004278411,0.0048189145,0.00095328694,0.000859915],"category_scores_gemma":[0.093330994,0.0002745777,0.00015018853,0.0025136732,0.0036972333,0.0006232607,0.0052804477,0.011522962,0.00017806292],"study_design_candidate":"randomized_trial","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.3341305,0.0778822,0.12113768,0.014006684,0.0006547078,0.0005375994,0.057790022,0.000006402675,0.0085737575,0.35158405,0.03213105,0.0015653496],"study_design_scores_gemma":[0.16470167,0.3306735,0.07193395,0.0633284,0.000037657086,0.000004075161,0.07635671,0.01607783,0.01290499,0.082277946,0.18031746,0.0013858221],"about_ca_topic_score_codex":0.00054271345,"about_ca_topic_score_gemma":0.004555881,"teacher_disagreement_score":0.2693061,"about_ca_system_score_codex":0.0006849791,"about_ca_system_score_gemma":0.0050397227,"threshold_uncertainty_score":0.9999706},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07155446,0.00068461325,0.001184478,0.006378169,0.29369903,0.0027473264,0.0004912295,0.0001129927,0.6231477],"genre_scores_gemma":[0.036449455,0.000026608435,0.0014735899,0.005557684,0.0061119227,0.000056603098,0.00009680291,0.00007518929,0.95015216],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99068624,0.000212793,0.0012795681,0.0019909437,0.0048468546,0.0009836215],"domain_scores_gemma":[0.9932364,0.00035005697,0.00078109774,0.0023315202,0.0022756166,0.0010252751],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.0063736383,0.00048685414,0.00085948384,0.00090515177,0.0015403801,0.011929617,0.030379336,0.00043679593,0.00077213417],"category_scores_gemma":[0.016286476,0.0004094163,0.000091560636,0.0023051777,0.00040346588,0.0009502039,0.01039504,0.0012815625,0.00059634366],"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":[0.0000076094666,0.000013956826,0.00063220883,0.0000030128977,0.00000927173,0.0000074317186,0.00075651205,0.000024487887,0.000047026228,0.000079047706,0.98052645,0.017892981],"study_design_scores_gemma":[0.0001348745,0.000125719,0.018967355,0.00018001796,0.00001881348,0.000012266226,0.0004274598,0.00017707034,0.00019282053,0.0023751846,0.9768028,0.0005855657],"about_ca_topic_score_codex":0.24286221,"about_ca_topic_score_gemma":0.6537419,"teacher_disagreement_score":0.41087967,"about_ca_system_score_codex":0.00095955736,"about_ca_system_score_gemma":0.009679837,"threshold_uncertainty_score":0.9998358},"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":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11289275,0.00003503983,0.0001560084,0.08618909,0.00036397914,0.00066949945,0.000002907619,0.00005132402,0.7996394],"genre_scores_gemma":[0.48603314,0.00003969787,0.007013837,0.0008542017,0.00009580404,0.0000067144256,0.000019199872,0.000018055913,0.50591934],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99861175,0.0002608299,0.00013071786,0.0004884002,0.00019468255,0.0003136367],"domain_scores_gemma":[0.99851125,0.0002915078,0.00023328412,0.0005248734,0.000088140674,0.00035092726],"candidate_categories":["sts","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.00072872295,0.00013224996,0.0001601894,0.00012557392,0.0013637178,0.011046079,0.015525095,0.000045181183,0.0002682171],"category_scores_gemma":[0.00015418985,0.000114945746,0.000025399726,0.0007026616,0.000099266836,0.01551044,0.042822048,0.00016439316,0.00009643585],"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":[0.000103948325,0.00012720634,0.18724203,0.000008708506,0.000055076936,0.000023718147,0.017104248,0.000030613508,0.000014524635,0.41362825,0.37594426,0.0057174317],"study_design_scores_gemma":[0.00061375607,0.00005567908,0.03168805,0.000027125507,0.00000828799,0.00001577624,0.011688536,0.00024451505,0.00010626658,0.002182498,0.9531537,0.00021577884],"about_ca_topic_score_codex":0.0037487785,"about_ca_topic_score_gemma":0.0015941189,"teacher_disagreement_score":0.5772095,"about_ca_system_score_codex":0.00018216447,"about_ca_system_score_gemma":0.0014822893,"threshold_uncertainty_score":0.99993634},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4803763,0.0037095414,0.0030652455,0.43174964,0.009758262,0.0022328957,0.000009810873,0.00040671244,0.06869159],"genre_scores_gemma":[0.98646975,0.00044701286,0.00019787405,0.004389865,0.0069115995,0.00008565325,0.000014773565,0.000018082623,0.0014653964],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9965275,0.0003873924,0.00059201807,0.00042649294,0.001709188,0.00035741724],"domain_scores_gemma":[0.9968806,0.001008123,0.0003086268,0.00048233045,0.0009118203,0.00040849583],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.004630015,0.00018302073,0.00032548016,0.00031108392,0.00035434662,0.000028605584,0.0005918871,0.0002250205,0.0033609867],"category_scores_gemma":[0.014056142,0.00014239362,0.00006670058,0.0013826257,0.0015786991,0.00030065668,0.000043516113,0.00020774707,0.00010811899],"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":[0.0009796361,0.0013426183,0.11781403,0.00053986325,0.00009314148,0.000031571406,0.12203845,0.0000374425,0.0010069945,0.06455932,0.5705924,0.12096451],"study_design_scores_gemma":[0.0048393225,0.005114344,0.058967076,0.028205127,0.00031563028,0.0000057914567,0.09055906,0.0041609583,0.0067026992,0.0052085924,0.7945506,0.0013708025],"about_ca_topic_score_codex":0.0016440307,"about_ca_topic_score_gemma":0.0032966787,"teacher_disagreement_score":0.50609344,"about_ca_system_score_codex":0.0002458562,"about_ca_system_score_gemma":0.001977165,"threshold_uncertainty_score":0.9975501},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014231549,0.033631075,0.00087713153,0.0885744,0.001276022,0.0049373764,0.00063038734,0.0008814581,0.8549606],"genre_scores_gemma":[0.87302226,0.11415392,0.00027249774,0.011721458,0.0005696068,0.000011974901,0.00009376905,0.000024812187,0.00012967578],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99836785,0.00070730963,0.00026920033,0.00015144932,0.00036001718,0.00014419622],"domain_scores_gemma":[0.9989871,0.00031129114,0.00021200623,0.000254263,0.00017567076,0.0000596648],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0013173539,0.000118872704,0.00022866146,0.00010485997,0.00043367414,0.00016365874,0.0003749423,0.000032284996,0.000030748328],"category_scores_gemma":[0.00378584,0.00009251004,0.00005338726,0.0002874458,0.00048972946,0.0024936984,0.0003191418,0.00016138278,0.000094165116],"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":[0.001257768,0.0010812711,0.0049480186,0.04573261,0.0004514274,0.00004915547,0.03401344,0.000017426479,0.000028287686,0.18682835,0.29734835,0.4282439],"study_design_scores_gemma":[0.00073084235,0.00083108136,0.001617817,0.012400158,0.00007252077,3.609869e-7,0.0013940072,0.00021965797,0.00014599983,0.0015466718,0.98069566,0.00034520935],"about_ca_topic_score_codex":0.00002020849,"about_ca_topic_score_gemma":0.0000061952137,"teacher_disagreement_score":0.85879076,"about_ca_system_score_codex":0.000030476842,"about_ca_system_score_gemma":0.00015933561,"threshold_uncertainty_score":0.45322776},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7965715,0.0010331711,0.0055306014,0.17937249,0.0011308735,0.0019019761,0.0006054888,0.00008381177,0.013770071],"genre_scores_gemma":[0.97899026,0.001455095,0.013626326,0.0030572594,0.0013293723,0.000014058015,0.0010513436,0.000013046601,0.00046326668],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99371487,0.0024070174,0.00075748906,0.00082882727,0.00066487474,0.0016269053],"domain_scores_gemma":[0.99655366,0.001321851,0.00032505326,0.0010160174,0.00023210443,0.0005513348],"candidate_categories":["sts","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.005283825,0.0003166039,0.00045381996,0.0006777779,0.0008172254,0.00023206961,0.008385644,0.00021321698,0.006742519],"category_scores_gemma":[0.0039440254,0.00019099507,0.00008925918,0.0010041025,0.0056386045,0.0016181178,0.006185744,0.0022541264,0.00040935678],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00095464225,0.00019338758,0.39685598,0.000012159345,0.00019064047,0.0005156489,0.0047336905,0.000021173997,0.00002420153,0.0111484695,0.0071853604,0.57816464],"study_design_scores_gemma":[0.0011798729,0.0030152865,0.19840857,0.000021009995,0.000010702759,0.0040235883,0.0020061592,0.010594392,0.000008880571,0.35556087,0.424846,0.00032465215],"about_ca_topic_score_codex":0.012807391,"about_ca_topic_score_gemma":0.16824289,"teacher_disagreement_score":0.57784,"about_ca_system_score_codex":0.000035265668,"about_ca_system_score_gemma":0.0013563947,"threshold_uncertainty_score":0.99706745},"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":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0074151726,0.9889322,0.002664497,0.00018416358,0.00014338312,0.00039782448,0.00015241567,0.0000036044087,0.00010675259],"genre_scores_gemma":[0.016441528,0.9715188,0.010982716,0.000015993175,0.00055749674,0.0000250864,0.00040203397,0.000037559428,0.000018802095],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978704,0.00016367156,0.00074520585,0.0008518028,0.00020056475,0.00016839679],"domain_scores_gemma":[0.9974741,0.00017991157,0.00055758987,0.0015945946,0.00014481027,0.00004902911],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00086913485,0.0002393032,0.00061465666,0.00012700354,0.00005365308,0.00001157387,0.0008657232,0.0002541161,0.0000072565667],"category_scores_gemma":[0.00072721817,0.0001777073,0.00003174242,0.00030327935,0.0007255386,0.0000033332255,0.000728879,0.0001907488,5.543489e-7],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000012791713,0.00002430913,0.0018107683,0.0016433074,0.00020389173,0.0000013661509,0.000203162,0.000028376531,0.0067073437,0.00014175138,0.00027797287,0.98894495],"study_design_scores_gemma":[0.0036630498,0.0011495391,0.03409641,0.020968225,0.005020495,0.00024286265,0.0013021664,0.016465025,0.0009893809,0.009529259,0.9052027,0.001370874],"about_ca_topic_score_codex":0.00029945115,"about_ca_topic_score_gemma":0.0002199727,"teacher_disagreement_score":0.9875741,"about_ca_system_score_codex":0.00006917489,"about_ca_system_score_gemma":0.00039754465,"threshold_uncertainty_score":0.72466934},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96411216,0.000032946868,0.005249992,0.0045923586,0.00006541063,0.0009896748,0.0064854193,0.000052447558,0.01841958],"genre_scores_gemma":[0.998037,0.000068165304,0.0001503687,0.0004469167,0.0000079467645,0.000014705686,0.00035026713,0.0000031330935,0.00092152465],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99899834,0.00003312877,0.00009795577,0.0003333375,0.00020545076,0.00033180497],"domain_scores_gemma":[0.999472,0.000048617698,0.000092692615,0.0002086472,0.00004731852,0.0001306779],"candidate_categories":["sts","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00032933982,0.000111575246,0.00014619589,0.000024484014,0.0030240696,0.00040138018,0.0006520763,0.000064153486,0.033503298],"category_scores_gemma":[0.000026872489,0.000103860344,0.000052665295,0.00026777294,0.00021308518,0.0011640916,0.005386208,0.000105054685,0.00055967894],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00042793388,0.0005398185,0.8540031,0.000029487972,0.000083420644,0.000002914764,0.0005538145,0.000052123687,0.0006059766,0.017351529,0.12596534,0.00038457735],"study_design_scores_gemma":[0.0026043258,0.00018968793,0.87513316,0.000030126768,0.00013572576,0.000010404188,0.0027833052,0.000048718175,0.00042996343,0.00067383714,0.11755721,0.000403538],"about_ca_topic_score_codex":0.0010546437,"about_ca_topic_score_gemma":0.0004900156,"teacher_disagreement_score":0.033924803,"about_ca_system_score_codex":0.00034205464,"about_ca_system_score_gemma":0.000016919927,"threshold_uncertainty_score":0.99827385},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.71493196,0.00016821439,0.23198242,0.039932307,0.0026528016,0.0017177156,0.000080426355,0.000097998,0.008436141],"genre_scores_gemma":[0.9979015,0.000012124833,0.0014862709,0.0005038815,0.000038953574,0.0000026575617,0.000008080502,0.0000025352851,0.000044019835],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99878424,0.0000027501849,0.00040387662,0.000093791336,0.0005819952,0.00013333598],"domain_scores_gemma":[0.9958691,0.00004990695,0.0014474123,0.0004269099,0.0022019104,0.000004759644],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0031373156,0.00005948046,0.00012516415,0.00052191946,0.0003021585,0.00027763314,0.0017990178,0.00006611408,0.0000019745034],"category_scores_gemma":[0.0019686737,0.000030156574,0.000028035623,0.0023100632,0.0001413633,0.007294774,0.00070519786,0.00012616205,0.0000053901495],"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":[0.000094779294,0.0001820444,0.1608528,0.00067420467,0.000103030754,9.941e-8,0.00042112492,0.032275233,0.007130864,0.759931,0.011220616,0.027114166],"study_design_scores_gemma":[0.0005056566,0.000007885771,0.008755722,0.000085642525,0.00008437864,0.000005808839,0.00071354804,0.9415924,0.0007568281,0.018725881,0.02865959,0.00010663648],"about_ca_topic_score_codex":0.000011433847,"about_ca_topic_score_gemma":0.0000014125388,"teacher_disagreement_score":0.9093172,"about_ca_system_score_codex":0.00005018395,"about_ca_system_score_gemma":0.00015059373,"threshold_uncertainty_score":0.52885354},"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":"codex-gemma-dda1882f352a","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.002966569,0.0047743428,0.0000071242457,0.00096032483,0.9518125,0.002659044,0.000057111043,0.0001000308,0.03666296],"genre_scores_gemma":[0.0044309883,0.0018121039,0.00024536476,0.0004558993,0.92170227,0.00021139701,0.0020207043,0.00021359668,0.06890766],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99374545,0.000080299724,0.0019292136,0.0014178205,0.002164869,0.00066233455],"domain_scores_gemma":[0.99221617,0.00027784833,0.0030970264,0.002301525,0.0008315089,0.0012758973],"candidate_categories":["metaepi_narrow","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00092913135,0.00074352935,0.0017991092,0.00029860713,0.00023189349,0.0023347426,0.0007965919,0.00090511364,0.002426994],"category_scores_gemma":[0.0019332885,0.0005750057,0.00079173973,0.00026074811,0.000056513083,0.0011232379,0.0012280621,0.0012909959,0.00014513211],"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":[0.0001767047,0.0007202472,0.04188241,0.00020877086,0.0012323376,0.11432191,0.00006397034,0.0000020379725,0.0000019752563,0.0000037646694,0.8387705,0.0026153415],"study_design_scores_gemma":[0.001734721,0.00041306813,0.009454251,0.0015900128,0.002587401,0.0074908477,0.000021099777,8.257523e-7,0.00004338497,0.0006362567,0.9754748,0.0005533654],"about_ca_topic_score_codex":0.0025365378,"about_ca_topic_score_gemma":0.0000515796,"teacher_disagreement_score":0.13670424,"about_ca_system_score_codex":0.0010037329,"about_ca_system_score_gemma":0.011378638,"threshold_uncertainty_score":0.99967015},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.020542827,0.0001842527,0.00014525511,0.0076582697,0.0017353339,0.0008065836,0.000608715,0.00012297064,0.9681958],"genre_scores_gemma":[0.75708735,0.0011589021,0.04148148,0.0066181594,0.01352514,0.00019608934,0.002657095,0.00013075901,0.17714502],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9960069,0.00043055968,0.00027169107,0.00073706184,0.0022020969,0.00035169235],"domain_scores_gemma":[0.99750096,0.00018842198,0.00025989415,0.0015219871,0.0003924993,0.00013622985],"candidate_categories":["scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science"],"category_scores_codex":[0.013060134,0.000165688,0.00025763523,0.0000348014,0.0005883957,0.0021133635,0.008215588,0.00026068947,0.02581695],"category_scores_gemma":[0.00092719763,0.00014643854,0.000039309944,0.00015355309,0.00024797738,0.00076777383,0.021152686,0.00026920895,0.00013851887],"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":[0.00006060497,0.0001380471,0.0145046385,0.00020458197,0.00031061107,0.000003641875,0.026047418,0.000009929229,0.000006691063,0.07066198,0.8021118,0.08594003],"study_design_scores_gemma":[0.00024791624,0.000039535546,0.000686761,0.00014572009,0.000059214675,9.115413e-8,0.010417206,0.0014470347,0.00001028102,0.021243494,0.9653275,0.00037522335],"about_ca_topic_score_codex":0.06605144,"about_ca_topic_score_gemma":0.1044881,"teacher_disagreement_score":0.7910508,"about_ca_system_score_codex":0.00031499699,"about_ca_system_score_gemma":0.0030340063,"threshold_uncertainty_score":0.9989225},"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":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10887103,0.000054876138,0.7881599,0.09499923,0.00089559413,0.0020152717,0.00025013194,0.0000492368,0.0047046924],"genre_scores_gemma":[0.11481516,0.00023501104,0.87836015,0.0049913367,0.00023459636,0.00003306465,0.00016699391,0.00002584154,0.0011378622],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9897661,0.00037729065,0.0008711441,0.0025339692,0.004717786,0.0017336644],"domain_scores_gemma":[0.9877329,0.0002715608,0.00027633173,0.009509586,0.0013400593,0.0008695852],"candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["metaresearch","scholarly_communication","open_science"],"category_scores_codex":[0.06316291,0.00020857298,0.0002867829,0.0027242962,0.000960773,0.1387359,0.1754812,0.000040549116,0.000043729324],"category_scores_gemma":[0.0125229815,0.00017960029,0.000013461474,0.009334749,0.00015236213,0.44698343,0.2792544,0.0009194107,0.000276133],"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":[0.00029717945,0.00051346974,0.05038571,0.0001336461,0.000088166125,0.00064796157,0.0009614271,0.030649228,0.011721232,0.12980376,0.59110713,0.18369107],"study_design_scores_gemma":[0.0005050562,0.000075974305,0.0064132325,0.00034577973,0.000002260421,0.00008665122,0.00035740397,0.8766887,0.00021250104,0.0010822989,0.11385876,0.00037134998],"about_ca_topic_score_codex":0.0019412035,"about_ca_topic_score_gemma":0.00092276454,"teacher_disagreement_score":0.8460395,"about_ca_system_score_codex":0.00054202357,"about_ca_system_score_gemma":0.004801103,"threshold_uncertainty_score":0.99579495},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92277265,0.0031682723,0.0000019409742,0.06958579,0.0001055535,0.00064171315,0.0000018368871,0.0000027704446,0.0037194625],"genre_scores_gemma":[0.98848075,0.007933521,0.00063140935,0.0002047477,0.00007495459,0.00006441229,0.00001250656,0.000009487823,0.0025882253],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.991897,0.004416438,0.00039585397,0.00049940124,0.002402628,0.00038869522],"domain_scores_gemma":[0.99115074,0.006135583,0.000112054004,0.0008636416,0.0014880793,0.00024988048],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.04800783,0.00009795167,0.00015804412,0.0012034258,0.0011592017,0.017476127,0.0026762479,0.00008680655,0.00012119763],"category_scores_gemma":[0.0057022837,0.00007605702,0.000011534702,0.00066116406,0.00030975998,0.017270243,0.0027243567,0.0016861642,0.000035703768],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028326872,0.0009160661,0.626372,0.00003742336,0.00002656785,8.7504714e-7,0.008608975,0.0000027819765,0.0017569285,0.17913313,0.0016965305,0.18116546],"study_design_scores_gemma":[0.00051827397,0.00040208912,0.8074129,0.00015424693,0.0000010155151,0.0000043894825,0.012735364,0.00008014466,0.000029851786,0.1731529,0.0054274034,0.00008143041],"about_ca_topic_score_codex":0.0002914932,"about_ca_topic_score_gemma":0.00066018425,"teacher_disagreement_score":0.18108404,"about_ca_system_score_codex":0.0001744054,"about_ca_system_score_gemma":0.00039592801,"threshold_uncertainty_score":0.9964747},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9955313,0.0008126932,0.00020764513,0.0013037829,0.0004433122,0.0009106401,0.00006028094,0.000006311641,0.0007239901],"genre_scores_gemma":[0.99892265,0.00014185451,0.00034326222,0.00006284908,0.000023143297,0.000008554824,0.0000015016543,0.000013897187,0.00048227905],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.97376555,0.0024317822,0.0014986485,0.0010111089,0.020212078,0.0010808457],"domain_scores_gemma":[0.9783849,0.008512352,0.0012215439,0.0017518249,0.009483315,0.00064602785],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics"],"category_scores_codex":[0.108231656,0.0002111957,0.0006495731,0.26909086,0.0003236766,0.007054362,0.011431104,0.00019942832,0.00043864132],"category_scores_gemma":[0.043196917,0.0001450322,0.00019400302,0.61527854,0.00093990914,0.0059272307,0.0033438117,0.00092968275,0.000024694027],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000714087,0.0004190861,0.9661016,0.000020500667,0.000007404975,0.0000059402514,0.0014834232,0.003067941,0.007809614,0.00016598376,0.0003029797,0.020544104],"study_design_scores_gemma":[0.0010863772,0.00021378273,0.93484783,0.00011935488,0.0000043483183,0.0000016582044,0.008116978,0.019445434,0.029963762,0.005993111,0.00004977768,0.00015761389],"about_ca_topic_score_codex":0.004475062,"about_ca_topic_score_gemma":0.0009897391,"teacher_disagreement_score":0.34618768,"about_ca_system_score_codex":0.00096834963,"about_ca_system_score_gemma":0.0019846612,"threshold_uncertainty_score":0.9939764},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.987853,0.002320982,0.0002651549,0.0012230084,0.00007046589,0.0002734824,0.00658334,0.0000048882453,0.0014056675],"genre_scores_gemma":[0.9881102,0.00006447875,0.0067697642,0.00020899143,0.00008053217,0.000014647748,0.0018131079,0.000015169102,0.0029230986],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99930626,0.000027327647,0.00013347006,0.00028797684,0.00011286272,0.00013212317],"domain_scores_gemma":[0.9988724,0.00047744586,0.000055537133,0.00047482873,0.000069365895,0.000050404215],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00010839007,0.00008065844,0.00014616152,0.000008711186,0.00015867298,0.0001143117,0.00057675876,0.000016832499,0.0004689432],"category_scores_gemma":[0.00005531286,0.000063323016,0.000017421922,0.00009203182,0.00006274667,0.00032802988,0.0017963389,0.000101329366,0.0000042785528],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003600053,0.00057574623,0.52078366,0.00028330673,0.00094263814,0.000035132372,0.0044930982,0.00021269282,0.047382567,0.0199685,0.033631537,0.37133116],"study_design_scores_gemma":[0.0023505988,0.000066656336,0.5899283,0.00022023765,0.00004097499,0.0000017800313,0.002715399,0.0006818019,0.004072703,0.151681,0.24788393,0.00035655982],"about_ca_topic_score_codex":0.00009917006,"about_ca_topic_score_gemma":0.00012534788,"teacher_disagreement_score":0.37097457,"about_ca_system_score_codex":0.0000040542727,"about_ca_system_score_gemma":0.000075105745,"threshold_uncertainty_score":0.5134599},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07095672,0.000985523,0.3737921,0.1184228,0.00072772405,0.0024893072,0.000018580231,0.000026406227,0.43258083],"genre_scores_gemma":[0.9649937,0.000012747565,0.008165293,0.011220453,0.0002527095,0.000035121466,0.0000060012912,0.0000067994565,0.015307165],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99016577,0.0026342832,0.0033970124,0.0004760389,0.0032473968,0.00007951816],"domain_scores_gemma":[0.996251,0.0005375215,0.0012696474,0.0013492085,0.00044978014,0.00014288089],"candidate_categories":["metaresearch","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch","insufficient_payload"],"category_scores_codex":[0.040531468,0.000083663625,0.001150396,0.000040228962,0.00006409329,0.0015297936,0.0023014427,0.00002131733,0.02246228],"category_scores_gemma":[0.033792302,0.00003766659,0.00039125854,0.0007430561,0.000008923921,0.000546633,0.00029638052,0.000045271056,0.050473128],"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":[0.0000021663823,0.000007425195,0.0028189179,0.000003404961,0.00002443155,0.000001580299,0.00088269357,0.000060582373,0.00009648615,0.0036013508,0.97726566,0.015235324],"study_design_scores_gemma":[0.00007008615,0.000017399472,0.0011607239,0.000002675769,0.00002041612,4.1393383e-7,0.0005485929,0.0070622,0.00015424991,0.0016687384,0.98923653,0.000057960508],"about_ca_topic_score_codex":0.0000035353244,"about_ca_topic_score_gemma":0.0000053917806,"teacher_disagreement_score":0.894037,"about_ca_system_score_codex":0.0000043882633,"about_ca_system_score_gemma":0.000028527167,"threshold_uncertainty_score":0.9995067},"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":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08218334,0.0015352829,0.89207244,0.023233159,0.000043934568,0.00082923763,0.000053215863,0.000044069657,0.000005344013],"genre_scores_gemma":[0.36566362,0.0015326309,0.6326327,0.000107257445,0.000018591776,0.000024746418,0.000013181388,0.0000066485472,6.4346096e-7],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9984534,0.00012008636,0.00016980796,0.0007911927,0.00026028234,0.00020523078],"domain_scores_gemma":[0.9972489,0.00058550376,0.00015406145,0.0019141356,0.000011670431,0.00008576185],"candidate_categories":["open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.0014731401,0.000112756716,0.00015022373,0.000083319515,0.00021285165,0.00031909742,0.007169334,0.00005789456,0.000003480924],"category_scores_gemma":[0.0051391157,0.00009754887,0.000006382635,0.00012766135,0.0001271683,0.00070334174,0.048483483,0.00020228907,4.8048736e-7],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003134203,0.00018956518,0.0059228777,0.00037774976,0.00005199413,0.0000062770637,0.00036118785,0.0035736423,0.01573078,0.004547304,0.118619345,0.85030586],"study_design_scores_gemma":[0.00085450755,0.00047931704,0.0029157097,0.000088085406,0.00001513011,0.00000354807,0.0000483777,0.947863,0.009518444,0.008537015,0.0295047,0.00017216992],"about_ca_topic_score_codex":0.000016657483,"about_ca_topic_score_gemma":1.7985033e-7,"teacher_disagreement_score":0.9442893,"about_ca_system_score_codex":0.000030019573,"about_ca_system_score_gemma":0.000035267254,"threshold_uncertainty_score":0.9982024},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00055632455,0.00043943457,0.9210995,0.06060238,0.0037909069,0.00096588593,0.0029899115,0.0003062158,0.009249465],"genre_scores_gemma":[0.33624765,0.000683527,0.33517107,0.0023227911,0.004216848,0.0004623708,0.020623216,0.00013046096,0.30014205],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978414,0.00003923405,0.00020628878,0.00096672383,0.0006246713,0.00032163932],"domain_scores_gemma":[0.9943673,0.0004363721,0.000115271345,0.0048468993,0.00010484731,0.00012933313],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.0021477712,0.00007645638,0.00006558477,0.000031548683,0.0009497717,0.038372423,0.015265726,0.000013048014,0.0000019674528],"category_scores_gemma":[0.006170863,0.000050607465,0.000023469138,0.0006003391,0.0002592145,0.040000178,0.01025894,0.00008066493,0.00007849642],"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":[0.0000182429,0.000029928793,0.00019681067,0.000037075853,0.000034816898,0.000010054145,0.00024888478,0.0000073826836,0.00018879249,0.42583832,0.51791066,0.055479016],"study_design_scores_gemma":[0.00007565916,0.000024418883,0.0001751642,0.0000030989454,0.000003187548,0.0000011731538,0.000079109515,0.046195887,0.00023601766,0.00058139116,0.9525578,0.00006707858],"about_ca_topic_score_codex":0.0000036391814,"about_ca_topic_score_gemma":0.000009160065,"teacher_disagreement_score":0.5859284,"about_ca_system_score_codex":0.000011219001,"about_ca_system_score_gemma":0.00012214547,"threshold_uncertainty_score":0.9977459},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008790772,0.0000051380352,0.31304988,0.0062585995,0.0007138727,0.00056952896,0.000013802264,0.0012187464,0.66937965],"genre_scores_gemma":[0.8146521,0.00003898246,0.014104762,0.0025278674,0.00047563054,0.000085042964,0.00004252069,0.00006154562,0.16801155],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99454135,0.00069655693,0.00071490207,0.002828751,0.00055606297,0.0006624003],"domain_scores_gemma":[0.9941943,0.00058320205,0.00029590572,0.0042433823,0.00008326068,0.00059989985],"candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.001786226,0.00039147516,0.00049668923,0.00010211063,0.00024281055,0.000806144,0.0051679695,0.00023697561,0.12386799],"category_scores_gemma":[0.0012050868,0.00042559754,0.0002802173,0.0008447173,0.000096432545,0.001054236,0.0049981223,0.00062459765,0.8102171],"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":[0.00011296912,0.00041642078,0.017711006,0.00025430115,0.00028057653,0.00028600916,0.0074763615,0.017424522,0.0014634656,0.04334028,0.8492624,0.061971676],"study_design_scores_gemma":[0.0008141021,0.0000027101241,0.0035923286,0.00014751068,0.000026468408,0.00007366562,0.00011221095,0.07886977,0.0023628757,0.005354734,0.90779555,0.000848104],"about_ca_topic_score_codex":0.000026638214,"about_ca_topic_score_gemma":0.0000017528698,"teacher_disagreement_score":0.80586135,"about_ca_system_score_codex":0.00006296493,"about_ca_system_score_gemma":0.0002379959,"threshold_uncertainty_score":0.9998196},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95864135,0.00036539682,0.038490627,0.0006863337,0.000033877135,0.0001160934,0.00007507339,0.000029785213,0.0015614523],"genre_scores_gemma":[0.9904517,0.00014227064,0.007345712,0.0013409994,0.00008586944,0.0000032938262,0.0004676985,0.000018632412,0.00014384204],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993905,0.00010491049,0.00017889142,0.000111023415,0.00011575054,0.00009894571],"domain_scores_gemma":[0.9995962,0.000005047015,0.000059387905,0.00021517821,0.0000391013,0.00008509905],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00006058688,0.00007775696,0.000106278334,0.000016667836,0.00005134346,0.000008091336,0.00014719508,0.00007373212,0.00002653111],"category_scores_gemma":[0.0000425838,0.000080512655,0.000054468404,0.00007441637,0.00003666906,0.0000036912365,0.00013377715,0.00008200557,0.0000037149584],"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":[0.0020803409,0.0005007823,0.09016239,0.00045909538,0.00038377038,0.000009293246,0.0026354617,0.29601255,0.48836723,0.01008005,0.09667826,0.012630785],"study_design_scores_gemma":[0.0069405944,0.0061336793,0.21631442,0.00010902003,0.0001661779,0.000032593096,0.0020712458,0.48584467,0.16913293,0.01053898,0.10087577,0.0018398949],"about_ca_topic_score_codex":0.000016887563,"about_ca_topic_score_gemma":0.000010598054,"teacher_disagreement_score":0.31923428,"about_ca_system_score_codex":0.0000038667545,"about_ca_system_score_gemma":0.000037294787,"threshold_uncertainty_score":0.32832107},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.988557,0.0021695714,0.00028285908,0.007909392,0.00005801684,0.00018417774,0.00027743576,0.0000072760517,0.00055429345],"genre_scores_gemma":[0.9899385,0.00031044055,0.0010202898,0.0080811605,0.00042701137,0.000007099907,0.00014170342,0.000013154725,0.00006066144],"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","domain_scores_codex":[0.99935895,0.000043585183,0.00017220604,0.00025915884,0.0000506685,0.00011541116],"domain_scores_gemma":[0.9990664,0.000022527907,0.00011137962,0.0007526084,0.00002479951,0.00002228406],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003128755,0.000084222076,0.00015497647,0.0000039914985,0.00005915448,0.000021740792,0.0009057795,0.000037126818,0.000010082244],"category_scores_gemma":[0.00009927296,0.000051128125,0.000030103902,0.000034573106,0.00019553746,0.000003876497,0.0033738678,0.00005788654,0.0000024171152],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00064517406,0.000009664564,0.04075389,0.00004774316,0.00011009013,7.783874e-7,0.00045463364,0.00002131778,0.95351774,0.00051875977,0.0032746764,0.0006455377],"study_design_scores_gemma":[0.010160312,0.00019584247,0.087781474,0.000053603315,0.00022167606,0.00013068899,0.00038169068,0.004964993,0.27494106,0.0005457292,0.6199602,0.00066277565],"about_ca_topic_score_codex":0.000014234692,"about_ca_topic_score_gemma":0.000026079428,"teacher_disagreement_score":0.6785767,"about_ca_system_score_codex":0.000004257604,"about_ca_system_score_gemma":0.000020917007,"threshold_uncertainty_score":0.4205285},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.71598464,0.00504351,0.16914362,0.028468762,0.004407408,0.017588083,0.010382738,0.00028203343,0.048699234],"genre_scores_gemma":[0.9697585,0.0026681372,0.018397523,0.00051383185,0.0024740542,0.0010921961,0.0035517404,0.00017143576,0.0013725418],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99707973,0.0005827233,0.0004992386,0.00040095625,0.00064883125,0.0007885458],"domain_scores_gemma":[0.9958278,0.0011443556,0.00014922525,0.001510376,0.00111711,0.00025116056],"candidate_categories":["sts","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.0031832552,0.00027180873,0.0002668941,0.00010136278,0.0023511485,0.0006865434,0.0019582526,0.00053582713,0.00003730912],"category_scores_gemma":[0.0029370124,0.0002352817,0.00020064691,0.00024744385,0.0005304261,0.0000043566583,0.005400854,0.0024956425,0.00003193123],"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":[0.00038028727,0.0001514609,0.0024505428,0.00093625503,0.0006052154,0.0000042366873,0.0015066128,0.0018217483,0.06362034,0.050926134,0.8666992,0.010897975],"study_design_scores_gemma":[0.00094993814,0.00051689736,0.0015750842,0.000109158034,0.000047265832,0.000007631626,0.0010755045,0.0052086026,0.021319935,0.1160394,0.8525912,0.0005593792],"about_ca_topic_score_codex":0.00046157514,"about_ca_topic_score_gemma":0.00029570455,"teacher_disagreement_score":0.25377393,"about_ca_system_score_codex":0.00020398905,"about_ca_system_score_gemma":0.001328074,"threshold_uncertainty_score":0.9998056},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008069743,0.0051000393,0.0006247596,0.95629823,0.0019008124,0.010239081,0.0015939137,0.00012536896,0.01604804],"genre_scores_gemma":[0.9345994,0.011257224,0.016452076,0.016909862,0.0025950386,0.0015545095,0.01060256,0.00040401422,0.0056253327],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9834565,0.0045741703,0.0014393544,0.004462591,0.004351419,0.0017159515],"domain_scores_gemma":[0.9589548,0.009404747,0.00055781147,0.029163552,0.0012418791,0.00067721517],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","open_science"],"category_scores_codex":[0.056068134,0.00047607016,0.0010194137,0.00031971306,0.0024286692,0.003514814,0.04529882,0.00047770172,0.00059895054],"category_scores_gemma":[0.026467841,0.00033872493,0.00016985896,0.003939997,0.0012228234,0.0005540189,0.42156208,0.006608725,0.00007207488],"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":[0.0006635708,0.0015766802,0.045603026,0.005394991,0.001195401,0.00043783974,0.0026398285,0.00063700404,0.0010589029,0.0058441055,0.8791017,0.055846956],"study_design_scores_gemma":[0.0014602316,0.000091692804,0.0059810556,0.011218177,0.00016159013,0.000060746297,0.0029751335,0.012398231,0.00026237406,0.0031026609,0.96181065,0.0004774678],"about_ca_topic_score_codex":0.026618304,"about_ca_topic_score_gemma":0.0072484068,"teacher_disagreement_score":0.9393884,"about_ca_system_score_codex":0.0024349985,"about_ca_system_score_gemma":0.006741542,"threshold_uncertainty_score":0.9999065},"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":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00060323294,0.000015423484,0.88475394,0.06977154,0.00022377673,0.00052985246,0.0000019788051,0.00009565241,0.044004567],"genre_scores_gemma":[0.9142701,0.000013258019,0.019574674,0.055736076,0.0004082254,0.00013994904,0.000023608147,0.000007681796,0.009826431],"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992756,0.000026085545,0.00021535534,0.00015875163,0.00018860328,0.00013556697],"domain_scores_gemma":[0.9994485,0.00005168263,0.0000740078,0.00028196533,0.000059529488,0.000084363244],"candidate_categories":["scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.00032868382,0.000056773664,0.00007410945,0.000016732718,0.00049518346,0.01573731,0.0050815972,0.000018591067,0.00003372193],"category_scores_gemma":[0.000016474647,0.000032716758,0.000027507454,0.00033658242,0.000025918742,0.039803017,0.0010856396,0.000027093483,0.0002899698],"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":[0.0000044168137,0.0000063213524,0.000022064522,0.0000046763757,0.000006272451,1.5899447e-7,0.0029275469,0.005332888,0.0008208251,0.38555616,0.5865657,0.018753005],"study_design_scores_gemma":[0.00009613423,0.000027049007,0.0000073050765,9.715274e-7,1.909811e-7,5.5546576e-7,0.000042962507,0.83680934,0.0014084234,0.0018179348,0.15973629,0.000052840147],"about_ca_topic_score_codex":0.0000151067625,"about_ca_topic_score_gemma":0.000001294736,"teacher_disagreement_score":0.91366684,"about_ca_system_score_codex":0.000006525361,"about_ca_system_score_gemma":0.00012251384,"threshold_uncertainty_score":0.98528445},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.052759483,0.00003361569,0.0003616495,0.0015192997,0.000102077676,0.00045987577,0.0000012221232,0.00001526581,0.9447475],"genre_scores_gemma":[0.9940845,0.00001735306,0.0003061424,0.002335636,0.00007455659,8.9003515e-7,0.000009446035,0.000004884298,0.0031665887],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99939406,0.0000141582805,0.00010727649,0.00018693868,0.00016929935,0.00012825431],"domain_scores_gemma":[0.9997027,0.000022740138,0.00008125122,0.00013075261,0.000051579347,0.000010993511],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0002975396,0.00008069559,0.00011497133,0.00015204864,0.00016881409,0.0003603746,0.001629962,0.000046933328,0.0034737778],"category_scores_gemma":[0.000023802042,0.000058558722,0.000042139432,0.00038713307,0.000079465106,0.0033829343,0.00096062705,0.00012728853,0.0002481779],"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":[0.00074713025,0.0008895969,0.01388496,0.000101484235,0.000056113586,0.000012740109,0.015146255,0.00002379782,0.00003684262,0.15899496,0.80593944,0.004166662],"study_design_scores_gemma":[0.0005310529,0.000016075548,0.076201975,0.00007547444,0.000029227636,9.498381e-7,0.005433953,0.003104317,0.000029939478,0.0024324022,0.9119728,0.00017182116],"about_ca_topic_score_codex":0.0023091908,"about_ca_topic_score_gemma":0.00013500801,"teacher_disagreement_score":0.9415809,"about_ca_system_score_codex":0.000029289973,"about_ca_system_score_gemma":0.000021576174,"threshold_uncertainty_score":0.9974372},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4677164,0.00044577252,0.4944633,0.036315557,0.00020469785,0.000375091,0.00001281404,0.00023362637,0.00023275135],"genre_scores_gemma":[0.9515356,0.00036482996,0.047804955,0.000102422324,0.00008280528,0.000010022996,0.000006581169,0.00001562647,0.00007718065],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9976802,0.00008557047,0.00024253622,0.00043442697,0.00023464502,0.0013226616],"domain_scores_gemma":[0.9842106,0.000031692543,0.00029420463,0.015295146,0.00007869393,0.000089632405],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["open_science"],"category_scores_codex":[0.005594434,0.00015199586,0.00017545988,0.00013946964,0.001992707,0.0016037181,0.06258564,0.0000853992,0.0000013349214],"category_scores_gemma":[0.01023959,0.00014431206,0.00001930593,0.00014841167,0.00007732425,0.0019114795,0.15144335,0.0024056984,0.000014135358],"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":[0.00018708542,0.0003602335,0.09110229,0.00011801454,0.0007824478,0.00006320168,0.002866326,0.00013812447,0.017147955,0.18785316,0.041901253,0.65747994],"study_design_scores_gemma":[0.000564076,0.00035401934,0.011815787,0.000067076166,0.000016077,0.00046163666,0.00044694953,0.04836645,0.00053482724,0.93482167,0.0022301301,0.00032130838],"about_ca_topic_score_codex":0.00042368134,"about_ca_topic_score_gemma":0.0029512749,"teacher_disagreement_score":0.7469685,"about_ca_system_score_codex":0.0005985493,"about_ca_system_score_gemma":0.00088776473,"threshold_uncertainty_score":0.9998958},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.004326582,0.013786429,0.8457931,0.1324113,0.00040763486,0.0021000195,0.000008400569,0.00006904511,0.0010974285],"genre_scores_gemma":[0.9742684,0.019063413,0.0056061205,0.00072458293,0.00012589175,0.00002401027,0.0000018570211,0.000013922741,0.00017182095],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99363744,0.0015512194,0.0005171761,0.0008681634,0.0013162624,0.002109753],"domain_scores_gemma":[0.99542284,0.0017744993,0.00037278325,0.0020585593,0.00014575232,0.00022554414],"candidate_categories":["metaresearch","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","scholarly_communication","open_science"],"category_scores_codex":[0.061656654,0.00014657083,0.00026109506,0.0001388639,0.0012455498,0.013877629,0.027298264,0.0000580938,0.00000139029],"category_scores_gemma":[0.017254654,0.00009574109,0.00002039464,0.001391372,0.00018025671,0.02859124,0.011653364,0.004729029,0.00002097571],"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":[0.000015967516,0.000018332337,0.0002707772,0.00038976,0.00005796854,0.000017206681,0.0006986315,0.0000128694965,0.000016828275,0.99667543,0.00014827997,0.0016779653],"study_design_scores_gemma":[0.0015199997,0.00083391916,0.0011410674,0.0035186545,0.00007914107,0.0006577187,0.015902441,0.2010194,0.00002661794,0.76375717,0.010791559,0.00075232814],"about_ca_topic_score_codex":0.00010831663,"about_ca_topic_score_gemma":0.0013708923,"teacher_disagreement_score":0.9699418,"about_ca_system_score_codex":0.00045475154,"about_ca_system_score_gemma":0.0048849485,"threshold_uncertainty_score":0.9975671},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.932056,0.00009838599,0.000003292287,0.0060913907,0.00012591037,0.0005269355,0.000016140688,0.000032504577,0.061049417],"genre_scores_gemma":[0.98986125,0.0004128512,0.00020468004,0.00041317948,0.0001972333,0.00011730942,0.0000045833704,0.000011452936,0.008777434],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9984107,0.00021267451,0.00024533138,0.00037688142,0.0003813871,0.00037302365],"domain_scores_gemma":[0.9981593,0.0003955868,0.000118368764,0.00009569684,0.001019462,0.00021158736],"candidate_categories":["scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0009054772,0.00010402756,0.00025100494,0.00016514852,0.00026400233,0.0021622605,0.0013366014,0.00016420743,0.0011537243],"category_scores_gemma":[0.004422605,0.000106818756,0.00004423845,0.001279803,0.0004806806,0.005533127,0.0007056746,0.00031596885,0.000017439821],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000064856154,0.00016440451,0.62816745,0.00003441469,0.000013963009,0.000003791715,0.34073204,7.9589235e-8,0.00008734866,0.015341681,0.009615067,0.005774876],"study_design_scores_gemma":[0.0003122571,0.00006097906,0.6179291,0.00015165782,0.0000059264466,7.0044985e-7,0.2971241,0.0000028736645,0.0006110287,0.016030805,0.067605235,0.00016537157],"about_ca_topic_score_codex":0.005549689,"about_ca_topic_score_gemma":0.0013596407,"teacher_disagreement_score":0.057990164,"about_ca_system_score_codex":0.000058400678,"about_ca_system_score_gemma":0.0013493852,"threshold_uncertainty_score":0.9997594},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4131716,0.054681726,0.39056024,0.12894528,0.008319896,0.003115462,0.00024096807,0.000035559322,0.0009293025],"genre_scores_gemma":[0.829801,0.0025010277,0.16393223,0.00227559,0.00066219416,0.000061616025,0.00049310375,0.000034676596,0.000238529],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9993563,0.00010118228,0.00019560075,0.00015541799,0.00008841424,0.00010306332],"domain_scores_gemma":[0.99857634,0.0008563332,0.0001129386,0.00034801796,0.00007864303,0.00002774788],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00060806924,0.000050126437,0.00021446472,0.000020463314,0.00007475164,0.000014184389,0.00038619226,0.00005751208,0.0000066988237],"category_scores_gemma":[0.0013065893,0.000030645664,0.00002193238,0.000109549146,0.00023357193,0.00003714357,0.00017316637,0.00020255103,5.191166e-8],"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":[0.0005412491,0.00015522515,0.1360742,0.000105760766,0.000104995204,0.000012294331,0.00040848294,0.00018762844,0.004366425,0.00061835407,0.053055152,0.8043702],"study_design_scores_gemma":[0.0045049884,0.0003574827,0.0413614,0.00008693904,0.00024598278,0.0000630377,0.004262027,0.103634216,0.0020148605,0.0037774495,0.8396022,0.000089427565],"about_ca_topic_score_codex":0.00004059739,"about_ca_topic_score_gemma":0.0000327315,"teacher_disagreement_score":0.8042808,"about_ca_system_score_codex":0.000021453023,"about_ca_system_score_gemma":0.00035281645,"threshold_uncertainty_score":0.1564204},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9812782,0.000015936825,0.0011565706,0.011977004,0.0004844512,0.0001477159,7.3914106e-7,0.000014457458,0.004924937],"genre_scores_gemma":[0.9963303,0.0000522851,0.0019514498,0.0014475086,0.00017169816,0.000008377558,0.0000050255117,0.0000037681843,0.000029570148],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99875605,0.000018027924,0.00064000563,0.00007387797,0.00033400825,0.00017801132],"domain_scores_gemma":[0.99829465,0.000038344024,0.0004371708,0.00013499756,0.0010849204,0.000009945001],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.001028179,0.00007248887,0.00014875374,0.0019605362,0.0002048711,0.0011056158,0.00073302136,0.00010010699,0.00037830908],"category_scores_gemma":[0.000621676,0.00005666784,0.00003812394,0.0033035798,0.00015070799,0.007022476,0.0006826519,0.0003924564,0.00019798522],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017227825,0.000048310896,0.30752665,0.000053457377,0.00010279289,0.00001689064,0.00077386387,0.00035591185,0.00042108938,0.050109703,0.02842238,0.61199665],"study_design_scores_gemma":[0.002751309,0.00011320683,0.34814826,0.0005139946,0.0000994246,0.0011247678,0.05961755,0.0021065169,0.012876425,0.11561924,0.4564519,0.00057741744],"about_ca_topic_score_codex":0.000063816165,"about_ca_topic_score_gemma":0.00009472696,"teacher_disagreement_score":0.61141926,"about_ca_system_score_codex":0.00010644224,"about_ca_system_score_gemma":0.00012476585,"threshold_uncertainty_score":0.99993134},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"}],"label_agreement":"split"},{"id":"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.88595307,0.00051696267,0.000046215868,0.01584983,0.00012917236,0.0001004737,0.0000026311557,0.000058201902,0.097343475],"genre_scores_gemma":[0.99457073,0.0028706698,0.00012849356,0.0013104149,0.00009362481,0.000010694484,0.000013251431,0.0000052425257,0.0009968926],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988141,0.00076752674,0.00008929546,0.000064838205,0.0001308349,0.00013338188],"domain_scores_gemma":[0.99931735,0.00026854724,0.0000393946,0.0002709332,0.00006179648,0.000041984837],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.0008657624,0.0000487203,0.00006894567,0.000015117072,0.0015288187,0.00013517457,0.00018189543,0.000045804078,0.00039674385],"category_scores_gemma":[0.00023282482,0.000041259667,0.000030405483,0.00016206273,0.00018827208,0.00016502554,0.00014109146,0.00023100761,0.000026837308],"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":[0.000034097706,0.00014646548,0.010753668,0.0000787609,0.000041399493,0.000010404232,0.51145214,0.0000029976065,0.011566148,0.07538459,0.0047834967,0.38574585],"study_design_scores_gemma":[0.00025155034,0.00002353089,0.012065906,0.000083358755,0.000022185817,0.000009109294,0.36300173,0.00008275677,0.00028346194,0.0044502188,0.6195607,0.00016547367],"about_ca_topic_score_codex":0.0023172915,"about_ca_topic_score_gemma":0.017526118,"teacher_disagreement_score":0.6147772,"about_ca_system_score_codex":0.000056545785,"about_ca_system_score_gemma":0.000031994874,"threshold_uncertainty_score":0.99977106},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9479204,0.0042287945,0.0005178048,0.010948379,0.0004474581,0.00059193675,0.0000051767024,0.0000056778904,0.03533441],"genre_scores_gemma":[0.99237984,0.00043336552,0.0056874314,0.00014189516,0.000028439239,0.000032788543,1.4626129e-7,0.0000034585216,0.00129265],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.96469176,0.018396266,0.0043328553,0.00643433,0.005604085,0.0005406974],"domain_scores_gemma":[0.97022873,0.012215503,0.003299846,0.011454862,0.002620866,0.00018018717],"candidate_categories":["metaresearch","sts","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.37310612,0.00019305604,0.0011843108,0.00043033017,0.00041812626,0.00043455858,0.0037427142,0.00010118701,0.004156955],"category_scores_gemma":[0.39923227,0.00008170154,0.0003017064,0.007155955,0.00319495,0.00040163263,0.00024000733,0.0007256275,0.00030588463],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002685632,0.0054618744,0.63569605,0.00033003074,0.00013092114,0.00019108596,0.06401878,0.00069033285,0.03464501,0.088226944,0.011906525,0.15843387],"study_design_scores_gemma":[0.00019573171,0.00029817765,0.91227764,0.000058045116,0.000021479544,0.00009668678,0.0058525642,0.002069968,0.00007956713,0.078402504,0.00052278687,0.00012486528],"about_ca_topic_score_codex":0.000015431038,"about_ca_topic_score_gemma":0.000013073965,"teacher_disagreement_score":0.27658156,"about_ca_system_score_codex":0.000042357668,"about_ca_system_score_gemma":0.00009722513,"threshold_uncertainty_score":0.9995178},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"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":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.29270557,0.0060080495,0.009956212,0.6797446,0.0044323527,0.0038630287,0.00008891509,0.000085608095,0.0031156663],"genre_scores_gemma":[0.92231727,0.020980768,0.0028961077,0.007164151,0.0132208085,0.0001403662,0.00017008263,0.00015075594,0.032959677],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.97386986,0.004414458,0.003800919,0.001519419,0.014678815,0.0017165133],"domain_scores_gemma":[0.49699134,0.44586724,0.0014944564,0.00971392,0.026580501,0.019352546],"candidate_categories":["metaresearch","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","open_science"],"category_scores_codex":[0.2252758,0.00023036812,0.001489731,0.0010235237,0.00031681298,0.00028107318,0.005448588,0.0008918537,0.0060668876],"category_scores_gemma":[0.8762583,0.00016633733,0.00072199816,0.0017641289,0.0026098068,0.00025597806,0.009511479,0.011450944,0.00018352228],"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":[0.09364605,0.005210745,0.009594926,0.0020113739,0.002307345,0.018636882,0.00062840816,0.0000044778626,0.00017454753,0.004013119,0.7762008,0.08757134],"study_design_scores_gemma":[0.09761091,0.010261327,0.0009374879,0.0017310893,0.00013522002,0.00042387354,0.00093457685,0.016974898,0.000117127674,0.0091958605,0.86147636,0.00020128733],"about_ca_topic_score_codex":0.00027235338,"about_ca_topic_score_gemma":0.00011512854,"teacher_disagreement_score":0.6725804,"about_ca_system_score_codex":0.0009248404,"about_ca_system_score_gemma":0.020675795,"threshold_uncertainty_score":0.9999324},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.022856157,0.0979009,0.00010070519,0.42918292,0.0004750595,0.002710815,0.00017964281,0.00008953255,0.4465043],"genre_scores_gemma":[0.5347802,0.41467884,0.0006302941,0.0006894488,0.0010737687,0.00020338214,0.00012080867,0.000033602522,0.047789626],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9596282,0.025764395,0.00086153334,0.0010020868,0.011107224,0.0016365108],"domain_scores_gemma":[0.9682428,0.00992643,0.00011696843,0.00090187346,0.020179832,0.00063205865],"candidate_categories":["metaresearch","sts","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","sts","research_integrity"],"category_scores_codex":[0.18752106,0.00021373144,0.00067240134,0.0005679396,0.0018980358,0.0003562797,0.0012321617,0.0014171561,0.0018201447],"category_scores_gemma":[0.09845751,0.00018218404,0.00019286422,0.0035082,0.0037236833,0.00043904703,0.0005327136,0.028544076,0.00036019037],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000933581,0.00031263754,0.000548961,0.0036464657,0.000041130923,0.00006160583,0.007197763,1.2506695e-7,0.00014897368,0.629034,0.32530773,0.03360724],"study_design_scores_gemma":[0.00018534984,0.0003207902,0.00051831815,0.009201858,0.000012541673,0.0000143996385,0.016110886,0.000016856848,0.0013248228,0.06082572,0.9112777,0.00019073008],"about_ca_topic_score_codex":0.018948874,"about_ca_topic_score_gemma":0.0045730025,"teacher_disagreement_score":0.58597,"about_ca_system_score_codex":0.0005153292,"about_ca_system_score_gemma":0.004282423,"threshold_uncertainty_score":0.99987924},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.893646,0.00177158,0.000029710216,0.042072654,0.00015029717,0.0005310362,0.000028014276,0.000004890895,0.06176579],"genre_scores_gemma":[0.98483115,0.00082962064,0.00013437525,0.0002678487,0.000087022534,0.000018434066,3.452256e-7,0.0000063941757,0.0138248075],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.982298,0.000953059,0.0005771977,0.0006708198,0.014537131,0.000963764],"domain_scores_gemma":[0.9599999,0.009186872,0.00025270053,0.001968057,0.028277485,0.0003149513],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","open_science"],"category_scores_codex":[0.076967746,0.00008472808,0.00017435085,0.018258698,0.0025459377,0.007745758,0.009931601,0.00006254338,0.000061384875],"category_scores_gemma":[0.32131895,0.000039769922,0.000047654514,0.49488565,0.0019416692,0.0012805751,0.013252809,0.00049284945,0.000074788804],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000029073693,0.00015332803,0.034760978,0.00001398899,0.00001292193,0.000007645207,0.0009934036,0.000009386551,0.03160311,0.7232306,0.035315722,0.17386983],"study_design_scores_gemma":[0.00064461306,0.00016897419,0.3388201,0.000040621337,0.00000171205,0.000022051754,0.0024430342,0.0026295201,0.1219528,0.32700714,0.20610656,0.00016287235],"about_ca_topic_score_codex":0.001407805,"about_ca_topic_score_gemma":0.00016899602,"teacher_disagreement_score":0.47662696,"about_ca_system_score_codex":0.00019797355,"about_ca_system_score_gemma":0.007047017,"threshold_uncertainty_score":0.9987526},"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":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10667544,0.09690798,0.040862277,0.649778,0.0009620232,0.0018416926,0.00021549,0.000021915608,0.102735154],"genre_scores_gemma":[0.82910854,0.021446384,0.14161916,0.0014373871,0.00024655712,0.000007355714,0.000007682524,0.000020094585,0.0061068484],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9881377,0.00013456766,0.00074250984,0.00045576907,0.010113021,0.0004164196],"domain_scores_gemma":[0.9745194,0.011566685,0.00037892914,0.0005406067,0.012131879,0.000862534],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.041428518,0.00012746024,0.00064281176,0.0030124413,0.00017723022,0.0016017917,0.001607953,0.000083296814,0.0026510446],"category_scores_gemma":[0.21775268,0.00006282571,0.000103424965,0.027802063,0.0018266055,0.0010958045,0.0011202344,0.00036170645,0.000008742417],"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":[0.00022376979,0.000120935896,0.00025126818,0.000052820757,0.000021246875,0.000026989692,0.00020926079,2.3292509e-7,0.00046348124,0.917307,0.036563832,0.04475915],"study_design_scores_gemma":[0.0009959594,0.00052657514,0.003213277,0.0003773771,0.00006482353,0.00037523874,0.0004868647,0.00007241403,0.016301965,0.9081633,0.0692633,0.00015887308],"about_ca_topic_score_codex":0.000026532958,"about_ca_topic_score_gemma":0.000003463689,"teacher_disagreement_score":0.7224331,"about_ca_system_score_codex":0.000015409923,"about_ca_system_score_gemma":0.00087711104,"threshold_uncertainty_score":0.99943465},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.45462823,0.00021266374,0.020633139,0.021717444,0.00087798684,0.0038570932,0.00006506793,0.00006641488,0.49794194],"genre_scores_gemma":[0.99555135,0.00035840884,0.0036068119,0.00020649431,0.00006769613,0.000047357793,0.00001260066,0.0000037334542,0.00014555159],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.9971339,0.0007648253,0.00028065805,0.0005898243,0.00092624954,0.00030451402],"domain_scores_gemma":[0.99835867,0.00014445452,0.0001788568,0.0005207663,0.00072745746,0.0000698158],"candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.005602171,0.00014484735,0.0002892395,0.00018927969,0.0006762664,0.0012983766,0.0043505556,0.00008072159,0.0005391616],"category_scores_gemma":[0.00045273875,0.000108024884,0.00005289933,0.000678365,0.0010212745,0.0005295109,0.005376855,0.00031014162,0.0000011222294],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00002381259,0.001189784,0.0016178772,0.000045511362,0.000032318116,0.000019129018,0.19468442,0.0017422605,0.00012383828,0.7879695,0.00011284469,0.012438691],"study_design_scores_gemma":[0.00022311762,0.000112099086,0.019487424,0.0003350561,0.00003119304,5.6638817e-7,0.93002796,0.0047850017,0.00006814758,0.043298595,0.0012940016,0.0003368381],"about_ca_topic_score_codex":0.24998409,"about_ca_topic_score_gemma":0.27198705,"teacher_disagreement_score":0.7446709,"about_ca_system_score_codex":0.00039289956,"about_ca_system_score_gemma":0.004506428,"threshold_uncertainty_score":0.9997384},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9831989,0.00007773844,0.0064793457,0.00045396076,0.00063934457,0.0076278304,0.0000021551045,0.00007277294,0.0014479851],"genre_scores_gemma":[0.89542955,0.0000023171717,0.09704514,0.0006304883,0.00024873184,0.0064612445,0.000018673114,0.000032897107,0.0001309591],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.9975095,0.0010239853,0.00050001964,0.00052555377,0.00015071167,0.00029024165],"domain_scores_gemma":[0.99833196,0.0005459367,0.00018634374,0.00026121413,0.000658642,0.000015902255],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01793626,0.00018376812,0.00030696802,0.00038544994,0.00036416223,0.00029181497,0.000311584,0.000087287364,0.000047869464],"category_scores_gemma":[0.004972566,0.00019672459,0.0000217093,0.0018916925,0.00003366174,0.0008708191,0.0054685394,0.00016913122,0.000013449442],"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":[0.00014819393,0.002809662,0.12549847,0.0010641981,0.00015788876,0.000027411568,0.31707174,0.000017530398,0.0003948976,0.019040667,0.0047199586,0.5290494],"study_design_scores_gemma":[0.0015408825,0.00038834958,0.36682442,0.0005200229,0.0000873991,0.0000012118653,0.5928224,0.0007350242,0.000335075,0.0042112926,0.031435207,0.0010987299],"about_ca_topic_score_codex":0.000054623546,"about_ca_topic_score_gemma":0.0015990308,"teacher_disagreement_score":0.52795064,"about_ca_system_score_codex":0.00014725105,"about_ca_system_score_gemma":0.0002247919,"threshold_uncertainty_score":0.80221957},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97710985,0.00088953576,0.00205192,0.009303352,0.0002566038,0.0007403863,0.009503441,0.000037135735,0.00010778686],"genre_scores_gemma":[0.9789385,0.000099192366,0.01814502,0.0002599098,0.00001516891,0.000009316887,0.0017568694,0.000009039199,0.00076698064],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9976369,0.000044819986,0.00019837862,0.0017907402,0.00016344877,0.00016570064],"domain_scores_gemma":[0.98995715,0.000044924953,0.00011040975,0.0097200265,0.00009863014,0.00006886664],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0017389944,0.000074910815,0.00016230528,0.000013965857,0.00027861982,0.00038245367,0.0020976078,0.00001643681,0.000017623679],"category_scores_gemma":[0.0016168478,0.000051979492,0.000009079192,0.00034067745,0.00087198993,0.00073388603,0.014951757,0.00011193956,0.0000011823142],"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":[0.00120456,0.001297933,0.23418726,0.0011080113,0.00015555492,0.00006166913,0.0009659761,0.0000049692617,0.22136533,0.0059276978,0.41619608,0.11752496],"study_design_scores_gemma":[0.000780884,0.00017075146,0.5823219,0.00032497212,0.00019970427,0.00010520746,0.0009842508,0.0027938688,0.1386297,0.008398251,0.2650311,0.00025944266],"about_ca_topic_score_codex":0.000051584786,"about_ca_topic_score_gemma":0.00013941227,"teacher_disagreement_score":0.3481346,"about_ca_system_score_codex":0.000006068723,"about_ca_system_score_gemma":0.00010308314,"threshold_uncertainty_score":0.9930151},"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":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10178874,0.04110595,0.36174482,0.3896334,0.0041512917,0.0009190433,0.0004894886,0.0002650615,0.099902235],"genre_scores_gemma":[0.7713932,0.01157275,0.09768484,0.113521114,0.0005495968,0.000024403074,0.00040949776,0.000015130765,0.0048294906],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99370533,0.00044398993,0.0010227655,0.00051419885,0.0040747873,0.00023893091],"domain_scores_gemma":[0.9841768,0.011592497,0.0004990475,0.0015908151,0.0018600888,0.0002807164],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.008201976,0.0001204227,0.00015709316,0.0055031125,0.0005392444,0.016095288,0.0018058986,0.00008850852,0.00034913604],"category_scores_gemma":[0.15154636,0.00008798307,0.00005004196,0.029915923,0.00012851205,0.48290467,0.0018683743,0.00024342665,0.00017931729],"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":[0.00068007753,0.0002813024,0.074622974,0.00012424069,0.000069279624,0.00036965095,0.00015281893,0.0015865185,0.0010089815,0.16138075,0.21983895,0.53988445],"study_design_scores_gemma":[0.00025325117,0.000053307886,0.030160783,0.00010156545,0.000013976898,0.0002283538,0.00043519004,0.037257478,0.0043665664,0.00039521707,0.92657125,0.00016308812],"about_ca_topic_score_codex":0.000012303627,"about_ca_topic_score_gemma":7.4854064e-8,"teacher_disagreement_score":0.7067323,"about_ca_system_score_codex":0.000017406584,"about_ca_system_score_gemma":0.0011005518,"threshold_uncertainty_score":0.9907037},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.89522845,0.09905679,0.0027238121,0.0016106012,0.00081218226,0.000097235315,0.000018606259,0.0000079063475,0.00044439948],"genre_scores_gemma":[0.9930813,0.0058374656,0.00016456631,0.000190147,0.00057604734,0.0000072227426,0.0000039496063,0.000020131863,0.00011917207],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9921703,0.0024971173,0.0020113494,0.0007348409,0.0020799383,0.0005065043],"domain_scores_gemma":[0.9920282,0.004248946,0.0009508184,0.0005985888,0.0017307445,0.00044274557],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.042738993,0.00025565329,0.0029687341,0.002039809,0.00032268686,0.009215571,0.0016945157,0.0002382838,0.000109012064],"category_scores_gemma":[0.01762059,0.00020516667,0.0021113933,0.0051616654,0.00024766917,0.009203007,0.0012435083,0.00076185894,0.000002920984],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000059460785,0.000021333892,0.83656895,0.00001531325,0.0025434338,0.0007337131,0.0001414011,0.0014613731,0.00003574588,0.000009764427,0.006170948,0.15223856],"study_design_scores_gemma":[0.0002860825,0.000005895681,0.9686421,0.000143627,0.00076708215,0.0013162827,0.00077172404,0.00022679975,0.00028333734,0.0021316593,0.02513632,0.0002890996],"about_ca_topic_score_codex":0.00008276211,"about_ca_topic_score_gemma":0.00002394779,"teacher_disagreement_score":0.15194947,"about_ca_system_score_codex":0.000074285526,"about_ca_system_score_gemma":0.0012542983,"threshold_uncertainty_score":0.99181294},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07451718,0.0071849725,0.18746822,0.027611816,0.00035840453,0.00038699553,0.00025236805,0.0014151382,0.7008049],"genre_scores_gemma":[0.8945523,0.003733162,0.024959905,0.0010269876,0.0005948782,9.273161e-8,0.0008345962,0.0015022869,0.072795786],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9967065,0.0016955258,0.00023022012,0.00045519447,0.0003012155,0.0006113731],"domain_scores_gemma":[0.99778974,0.00026575528,0.00007944053,0.00042887367,0.0010036308,0.00043258537],"candidate_categories":["sts","scholarly_communication","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00153271,0.00018848511,0.0001564109,0.000095572075,0.0016203866,0.0057319812,0.0008085459,0.00024027692,0.0032772392],"category_scores_gemma":[0.00531221,0.00018678168,0.00007791848,0.0007498578,0.00031346246,0.0016654435,0.0012821261,0.00058493234,0.0018705246],"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":[0.000035891208,0.00040362013,0.0000115345265,0.00023337094,0.000054580083,0.00052330433,0.013384225,0.0007411894,0.02019161,0.16124967,0.1423598,0.6608112],"study_design_scores_gemma":[0.0003097667,0.00016997056,0.0021327771,0.00028034375,0.000016823398,0.0022047805,0.00044524777,0.008160505,0.006033526,0.011304574,0.96868783,0.0002538655],"about_ca_topic_score_codex":0.00035887613,"about_ca_topic_score_gemma":0.000021272084,"teacher_disagreement_score":0.82632804,"about_ca_system_score_codex":0.00057658495,"about_ca_system_score_gemma":0.00004901403,"threshold_uncertainty_score":0.9996794},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97955996,0.0028918574,0.000010860537,0.00735698,0.0043469165,0.00050171826,0.000024112876,0.00003651985,0.005271085],"genre_scores_gemma":[0.99729705,0.00017122112,0.00001936136,0.00046977925,0.0005962242,0.0008732753,0.000004423561,0.000006109338,0.0005625862],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.99725807,0.0004548962,0.00038662818,0.0005364816,0.00090506475,0.00045885186],"domain_scores_gemma":[0.9989149,0.00034573083,0.00015782613,0.0003456103,0.000044171436,0.00019178064],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0061103227,0.00009180922,0.00019154866,0.0002753098,0.00016259323,0.00040821987,0.0012085168,0.000055347096,0.00025195422],"category_scores_gemma":[0.0014792089,0.00007384933,0.000037598846,0.00082672975,0.000052088926,0.0007811309,0.0015649631,0.00073165394,0.000011204995],"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":[0.000025126436,0.000022806262,0.32807785,0.0000011981285,0.0000024604726,6.8128134e-7,0.0008042786,0.00012607848,4.8924403e-7,0.0005385735,0.01749749,0.65290296],"study_design_scores_gemma":[0.00018940501,0.0000033243964,0.51559275,4.090656e-7,0.0000011310465,0.0000026131784,0.0019801895,0.00009556533,0.000002374711,0.007141616,0.47493422,0.00005644003],"about_ca_topic_score_codex":0.0001241176,"about_ca_topic_score_gemma":0.000059842423,"teacher_disagreement_score":0.6528465,"about_ca_system_score_codex":0.00012404931,"about_ca_system_score_gemma":0.00006376179,"threshold_uncertainty_score":0.39364737},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9832828,0.001343727,0.00078139856,0.010108868,0.000135017,0.001217469,0.0030639993,0.0000035845067,0.00006311195],"genre_scores_gemma":[0.9325738,0.005089524,0.050994046,0.0024160896,0.00028012594,0.0000068790264,0.008433265,0.00007917337,0.00012709954],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9960801,0.00096994994,0.0011066805,0.0010387265,0.00037735936,0.0004271732],"domain_scores_gemma":[0.9949785,0.0011093298,0.00055483327,0.0029145665,0.00014014014,0.00030264232],"candidate_categories":["open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.005061387,0.000267862,0.0012223129,0.00018401118,0.000103197155,0.0006083514,0.008614474,0.00010199818,0.0001627782],"category_scores_gemma":[0.0022522206,0.00021005515,0.000015063029,0.00024098062,0.0005223588,0.005052123,0.08488541,0.0010324416,0.0000045087036],"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":[0.003365574,0.00041775068,0.8841828,0.00011461942,0.00058216555,0.0033938154,0.039573763,0.00005798732,0.010529749,0.000016998125,0.0056336396,0.052131105],"study_design_scores_gemma":[0.010865361,0.00064845645,0.92027426,0.0010986804,0.00024261634,0.0023094937,0.012031584,0.019293351,0.00007712447,0.0005321213,0.03226121,0.0003657428],"about_ca_topic_score_codex":0.018734302,"about_ca_topic_score_gemma":0.04686787,"teacher_disagreement_score":0.07627094,"about_ca_system_score_codex":0.00003860974,"about_ca_system_score_gemma":0.0005216906,"threshold_uncertainty_score":0.9967494},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07882577,0.00035236002,0.00005067971,0.0040389257,0.0002998733,0.00011339669,0.000045620174,0.000178385,0.91609496],"genre_scores_gemma":[0.76905805,0.00020011366,0.00039248893,0.00017317216,0.000118880016,1.7784971e-7,0.000032264958,0.000017494787,0.23000735],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99865127,0.00032041062,0.00013019104,0.0003532426,0.00021740873,0.00032747615],"domain_scores_gemma":[0.99859405,0.00011219424,0.0002905831,0.0005708428,0.00016967021,0.00026264624],"candidate_categories":["sts","scholarly_communication","insufficient_payload"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.0006042184,0.00018673774,0.0002466217,0.0009107008,0.0019335636,0.009719148,0.0041176137,0.0001392685,0.002975842],"category_scores_gemma":[0.00008269023,0.00020387057,0.00008458962,0.0005947723,0.00027968021,0.017399767,0.0031016956,0.0005545229,0.000026682948],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00039116593,0.00020660592,0.025892055,0.000018120554,0.00022052982,0.00010553836,0.07291286,0.0000056797858,0.000024205376,0.6212194,0.26431096,0.014692897],"study_design_scores_gemma":[0.0007512498,0.000042369305,0.011881637,0.00006325627,0.00007422785,0.0000062067356,0.01894024,0.000023111623,0.000011362493,0.00056890777,0.96734643,0.00029099872],"about_ca_topic_score_codex":0.008373734,"about_ca_topic_score_gemma":0.018551232,"teacher_disagreement_score":0.7030355,"about_ca_system_score_codex":0.00026407398,"about_ca_system_score_gemma":0.00009793323,"threshold_uncertainty_score":0.9993658},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","scholarly_communication"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"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":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8660708,0.055705592,0.06718083,0.00081647385,0.00051485095,0.007732608,0.0016960407,0.0000087231665,0.0002740881],"genre_scores_gemma":[0.9809117,0.0033726671,0.010462924,0.00011450597,0.00002448213,0.0004661801,0.00053114264,0.000025971365,0.004090433],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9018013,0.047882624,0.042569306,0.0030287488,0.00436507,0.0003529776],"domain_scores_gemma":[0.9442109,0.0071327505,0.02782229,0.01865411,0.0019140015,0.0002659716],"candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.27804184,0.0005935515,0.016301151,0.0011819649,0.00007802321,0.0012667661,0.004693747,0.00019317715,0.0077818297],"category_scores_gemma":[0.18134187,0.00027685525,0.0036612644,0.0030354045,0.00012565225,0.00044778205,0.0035736593,0.00045125233,0.000116759955],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000031598538,0.00071794953,0.752382,0.22548775,0.008817143,0.00006766558,0.001990457,0.00016707306,0.00009808563,0.000039497627,0.005161421,0.0050393967],"study_design_scores_gemma":[0.001955005,0.00019828165,0.4172613,0.1702544,0.07282068,0.00004482669,0.023414345,0.28727308,0.00005694414,0.0022916982,0.021775493,0.002653938],"about_ca_topic_score_codex":0.00016862221,"about_ca_topic_score_gemma":0.0057583316,"teacher_disagreement_score":0.33512065,"about_ca_system_score_codex":0.00009976733,"about_ca_system_score_gemma":0.00020491109,"threshold_uncertainty_score":0.99996835},"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":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08038446,0.00025505573,0.88781005,0.0017246252,0.00040171115,0.0011864078,0.00036772026,0.0003225991,0.02754734],"genre_scores_gemma":[0.5710574,0.0044277124,0.4180894,0.0005148709,0.00006367437,0.00019124319,0.0049860193,0.000035966288,0.0006337243],"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99685216,0.00021899964,0.00040027066,0.0014318271,0.00081124116,0.00028547383],"domain_scores_gemma":[0.9880226,0.00013318093,0.0003093552,0.011418623,0.00003800708,0.000078193734],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.0017964349,0.00014574736,0.00023706524,0.0008866884,0.000045365556,0.0016079664,0.03574179,0.000033570337,0.00015454332],"category_scores_gemma":[0.00022078496,0.00014801776,0.000031077772,0.0016423505,0.000055817716,0.028728232,0.100694045,0.0008112019,0.0000051229395],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000042710068,0.001398733,0.6292022,0.0011703839,0.00034939934,0.00036225902,0.0018647027,0.032993898,0.000035245914,0.09927502,0.0027311032,0.23057437],"study_design_scores_gemma":[0.00024762485,0.00016003438,0.18708578,0.000038327766,0.00001474249,0.0000011413302,0.00030160704,0.6457174,0.0000045374504,0.0015386496,0.16435957,0.0005305788],"about_ca_topic_score_codex":0.0050916346,"about_ca_topic_score_gemma":0.0027133224,"teacher_disagreement_score":0.6127235,"about_ca_system_score_codex":0.000032044358,"about_ca_system_score_gemma":0.00014524012,"threshold_uncertainty_score":0.99942845},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics"],"domain":"reproducibility","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["open_science"],"domain":null,"study_design":"not_applicable","genre":"other","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9580271,0.0068989126,0.000097163924,0.024147283,0.00054346386,0.0002205699,0.00004140537,0.0000014970931,0.010022581],"genre_scores_gemma":[0.9973512,0.00037073778,0.0005824177,0.0009847975,0.00014937764,0.0000045087886,2.4642253e-7,0.000009028008,0.00054764416],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9829064,0.0013924032,0.0015311828,0.0003054855,0.013502398,0.00036214865],"domain_scores_gemma":[0.97244596,0.010717818,0.0024198296,0.00080149865,0.01314928,0.00046559505],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"category_scores_codex":[0.12039622,0.00012000133,0.0004404977,0.015015588,0.0008079619,0.216302,0.019032039,0.000094338255,0.0011703243],"category_scores_gemma":[0.1552121,0.000055208664,0.00018268565,0.08051053,0.0001526422,0.09362901,0.011727584,0.0033890556,0.0000018261685],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00027710426,0.00022826422,0.6778256,0.000013697048,0.00014606547,0.00012481232,0.0012676545,0.00050249166,0.0014064991,0.0046660216,0.10257793,0.21096386],"study_design_scores_gemma":[0.00068310683,0.00023768206,0.6800468,0.000072653,0.000039302344,0.0011920548,0.024517559,0.000095315176,0.00027923242,0.02583803,0.26684028,0.00015795062],"about_ca_topic_score_codex":0.00046985756,"about_ca_topic_score_gemma":0.000030283374,"teacher_disagreement_score":0.21080591,"about_ca_system_score_codex":0.00009742372,"about_ca_system_score_gemma":0.00097531226,"threshold_uncertainty_score":0.99974275},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9960725,0.0028753458,0.0000040141567,0.00008671245,0.00008574071,0.000285839,0.00036761924,0.0000014578751,0.00022079096],"genre_scores_gemma":[0.9893018,0.00007809859,0.009391581,0.00044684592,0.00008282398,0.00021389418,0.00044783272,0.000012160279,0.000024984485],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9988355,0.00025361942,0.00036001366,0.00027848515,0.00010278193,0.00016962111],"domain_scores_gemma":[0.99941087,0.000060352308,0.0001728255,0.00028240486,0.0000398068,0.000033749395],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004124662,0.000094524854,0.000204466,0.000051554278,0.00007977407,0.000009673917,0.0007023058,0.0000651951,0.00017374699],"category_scores_gemma":[0.000056279623,0.00008042415,0.000019824533,0.00013400326,0.000051654257,0.0000053515555,0.0010335958,0.00009453764,9.04999e-7],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00038202747,0.0008016588,0.084254384,0.000037945993,0.000056134293,0.00014714281,0.0016128469,0.002274663,0.87937325,0.0000030807141,0.0006440468,0.030412812],"study_design_scores_gemma":[0.0053235414,0.0015423257,0.64235,0.000064483494,0.000079041296,0.000083194165,0.011200327,0.0010354802,0.19389996,0.00018837754,0.14367707,0.00055615953],"about_ca_topic_score_codex":0.008869165,"about_ca_topic_score_gemma":0.009501156,"teacher_disagreement_score":0.6854733,"about_ca_system_score_codex":0.000048141977,"about_ca_system_score_gemma":0.0007918713,"threshold_uncertainty_score":0.99773085},"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":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005334571,0.0011994917,0.00026273407,0.0031826433,0.0015748577,0.0016697956,0.00022948965,0.00020911405,0.9863373],"genre_scores_gemma":[0.30080548,0.2053072,0.21918803,0.020973125,0.00082003203,0.001120259,0.0008760652,0.00039346484,0.25051633],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.993808,0.00006889676,0.0013374814,0.0015399188,0.0025298486,0.00071588124],"domain_scores_gemma":[0.99698776,0.00057666056,0.00068536494,0.0011602059,0.00014520972,0.0004448031],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","open_science"],"consensus_categories":["sts","scholarly_communication"],"category_scores_codex":[0.007734748,0.00042355942,0.0005123957,0.002193803,0.002661071,0.006654501,0.0049139634,0.00007406206,0.00041866853],"category_scores_gemma":[0.001539602,0.0003626839,0.000058515747,0.0016997877,0.003017109,0.7357208,0.013329377,0.0003833708,0.000030871364],"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":[0.00006432781,0.000012920603,0.00073257927,0.000016951673,0.0000026480668,0.000007375827,0.0021598318,0.00045985123,0.000013459747,0.9753493,0.00013269334,0.02104804],"study_design_scores_gemma":[0.00036217645,0.0002003349,0.0010798891,0.00012835435,0.0000058735627,0.000027366097,0.006034662,0.0006931417,0.00018192457,0.0029486849,0.98783034,0.0005072349],"about_ca_topic_score_codex":0.0000050954022,"about_ca_topic_score_gemma":0.000010254386,"teacher_disagreement_score":0.98769766,"about_ca_system_score_codex":0.00013578181,"about_ca_system_score_gemma":0.00054244814,"threshold_uncertainty_score":0.9998825},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[9.506371e-7,0.9815127,0.008255969,0.0026971174,0.00034665703,0.00629487,0.000059774266,0.0000037385655,0.0008282163],"genre_scores_gemma":[0.0000012560212,0.95471466,0.039340843,0.0006405404,0.000046515837,0.0009445004,0.000004375103,0.000022314529,0.004285029],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.86840796,0.10799047,0.015970046,0.0036641317,0.0035797271,0.0003876392],"domain_scores_gemma":[0.7561515,0.21083781,0.012400934,0.018749703,0.0013647284,0.00049535296],"candidate_categories":["metaresearch","metaepi_broad","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.5297139,0.0006820084,0.015907617,0.00023349286,0.00053234515,0.0013653069,0.006132409,0.00011211206,0.0074729896],"category_scores_gemma":[0.6484308,0.00020534938,0.0049496,0.0010613281,0.00017779587,0.0005117864,0.0044165356,0.0009775612,0.000021225798],"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":[8.4603556e-7,0.000035571997,0.00008591464,0.0852066,0.00012431845,2.4617825e-7,0.00006619717,9.4021075e-9,7.101838e-8,0.006084142,0.0049974285,0.90339863],"study_design_scores_gemma":[0.000038071725,0.000033596527,0.000017068514,0.0562207,0.0023702246,0.00004350206,0.00014113892,0.000028876384,1.7183733e-7,0.004883263,0.93599665,0.00022673527],"about_ca_topic_score_codex":0.000012133555,"about_ca_topic_score_gemma":0.0000036189622,"teacher_disagreement_score":0.9309992,"about_ca_system_score_codex":0.000041847783,"about_ca_system_score_gemma":0.00037026405,"threshold_uncertainty_score":0.9996714},"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":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2605863,0.00016052711,0.034442894,0.36918554,0.005324148,0.019957159,0.00033210087,0.00071432616,0.309297],"genre_scores_gemma":[0.92984056,0.00024683477,0.023617046,0.035268392,0.00047674478,0.0012868501,0.000032862343,0.000036778787,0.009193928],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99507487,0.0007211477,0.0011244536,0.0016590445,0.0006476324,0.0007728691],"domain_scores_gemma":[0.99706954,0.0003747615,0.0005608678,0.0010442025,0.00034656056,0.00060407777],"candidate_categories":["sts","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.008079193,0.00020122556,0.00040441952,0.00021489826,0.0015282744,0.00015288554,0.0014484623,0.00016645897,0.0016566676],"category_scores_gemma":[0.0026107077,0.00018248144,0.000031602518,0.000745245,0.00035361754,0.00052321906,0.0038308287,0.0010642015,0.0003350323],"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":[0.00022451706,0.0001968217,0.11243394,0.0107398555,0.000069943846,0.000012272978,0.14121951,0.00006209541,0.008479842,0.11323355,0.57909083,0.03423682],"study_design_scores_gemma":[0.0044579655,0.0005106724,0.31775305,0.0024216052,0.000117265525,0.000006737075,0.016367745,0.0025336342,0.0024227805,0.16430448,0.48735645,0.001747629],"about_ca_topic_score_codex":0.0017844944,"about_ca_topic_score_gemma":0.0004145621,"teacher_disagreement_score":0.66925424,"about_ca_system_score_codex":0.00027470538,"about_ca_system_score_gemma":0.0075439466,"threshold_uncertainty_score":0.9997716},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.57188374,0.060564265,0.00020907904,0.29249182,0.0073410803,0.00038948416,0.00022538916,0.000025288586,0.06686987],"genre_scores_gemma":[0.85893506,0.0032056442,0.0010127927,0.04232443,0.008315963,0.000010697907,0.00005258084,0.00014581005,0.08599699],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99629265,0.00060307194,0.0011525677,0.00025453346,0.00046976787,0.0012274131],"domain_scores_gemma":[0.99101275,0.000363198,0.00066389766,0.0004096883,0.0008978678,0.0066525987],"candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["research_integrity"],"category_scores_codex":[0.0031757161,0.00031611638,0.00072905375,0.000078901576,0.0006037428,0.000070292306,0.0006771825,0.0015826605,0.0015032283],"category_scores_gemma":[0.001913207,0.00031172857,0.00025649328,0.00035053355,0.0003790879,0.00050008454,0.000052672836,0.0052571925,0.00009264296],"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":[0.00004462192,0.00004518654,0.0022727067,0.0013354268,0.00013006091,0.0010613198,0.049855057,0.000034853372,0.0001564564,0.00015766808,0.9376043,0.0073023485],"study_design_scores_gemma":[0.0042182985,0.00009277856,0.00096589053,0.0020039699,0.00008711131,0.0005615957,0.03790944,0.000059558344,0.00039242374,0.00060511444,0.95281047,0.00029335462],"about_ca_topic_score_codex":0.021708101,"about_ca_topic_score_gemma":0.014975215,"teacher_disagreement_score":0.28705138,"about_ca_system_score_codex":0.0012673074,"about_ca_system_score_gemma":0.13103718,"threshold_uncertainty_score":0.9999335},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.5552062,0.050634168,0.0002924817,0.34278864,0.01645808,0.0005130442,0.000077257755,0.000015819569,0.034014337],"genre_scores_gemma":[0.9108906,0.0030100017,0.004146232,0.054017194,0.0034681514,0.0000045052857,0.000004238446,0.000039532493,0.024419557],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9949907,0.0015484983,0.0009784895,0.00029077643,0.00076509966,0.0014264064],"domain_scores_gemma":[0.992538,0.0005792316,0.00059005246,0.0002356003,0.0007452551,0.0053118714],"candidate_categories":["sts","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.010441759,0.00023273863,0.00057345536,0.00035935699,0.0018834753,0.00018778844,0.0009037028,0.0006099954,0.0007778919],"category_scores_gemma":[0.0021776282,0.000189931,0.00017463548,0.0011018874,0.0014832685,0.0015024131,0.00005068518,0.0026658184,0.00016929513],"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":[0.000042642878,0.00007384169,0.07347841,0.00042595193,0.000095115414,0.0009897631,0.21222728,0.00065870397,0.000009871943,0.014725966,0.6661677,0.0311048],"study_design_scores_gemma":[0.0019258758,0.0006679953,0.02949459,0.0012060861,0.000039390616,0.00023077131,0.0566427,0.00045169538,0.0000073699002,0.0015381324,0.9075737,0.00022167263],"about_ca_topic_score_codex":0.07093326,"about_ca_topic_score_gemma":0.3795933,"teacher_disagreement_score":0.35568443,"about_ca_system_score_codex":0.00009804869,"about_ca_system_score_gemma":0.15583277,"threshold_uncertainty_score":0.9996351},"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":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000018488645,0.6226917,0.0000021766245,0.00039176276,0.00009184279,0.37501833,0.000043885073,0.00003237073,0.0017260894],"genre_scores_gemma":[3.7603021e-7,0.57387614,0.0005804498,0.00041203568,0.0002325635,0.4239973,0.00045064083,0.000070855756,0.00037963502],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.89084965,0.095566265,0.0046296483,0.0014930301,0.004192188,0.0032692216],"domain_scores_gemma":[0.98331857,0.011772255,0.0012732672,0.0011503791,0.0014376334,0.0010478918],"candidate_categories":["metaresearch","metaepi_narrow","sts","research_integrity","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.15139557,0.000544741,0.003129376,0.0014512461,0.0040245806,0.000078503086,0.00061421935,0.000468263,0.00096675573],"category_scores_gemma":[0.006598445,0.0004532676,0.00021018265,0.0036711432,0.00021695852,0.00014609059,0.0024045128,0.009630601,0.00033129466],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000008753464,0.00008412996,0.00013584655,0.6044761,0.000010363152,0.00002493592,0.00060256606,7.8357075e-8,2.6437315e-8,0.00035037997,0.0023926166,0.39191422],"study_design_scores_gemma":[0.0004694892,0.0006176029,0.0000032690139,0.5943104,0.000007665149,0.0000017826532,0.0012360783,0.000011130233,3.9621177e-8,0.00010547004,0.4030603,0.00017675365],"about_ca_topic_score_codex":0.0012024099,"about_ca_topic_score_gemma":0.0010310196,"teacher_disagreement_score":0.40066767,"about_ca_system_score_codex":0.003988361,"about_ca_system_score_gemma":0.023708304,"threshold_uncertainty_score":0.9999465},"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":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.4280539,0.05518218,0.49511436,0.008542089,0.00069415825,0.00891237,0.0032585284,0.000013374381,0.00022906909],"genre_scores_gemma":[0.990086,0.0010733149,0.0070352727,0.000048722486,0.00008868679,0.00076778716,0.00010966506,0.0000089005725,0.0007816064],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9842309,0.003857582,0.0069160517,0.0010958305,0.003799295,0.00010035316],"domain_scores_gemma":[0.7891742,0.18718788,0.009627989,0.0027418525,0.011153212,0.000114849434],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.08044572,0.0002720266,0.0027801825,0.00017671552,0.00013906659,0.00047619548,0.0011735597,0.00009565249,0.0032264157],"category_scores_gemma":[0.510906,0.00009929346,0.0015910434,0.00037932865,0.00015764624,0.00029257938,0.0006244912,0.00014317037,0.000030063307],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006608564,0.00026548124,0.006226237,0.0015695654,0.006582861,5.063282e-7,0.012878063,0.00019808116,0.000088085566,0.0006627917,0.028211894,0.94325036],"study_design_scores_gemma":[0.00021940164,0.00021875497,0.13912842,0.017664563,0.026350748,0.0000019105282,0.117759936,0.59904563,0.0005979845,0.09803433,0.00042530097,0.0005530024],"about_ca_topic_score_codex":0.000053629516,"about_ca_topic_score_gemma":0.0008674036,"teacher_disagreement_score":0.94269735,"about_ca_system_score_codex":0.000029780538,"about_ca_system_score_gemma":0.00008641383,"threshold_uncertainty_score":0.9976848},"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":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.40649134,0.000028499342,0.51274604,0.0040973597,0.0011988063,0.0020617805,0.00003989646,0.00004397961,0.07329232],"genre_scores_gemma":[0.916262,0.0000020633288,0.08238243,0.00050977105,0.000053780437,0.000049578637,0.0000130972885,0.000010846171,0.0007164625],"study_design_codex":"design_other","study_design_gemma":"qualitative","domain_scores_codex":[0.9790892,0.007543263,0.0019276782,0.0035644758,0.0069665303,0.0009088327],"domain_scores_gemma":[0.9668208,0.027822226,0.00061403686,0.003068536,0.0013947393,0.00027966028],"candidate_categories":["metaresearch","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","open_science"],"category_scores_codex":[0.19541782,0.0003038361,0.00083225133,0.0019077527,0.0010319666,0.0067174323,0.026588324,0.00009184776,0.00007402017],"category_scores_gemma":[0.10696557,0.00018856191,0.00013064095,0.0051939134,0.0010123327,0.00066586584,0.07510582,0.0025305978,0.00019320424],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00022995973,0.0003327289,0.0058844173,0.000042845633,0.000045309214,0.00003724985,0.12650736,0.019728184,0.0007864307,0.033073083,0.026211675,0.78712076],"study_design_scores_gemma":[0.00053016265,0.00028697294,0.017754411,0.0016101727,0.00002479377,0.0000045812985,0.49445358,0.4543878,0.0012494201,0.025199465,0.0039080833,0.0005905686],"about_ca_topic_score_codex":0.004243039,"about_ca_topic_score_gemma":0.00042612292,"teacher_disagreement_score":0.7865302,"about_ca_system_score_codex":0.00018367147,"about_ca_system_score_gemma":0.0007951929,"threshold_uncertainty_score":0.9997706},"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":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008138657,0.015428608,0.004884259,0.68831915,0.0019870477,0.002833601,0.002992884,0.00018043928,0.28256014],"genre_scores_gemma":[0.3863692,0.1944148,0.15487966,0.031339336,0.0041598734,0.001224394,0.025749343,0.0005835658,0.20127982],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9188565,0.07625958,0.0021346838,0.0011002527,0.00038872886,0.0012602279],"domain_scores_gemma":[0.9243685,0.016589012,0.0044193603,0.052299704,0.0016160841,0.0007073631],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","open_science","research_integrity"],"category_scores_codex":[0.07088956,0.0006212343,0.0013042398,0.00020120983,0.009018118,0.0036462755,0.058667682,0.0021957105,0.0007473893],"category_scores_gemma":[0.04019186,0.00072585006,0.00013583877,0.00037241526,0.001052816,0.0102768,0.10766323,0.008511996,0.00072014134],"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":[0.00006582451,0.000702363,0.0111446995,0.0013601303,0.00024577836,0.000008997576,0.016950995,6.914243e-7,0.000016160146,0.2442877,0.4878331,0.23738353],"study_design_scores_gemma":[0.0010856979,0.000028229651,0.013114601,0.0038029745,0.00012460264,0.00005947287,0.0036175905,0.00049464346,0.000001601559,0.015554138,0.9614494,0.0006670443],"about_ca_topic_score_codex":0.12949137,"about_ca_topic_score_gemma":0.35964295,"teacher_disagreement_score":0.6569798,"about_ca_system_score_codex":0.0008819322,"about_ca_system_score_gemma":0.015504347,"threshold_uncertainty_score":0.9995192},"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.10466748,0.00028340946,0.09574472,0.017522115,0.011438303,0.009417011,0.7517484,0.0005270113,0.008651554],"genre_scores_gemma":[0.066362396,0.0000126285695,0.0368044,0.009346709,0.00036113942,0.0003702934,0.7949009,0.0001288231,0.0917127],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9967853,0.00008566306,0.00032709938,0.0014066756,0.000941292,0.00045395966],"domain_scores_gemma":[0.996333,0.000029221621,0.00013387768,0.003238953,0.00005859559,0.00020635407],"candidate_categories":["open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0019867155,0.00016999732,0.00020640575,0.00026888808,0.00074079086,0.0005445679,0.0023440702,0.000016627138,0.0009140257],"category_scores_gemma":[0.00016500558,0.00017013693,0.000025166915,0.000814677,0.00006227687,0.0006860632,0.016778458,0.00027092878,0.00028981402],"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":[0.000052697047,0.00019195583,0.00019417708,0.000022999518,0.000031015974,0.000059292637,0.00017920817,0.00002812342,0.067858,0.000007394835,0.90597534,0.025399797],"study_design_scores_gemma":[0.0014840504,0.00010680331,0.000158371,0.000016298207,0.00010074848,0.00003645288,0.0031055855,0.01720556,0.0038229171,0.000002379175,0.9737935,0.00016732133],"about_ca_topic_score_codex":0.00013711848,"about_ca_topic_score_gemma":0.000013581286,"teacher_disagreement_score":0.083061144,"about_ca_system_score_codex":0.00016080227,"about_ca_system_score_gemma":0.00013170391,"threshold_uncertainty_score":0.9999993},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9632468,0.0001104228,0.00003289734,0.0034968767,0.0006489969,0.0042603714,0.000040856783,0.00009593408,0.028066834],"genre_scores_gemma":[0.9974392,0.00012365039,0.000998456,0.000045869718,0.00012238164,0.0009022333,0.00001043594,0.00002646023,0.00033130142],"study_design_codex":"observational","study_design_gemma":"qualitative","domain_scores_codex":[0.9919294,0.0014742322,0.000807026,0.001185938,0.0038697228,0.0007336857],"domain_scores_gemma":[0.9949158,0.0016774974,0.0013126313,0.0013457804,0.00052036543,0.00022794397],"candidate_categories":["metaresearch","open_science"],"consensus_categories":["metaresearch","open_science"],"category_scores_codex":[0.04910762,0.00022676423,0.00060218415,0.0015546504,0.00064513314,0.00084965944,0.013754592,0.00017915996,0.0004072177],"category_scores_gemma":[0.029792558,0.00022511912,0.00006119716,0.004815175,0.0017431662,0.0010109438,0.017068373,0.0017406007,0.0000033511594],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024020867,0.0016111142,0.7672553,0.0003093462,0.00007041652,0.00059129193,0.21255837,0.0004351295,0.0024137546,0.0057034,0.0020899961,0.006721648],"study_design_scores_gemma":[0.0024220368,0.0011739308,0.22271651,0.0011293463,0.00013533434,0.000007629951,0.70586,0.0036314582,0.00045283863,0.033821613,0.02703319,0.001616122],"about_ca_topic_score_codex":0.14347291,"about_ca_topic_score_gemma":0.017526155,"teacher_disagreement_score":0.5445388,"about_ca_system_score_codex":0.0007132867,"about_ca_system_score_gemma":0.004919831,"threshold_uncertainty_score":0.99158144},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.79718626,0.0004552088,0.00010127779,0.19750631,0.0035376186,0.00012186594,0.000014943454,0.0000020997934,0.0010744445],"genre_scores_gemma":[0.9709559,0.00123163,0.000017850647,0.026991755,0.00078441517,0.0000010805803,0.0000037064508,0.0000031449044,0.000010507839],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.8009734,0.0010856015,0.00088792416,0.00012712146,0.19670783,0.00021808466],"domain_scores_gemma":[0.9978026,0.0006421342,0.00076883874,0.000081320984,0.0005323456,0.00017276446],"candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.14540039,0.00010285506,0.0002796738,0.00041068095,0.0006796742,0.001748092,0.097222574,0.0001224758,0.0008235837],"category_scores_gemma":[0.0027188987,0.00006985796,0.00014095812,0.00036205692,0.00045142256,0.00050496083,0.16804832,0.0012763579,8.932978e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00033339125,0.002242685,0.7332306,0.000009650057,0.0005932833,0.0045503974,0.18404837,0.000022704497,0.0000033195515,0.009559078,0.02596004,0.039446473],"study_design_scores_gemma":[0.01374422,0.0008667953,0.56708664,0.0011420681,0.00023532097,0.0011015316,0.29092205,0.000029305977,0.0000030118267,0.068182476,0.05644721,0.00023939474],"about_ca_topic_score_codex":0.0015570191,"about_ca_topic_score_gemma":0.00244238,"teacher_disagreement_score":0.19682917,"about_ca_system_score_codex":0.00042845288,"about_ca_system_score_gemma":0.0005763259,"threshold_uncertainty_score":0.9992882},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.98193336,0.0005344467,0.00008430384,0.004769085,0.00040024484,0.00016189086,0.00009376124,5.991246e-7,0.012022306],"genre_scores_gemma":[0.99910307,0.000023971676,0.00010683007,0.0006566396,0.0000063825137,0.0000017223244,4.201792e-7,8.2815836e-7,0.000100108766],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9941161,0.00010755877,0.00074021047,0.000086455286,0.0047366074,0.00021301521],"domain_scores_gemma":[0.99732393,0.00060004863,0.0005658047,0.00013586585,0.00095707225,0.00041727207],"candidate_categories":["bibliometrics","scholarly_communication","open_science"],"consensus_categories":["bibliometrics","scholarly_communication"],"category_scores_codex":[0.014976857,0.000039901646,0.00011642354,0.019613521,0.00050581526,0.0043733413,0.008083475,0.0000074431005,0.000798042],"category_scores_gemma":[0.0065942663,0.000025645973,0.000017949405,0.06342222,0.00021942589,0.01602819,0.0011283635,0.00021900532,0.0000010910301],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00000891439,0.0000044839517,0.9500354,0.0000068296604,0.0000012571198,0.0000026974772,0.00069537986,0.004087092,0.00000498595,0.002925512,0.015519097,0.026708364],"study_design_scores_gemma":[0.00014803381,0.0000319096,0.9424856,0.000008006652,5.09194e-7,0.000026243235,0.0021805211,0.011115415,0.00008972547,0.0013207501,0.042552426,0.000040913925],"about_ca_topic_score_codex":0.3814228,"about_ca_topic_score_gemma":0.28490916,"teacher_disagreement_score":0.096513614,"about_ca_system_score_codex":0.0003702256,"about_ca_system_score_gemma":0.042043403,"threshold_uncertainty_score":0.9977341},"labels":[{"model":"gemma","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science","scholarly_communication"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":true,"confidence":"medium"}],"label_agreement":"split"},{"id":"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.73051745,0.00015970692,0.05343706,0.07543147,0.0033939613,0.009927975,0.000677959,0.0027539872,0.12370043],"genre_scores_gemma":[0.96925575,0.00010594845,0.012964867,0.006987685,0.0019618506,0.0000012728307,0.0011919844,0.0006404848,0.006890171],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99758905,0.00038555433,0.00043323438,0.00058498536,0.0006336756,0.00037347816],"domain_scores_gemma":[0.99745125,0.00011562846,0.00026158543,0.00090456835,0.001081803,0.00018516573],"candidate_categories":["sts","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.003890179,0.00015817265,0.00014839077,0.00020362229,0.0073832027,0.083934255,0.006632553,0.000085062886,0.0006289671],"category_scores_gemma":[0.0038300296,0.00008838326,0.00002054139,0.0018226808,0.00032598988,0.06207376,0.011579658,0.00036693766,0.00032319856],"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":[0.00054674165,0.00009333663,0.000040083047,0.00008884511,0.00006647149,0.000016376593,0.053007767,0.00002862412,0.003178537,0.0037667896,0.8271848,0.11198166],"study_design_scores_gemma":[0.00062949385,0.0003856935,0.0032005636,0.00006959944,0.0000040806535,0.00046926006,0.0045427657,0.015434687,0.000185333,0.0000953272,0.9747953,0.00018790433],"about_ca_topic_score_codex":0.0002656725,"about_ca_topic_score_gemma":0.000008691878,"teacher_disagreement_score":0.2387383,"about_ca_system_score_codex":0.00008069882,"about_ca_system_score_gemma":0.00007688246,"threshold_uncertainty_score":0.99874204},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9889781,0.00044185933,0.000005147369,0.0063891304,0.0006459115,0.00084880943,0.00034957362,0.000015115865,0.0023263593],"genre_scores_gemma":[0.9966982,0.0014748081,0.00024697988,0.0002937009,0.000108438355,0.00005699271,0.00016320801,0.000015115514,0.0009425456],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9946779,0.00034811115,0.0005506441,0.0019893318,0.0021376566,0.00029634667],"domain_scores_gemma":[0.98074555,0.013558146,0.0008036204,0.0035246573,0.0010796691,0.00028833255],"candidate_categories":["metaresearch","sts","open_science","research_integrity"],"consensus_categories":["metaresearch","open_science"],"category_scores_codex":[0.051197764,0.00016558019,0.00039804683,0.00009357105,0.0006131484,0.0005518051,0.005785931,0.00027557206,0.00036704587],"category_scores_gemma":[0.23725623,0.0001101935,0.000028597075,0.00034985872,0.00478603,0.00040101248,0.104340106,0.00407281,0.000003105034],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044555933,0.0001410961,0.9980718,0.00017538946,0.00004555433,0.00006574608,0.0000707892,0.00002122338,0.00006152448,0.00041916777,0.00010130394,0.00038086984],"study_design_scores_gemma":[0.00044617857,0.00032010977,0.9941644,0.00004456891,0.000027164888,0.00003694269,0.000011716909,0.0019547518,0.0000108058475,0.0019269668,0.00095914106,0.00009722536],"about_ca_topic_score_codex":0.013006058,"about_ca_topic_score_gemma":0.010526363,"teacher_disagreement_score":0.18605846,"about_ca_system_score_codex":0.00008524946,"about_ca_system_score_gemma":0.0021625045,"threshold_uncertainty_score":0.99959326},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9176459,0.00023323366,0.000044519995,0.03689743,0.0006082269,0.0008180629,0.0003311945,0.000020611316,0.043400854],"genre_scores_gemma":[0.9952937,0.0000671587,0.00017981342,0.00086191215,0.00008153009,0.000117790754,0.0000017112254,0.000019388488,0.0033770306],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.988986,0.00015446624,0.0010902293,0.00077551446,0.008091196,0.00090255466],"domain_scores_gemma":[0.6683719,0.0094945,0.0039526387,0.0021828942,0.31460696,0.0013910844],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.027384568,0.00024336223,0.0005475616,0.008188741,0.0014231704,0.18672621,0.049079955,0.00009682442,0.00032838449],"category_scores_gemma":[0.39658353,0.00014141992,0.00024202692,0.06696718,0.000481455,0.044111717,0.039861206,0.0007295772,0.000012455159],"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":[0.00038983824,0.00029928747,0.51587284,0.000037206304,0.00009056764,0.0000028794052,0.019586805,0.00005169081,0.0025936821,0.064830355,0.29292247,0.10332239],"study_design_scores_gemma":[0.00044976277,0.0003484255,0.27894258,0.000030636962,0.000016200322,0.000013755645,0.0075273574,0.00044646318,0.0059429593,0.04936454,0.6566427,0.00027467377],"about_ca_topic_score_codex":0.0014109425,"about_ca_topic_score_gemma":0.000069287365,"teacher_disagreement_score":0.36919898,"about_ca_system_score_codex":0.0001805284,"about_ca_system_score_gemma":0.0009618522,"threshold_uncertainty_score":0.99987686},"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":"evaluation","model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.946228,0.00019414948,0.0000047915405,0.011630291,0.000304118,0.00062173116,0.0004821424,0.000016094958,0.04051868],"genre_scores_gemma":[0.99740434,0.000057527453,0.00013016416,0.0003897116,0.00005239371,0.000053303746,0.000005548153,0.000022863474,0.0018841395],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9878257,0.00026653352,0.0010579552,0.00088120287,0.008811504,0.0011570993],"domain_scores_gemma":[0.4715041,0.011910197,0.010001197,0.0020966132,0.5014367,0.0030511643],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"category_scores_codex":[0.03238912,0.00027473652,0.00063678843,0.01848057,0.0012284273,0.17360933,0.0383428,0.00016002856,0.00079328235],"category_scores_gemma":[0.6587914,0.00018832224,0.000219428,0.06823122,0.0005159641,0.09840728,0.021599013,0.001267248,0.0000050333815],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007258553,0.00010494194,0.9278254,0.000033935215,0.000055956825,0.0000032105036,0.017045682,0.000022072358,0.0017312868,0.0036763723,0.037544973,0.011883606],"study_design_scores_gemma":[0.00044786616,0.00027853952,0.6672092,0.00005260625,0.00002078585,0.000012035415,0.035343338,0.0006691838,0.0048020985,0.010800492,0.2800452,0.00031866762],"about_ca_topic_score_codex":0.16000754,"about_ca_topic_score_gemma":0.007085081,"teacher_disagreement_score":0.6264023,"about_ca_system_score_codex":0.00061817555,"about_ca_system_score_gemma":0.0046151006,"threshold_uncertainty_score":0.99635905},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9289218,0.000055487526,0.00025307309,0.0035135874,0.0006270278,0.000637174,0.00006555517,0.000015884283,0.065910414],"genre_scores_gemma":[0.9179739,0.00008678402,0.000034233868,0.000096276824,0.00027871062,0.00000229383,0.000002227934,0.00002205456,0.08150352],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98936415,0.0010139847,0.00042184882,0.0013749577,0.005640888,0.0021841899],"domain_scores_gemma":[0.9901372,0.0017493857,0.00030627675,0.0010565328,0.0036479717,0.0031026371],"candidate_categories":["bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["bibliometrics"],"category_scores_codex":[0.008500346,0.00027550763,0.000458903,0.04005377,0.004204301,0.024128351,0.010518026,0.00017003922,0.00027450605],"category_scores_gemma":[0.0036856392,0.00022566406,0.00009761263,0.09715174,0.0023350623,0.0053581665,0.0038710928,0.001195173,0.00008961201],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00026846922,0.00009318191,0.9021432,0.000010982027,0.000061950814,0.0011592173,0.00019400638,0.0000033672109,0.0042910664,0.000259351,0.08823765,0.003277591],"study_design_scores_gemma":[0.0008442884,0.000769649,0.8986996,0.000047417067,0.00000949922,0.00022494174,0.0008847358,0.00011125845,0.005696054,0.0006508327,0.09175076,0.0003109933],"about_ca_topic_score_codex":0.38666764,"about_ca_topic_score_gemma":0.3215717,"teacher_disagreement_score":0.06509593,"about_ca_system_score_codex":0.0014679633,"about_ca_system_score_gemma":0.0034258387,"threshold_uncertainty_score":0.99709207},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"},{"model":"gpt","categories":["metaresearch","bibliometrics","scholarly_communication","open_science"],"domain":"incentives","study_design":"observational","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"low"}],"label_agreement":"split"},{"id":"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.07491712,0.0065342635,0.47169992,0.0003089657,0.004777917,0.021865757,0.048974935,0.007507854,0.36341327],"genre_scores_gemma":[0.9639256,0.00030639055,0.0132331615,0.00003759809,0.00063944753,0.0020496615,0.01794254,0.00041024855,0.0014553607],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99179286,0.0018080452,0.0009315168,0.0019872172,0.0020637554,0.0014165906],"domain_scores_gemma":[0.9925668,0.00006560665,0.00016851313,0.005665722,0.0009776988,0.00055564556],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.004844289,0.0007047547,0.0009884505,0.000274058,0.0007360742,0.003541842,0.017190687,0.0007796911,0.00018421888],"category_scores_gemma":[0.0002600814,0.0007536719,0.00008836018,0.0022562118,0.000061870574,0.0011949876,0.032451328,0.002260809,0.00008641542],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00006909363,0.00011121095,0.00021567295,0.00091428694,0.00021314132,0.000058408437,0.00006675579,0.94651884,0.000049853217,0.009062228,0.041087832,0.0016326964],"study_design_scores_gemma":[0.00048395654,0.00011331671,0.000054921227,0.0005047837,0.000031344014,0.00002695525,0.00053770206,0.9842532,0.0000685092,0.00024726946,0.012966413,0.0007115993],"about_ca_topic_score_codex":0.018031634,"about_ca_topic_score_gemma":0.0027692013,"teacher_disagreement_score":0.88900846,"about_ca_system_score_codex":0.008435898,"about_ca_system_score_gemma":0.002030061,"threshold_uncertainty_score":0.99949145},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25071985,0.00048895535,0.0001951136,0.0011883965,0.0011129761,0.0003484203,0.00010696595,0.00014263744,0.7456967],"genre_scores_gemma":[0.8476296,0.00008096775,0.000818221,0.00067480607,0.0003808999,0.000001842469,0.00018168826,0.000026614214,0.15020536],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987874,0.00007035403,0.00028062356,0.00025894103,0.00036351613,0.00023915744],"domain_scores_gemma":[0.9986766,0.00004257646,0.0004071812,0.00035950687,0.00031677927,0.00019736393],"candidate_categories":["sts","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00014061214,0.0001926017,0.00033705914,0.00011051291,0.0013595783,0.0023824428,0.0018875394,0.00007779235,0.002418498],"category_scores_gemma":[0.000009710429,0.00019366045,0.00009258757,0.00014229825,0.00027653156,0.008432803,0.0019338375,0.00012173987,0.00009501158],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0003972305,0.0002619496,0.0019388151,0.00018340505,0.00041186198,0.000061375016,0.1385346,0.000059738133,0.000070972346,0.12275148,0.7324475,0.002881072],"study_design_scores_gemma":[0.00096894416,0.0001831247,0.0011859434,0.000042025258,0.0000616593,0.0000071715695,0.06986525,0.00024035755,0.00012837222,0.00049444305,0.9265427,0.0002800265],"about_ca_topic_score_codex":0.01258066,"about_ca_topic_score_gemma":0.0030799562,"teacher_disagreement_score":0.59690976,"about_ca_system_score_codex":0.00008468914,"about_ca_system_score_gemma":0.00024221519,"threshold_uncertainty_score":0.9999405},"labels":[{"model":"gemma","categories":["open_science","bibliometrics"],"domain":null,"study_design":"not_applicable","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9797359,0.00004012203,0.000016941445,0.00066296215,0.00012907841,0.000570018,0.018705366,0.000009823865,0.0001297909],"genre_scores_gemma":[0.98907214,0.000072375595,0.00047229647,0.0008504962,0.000024975385,0.000010168631,0.009386744,0.000020367184,0.00009043954],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99595815,0.00036549632,0.00035063556,0.00070804957,0.0018347014,0.000782961],"domain_scores_gemma":[0.9972745,0.00012104836,0.00037096912,0.0014344697,0.00020282436,0.0005961609],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0028570574,0.00014935827,0.00032222274,0.00033727268,0.00090120675,0.00028547406,0.0027419995,0.000011490333,0.0014529999],"category_scores_gemma":[0.00049174746,0.00010930303,0.00002019123,0.001281552,0.00022504393,0.0016055442,0.0024347277,0.00069991924,0.000043827524],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.008404824,0.00473583,0.6944162,0.000017032367,0.00018629867,0.00027551482,0.00008853035,0.000025323781,0.0015451256,0.000014213309,0.27174625,0.01854486],"study_design_scores_gemma":[0.0047388417,0.010359891,0.9714262,0.00004344098,0.000016773465,0.000085299514,0.00003786686,0.00033088328,0.0005484685,0.0000127575195,0.012283739,0.000115850686],"about_ca_topic_score_codex":0.00360217,"about_ca_topic_score_gemma":0.0016346319,"teacher_disagreement_score":0.27701,"about_ca_system_score_codex":0.00040260685,"about_ca_system_score_gemma":0.005009867,"threshold_uncertainty_score":0.9994598},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.7783497,0.02276781,0.0000064778533,0.015298083,0.18164387,0.000689289,0.000024970625,0.000006284223,0.0012135356],"genre_scores_gemma":[0.90153515,0.058992956,0.0013218358,0.001517779,0.03643344,0.0000057957413,0.000009852988,0.00003379834,0.00014937288],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9881379,0.0000010476571,0.0069558397,0.0010320436,0.002264091,0.0016090893],"domain_scores_gemma":[0.98180556,0.00001210205,0.01516484,0.0004004009,0.0011977183,0.0014193868],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.08958244,0.00045960248,0.0029350922,0.008203123,0.0023619502,0.006747282,0.006664795,0.00019228114,0.00016845092],"category_scores_gemma":[0.000056480556,0.00041643085,0.00020930979,0.006659719,0.0015875549,0.020273346,0.0014669754,0.0033512663,8.1689683e-7],"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":[0.000058987338,0.0001369303,0.95200735,0.00018064512,0.000009267205,0.00016845348,0.0006158272,0.0006004096,1.243778e-7,0.0055904267,0.040163636,0.00046794166],"study_design_scores_gemma":[0.0009912534,0.0005874343,0.73841625,0.00116712,0.000007578788,0.00039508074,0.0007796968,0.0006427772,8.6299417e-7,0.0026851296,0.2539992,0.00032758468],"about_ca_topic_score_codex":0.006100017,"about_ca_topic_score_gemma":0.00071584823,"teacher_disagreement_score":0.21383557,"about_ca_system_score_codex":0.0020033312,"about_ca_system_score_gemma":0.023343543,"threshold_uncertainty_score":0.99982876},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9845085,0.00007561501,0.000002727295,0.009097034,0.000066672284,0.00044471625,0.00003055118,0.000009425561,0.005764748],"genre_scores_gemma":[0.9993123,0.0000367114,0.00033726703,0.000110421315,0.000007272898,0.000058592163,6.6283513e-7,0.0000019617266,0.00013485369],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9939381,0.000031832595,0.00061811,0.0002904762,0.004754495,0.0003670017],"domain_scores_gemma":[0.986647,0.0028685096,0.0033156613,0.00010272363,0.007018161,0.000047958518],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","scholarly_communication"],"category_scores_codex":[0.055342477,0.000069432586,0.00020345261,0.019848257,0.0020194498,0.012004744,0.0047194175,0.000078439094,0.0000014675404],"category_scores_gemma":[0.20841368,0.000049711474,0.000018826531,0.078638405,0.001560307,0.026061628,0.009099861,0.0002905632,1.5427531e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000030556946,0.0000134565325,0.91181886,0.000027194677,0.000003977937,8.112961e-9,0.0041540554,0.0000031996226,0.0016777928,0.06952808,0.0010350879,0.011735207],"study_design_scores_gemma":[0.00083286065,0.00033362902,0.7160398,0.000022925211,0.000007982082,0.000003475047,0.09521636,0.00335206,0.006828561,0.16305554,0.014211025,0.0000957852],"about_ca_topic_score_codex":0.00021603183,"about_ca_topic_score_gemma":0.000022163647,"teacher_disagreement_score":0.1957791,"about_ca_system_score_codex":0.00015647139,"about_ca_system_score_gemma":0.00042697717,"threshold_uncertainty_score":0.9992798},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"reporting","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.63150877,0.0007723538,0.13858606,0.2104592,0.0021673115,0.004055316,0.00008102756,0.00007407339,0.012295901],"genre_scores_gemma":[0.96476364,0.000009202348,0.029680528,0.0017094291,0.00020438313,0.00018285483,9.921761e-8,0.000011905859,0.0034379307],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9927505,0.00020691921,0.0007431399,0.000604657,0.0039500594,0.0017447178],"domain_scores_gemma":[0.9912987,0.0007329543,0.0012715269,0.0014667549,0.0049587316,0.00027136752],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","sts","open_science"],"category_scores_codex":[0.052065942,0.00017491641,0.00025329526,0.0005399515,0.018016063,0.003535505,0.028432572,0.000023165345,0.00002428725],"category_scores_gemma":[0.010512092,0.00010743109,0.00008414185,0.015889967,0.004078682,0.0055599706,0.016570292,0.0009521368,0.0000064726887],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000023502611,0.00013751604,0.00015366069,0.000017512479,0.000013223476,0.00023475163,0.01687003,0.0017193194,0.00127797,0.9363799,0.01573143,0.027441181],"study_design_scores_gemma":[0.0012679803,0.00085632637,0.007919715,0.0000950923,0.000013401102,0.0099738045,0.03400662,0.0070122243,0.021488298,0.103647016,0.81306267,0.0006568363],"about_ca_topic_score_codex":0.0034517467,"about_ca_topic_score_gemma":0.00008520271,"teacher_disagreement_score":0.8327329,"about_ca_system_score_codex":0.0039327284,"about_ca_system_score_gemma":0.025797267,"threshold_uncertainty_score":0.999891},"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":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016880329,0.000054905,0.0019747857,0.027268203,0.006311778,0.019310135,0.0013266488,0.00032582178,0.9417397],"genre_scores_gemma":[0.00035107267,0.00013808369,0.023550665,0.007300949,0.0012676261,0.014961282,0.00044203826,0.000108580236,0.9518797],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.986947,0.0026505073,0.0022242863,0.0008630451,0.0049125827,0.002402585],"domain_scores_gemma":[0.98798543,0.0067116846,0.0007187035,0.000719287,0.003325655,0.0005392504],"candidate_categories":["metaresearch","sts","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","insufficient_payload"],"category_scores_codex":[0.057362705,0.0002548457,0.0005257,0.0008327754,0.005080249,0.00002795095,0.00085455953,0.00036448814,0.06023401],"category_scores_gemma":[0.011196502,0.00021988625,0.00007549682,0.0008694115,0.0001196621,0.000110790104,0.0010975625,0.002667932,0.0015288006],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000018350245,0.00002065262,0.0014420531,0.0025266681,0.000015820977,0.0000035650328,0.0072678397,4.0729068e-7,0.0000071244485,0.0024937708,0.98242456,0.0037791696],"study_design_scores_gemma":[0.00045163734,0.00011026644,0.0013482483,0.00009547646,0.0000040023456,0.0000013822975,0.008588205,0.0000044697463,0.000021869253,0.0001685379,0.9889815,0.00022438992],"about_ca_topic_score_codex":0.003174705,"about_ca_topic_score_gemma":0.0050112777,"teacher_disagreement_score":0.16726807,"about_ca_system_score_codex":0.0073063327,"about_ca_system_score_gemma":0.1745744,"threshold_uncertainty_score":0.99963295},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3634628,0.0024030423,0.0019296424,0.16550995,0.0023110455,0.12879448,0.19515009,0.00045095195,0.13998799],"genre_scores_gemma":[0.8747709,0.0013277248,0.0051721907,0.00036427326,0.0007022879,0.0052695978,0.080794305,0.00028369442,0.031315047],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99045503,0.0023022138,0.001351933,0.0018172478,0.0024583822,0.0016152113],"domain_scores_gemma":[0.9934971,0.0011076662,0.0002981749,0.0035867486,0.0006885637,0.0008217653],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["open_science"],"category_scores_codex":[0.01816784,0.00043990405,0.001575739,0.0019187084,0.00023525995,0.0013595563,0.008689972,0.0005284358,0.004493767],"category_scores_gemma":[0.0017456607,0.00040686427,0.00010642082,0.0014604126,0.00029632982,0.0005691056,0.021003323,0.0034140295,0.00072374707],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.08741723,0.018093918,0.06750414,0.015170677,0.0053150584,0.0069493228,0.0068678884,0.008528499,0.14508568,0.05927054,0.5502121,0.02958498],"study_design_scores_gemma":[0.017850336,0.0032601436,0.62079257,0.0026002997,0.00017369629,0.000027247359,0.0011242988,0.014744273,0.0012109057,0.028921852,0.307861,0.0014333974],"about_ca_topic_score_codex":0.090083905,"about_ca_topic_score_gemma":0.014312757,"teacher_disagreement_score":0.5532884,"about_ca_system_score_codex":0.0005269145,"about_ca_system_score_gemma":0.0029614978,"threshold_uncertainty_score":0.9998383},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.622133,0.00019214491,0.0003230141,0.3721273,0.00011590345,0.0044593066,0.0005149349,0.000079249,0.000055131673],"genre_scores_gemma":[0.9866263,0.00087136443,0.00007395519,0.0102097215,0.00017605911,0.00032700214,0.0016124309,0.000036711663,0.00006641053],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9873908,0.0062228227,0.0010559034,0.0012098595,0.0020715646,0.002049042],"domain_scores_gemma":[0.9521856,0.043680202,0.00014308463,0.0015434145,0.0008626249,0.0015850902],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.14217275,0.0001832752,0.0006453003,0.0008284406,0.001106075,0.00046788392,0.0008949802,0.00018215853,0.000008293045],"category_scores_gemma":[0.07253789,0.0001364293,0.000023822144,0.0045720567,0.00034565647,0.0001835682,0.000943987,0.0019418383,0.000005614914],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00024035224,0.00035329367,0.96423554,0.0010766939,0.00002164239,0.000009158723,0.0027462256,4.5778776e-7,1.5001294e-7,0.004188377,0.024542434,0.0025856674],"study_design_scores_gemma":[0.0017635555,0.0015881883,0.95435834,0.00012112177,2.4653346e-7,0.0000040130813,0.0052567367,0.00069494965,5.425757e-9,0.0005306576,0.03556689,0.0001152845],"about_ca_topic_score_codex":0.0039975303,"about_ca_topic_score_gemma":0.12488383,"teacher_disagreement_score":0.36449334,"about_ca_system_score_codex":0.00036858546,"about_ca_system_score_gemma":0.0036807598,"threshold_uncertainty_score":0.93527454},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00037221552,0.15012364,0.0886226,0.28079817,0.021165935,0.02727096,0.000034388126,0.0036801214,0.427932],"genre_scores_gemma":[0.0017034083,0.00042922774,0.07184015,0.00063386356,0.0025555817,0.0005781592,0.00058479904,0.00009446509,0.9215804],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9974589,0.00029692493,0.0004339348,0.00116124,0.0003024914,0.00034650648],"domain_scores_gemma":[0.99745446,0.0006066997,0.0007132748,0.00072545325,0.0002968856,0.00020322543],"candidate_categories":["metaepi_narrow","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0012813527,0.00039094442,0.0004557206,0.000753463,0.00079297606,0.0075236973,0.004255402,0.00033586932,0.000078080586],"category_scores_gemma":[0.0006721516,0.00037244626,0.000042690626,0.0003963571,0.000041458294,0.002844689,0.0021154433,0.0011403252,0.000018752062],"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":[0.00007195081,0.000114968934,0.000070253045,0.0005095236,0.00005910861,0.000003938647,0.010419338,0.0005267529,0.000025359646,0.864983,0.08691973,0.036296044],"study_design_scores_gemma":[0.00028240785,0.00017108649,0.00018740223,0.0019307593,0.00004021924,0.000014774281,0.007940003,0.0007238494,0.0000068413046,0.038897187,0.9492254,0.0005800625],"about_ca_topic_score_codex":0.004923188,"about_ca_topic_score_gemma":0.00029163406,"teacher_disagreement_score":0.8623057,"about_ca_system_score_codex":0.0005074882,"about_ca_system_score_gemma":0.0034944257,"threshold_uncertainty_score":0.99987274},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"editorial","genre_consensus":"editorial","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0018681068,0.029613014,0.0000068228574,0.0061852,0.9517626,0.00007923007,0.0003743988,0.00005067181,0.010059917],"genre_scores_gemma":[0.0012956266,0.026712008,0.000024237044,0.000318032,0.9527592,0.000002128861,0.00009956692,0.00013193124,0.01865727],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99296653,0.00010723086,0.0025474066,0.00029960714,0.0036331578,0.00044605797],"domain_scores_gemma":[0.9869439,0.0016579909,0.004819245,0.00033567278,0.005895643,0.00034755486],"candidate_categories":["metaresearch","metaepi_narrow","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.0025821922,0.0005206779,0.001432954,0.00016063318,0.00018947839,0.00052052975,0.0020249055,0.0006428383,0.00052873313],"category_scores_gemma":[0.017476818,0.00043370033,0.0009952052,0.00011176593,0.00020683199,0.0005225441,0.00009978135,0.0063437526,0.000011005316],"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":[0.000052936204,0.00006487526,0.0000018514962,0.0062714457,0.00015164765,0.0004894699,0.0008113763,0.0000063532184,0.0018543535,0.0000058973706,0.9899751,0.00031472207],"study_design_scores_gemma":[0.0004718544,0.00012486616,0.0000033991275,0.011524746,0.0006220498,0.0003440117,0.0006777882,0.0000012828971,0.00035619704,0.000083824205,0.98545074,0.0003392648],"about_ca_topic_score_codex":0.0000053494396,"about_ca_topic_score_gemma":0.0000014461222,"teacher_disagreement_score":0.014894625,"about_ca_system_score_codex":0.00009477984,"about_ca_system_score_gemma":0.0010568688,"threshold_uncertainty_score":0.9998115},"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":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00008925991,0.000042921893,0.00007576008,0.013197804,0.0011297079,0.0013437221,0.00003086987,0.0001284316,0.9839615],"genre_scores_gemma":[0.0014514287,0.00019145149,0.00044605514,0.0024133609,0.0011642032,0.00002982777,0.000023766397,0.000058838814,0.9942211],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9984697,0.00022540396,0.00022224987,0.00033229965,0.0004043741,0.00034598517],"domain_scores_gemma":[0.9943702,0.0045000366,0.00022894055,0.0002457924,0.00050775916,0.00014727512],"candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.004484343,0.00016357927,0.00032331512,0.00009415176,0.0033713446,0.0025781873,0.0016452259,0.0011429281,0.000014976816],"category_scores_gemma":[0.007490101,0.00017727181,0.000083930216,0.000059904763,0.00045774135,0.0002589725,0.0017014297,0.001237611,0.000010668946],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000021513413,0.0000071792206,0.000024834406,0.00016898631,0.00010944927,0.000004235305,0.071716756,7.0240003e-7,0.0000031149386,0.5110776,0.406185,0.010680628],"study_design_scores_gemma":[0.00017472017,0.00003509593,0.000048333346,0.00016713413,0.00003611254,1.6579297e-7,0.00096656854,0.000006016962,0.000003833775,0.5080731,0.49029693,0.00019201072],"about_ca_topic_score_codex":0.0047676717,"about_ca_topic_score_gemma":0.025426144,"teacher_disagreement_score":0.08411193,"about_ca_system_score_codex":0.0002231881,"about_ca_system_score_gemma":0.0033636482,"threshold_uncertainty_score":0.99845725},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000011919342,0.97849303,0.000046980564,0.0017759671,0.00044814873,0.003248423,0.0002006361,0.000034662004,0.015740216],"genre_scores_gemma":[0.000059518916,0.99043804,0.0007820744,0.00010481422,0.0001788051,0.00019472156,0.000010538784,0.00008201458,0.008149494],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9254473,0.007679043,0.0066927625,0.0037446814,0.052935082,0.0035011354],"domain_scores_gemma":[0.9197303,0.03306343,0.004103703,0.007981482,0.033588383,0.001532719],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"category_scores_codex":[0.41561195,0.00047564783,0.0038841157,0.014945129,0.0019217549,0.021407334,0.116024345,0.00039304156,0.0011857807],"category_scores_gemma":[0.49917465,0.00030623478,0.0006131452,0.10404132,0.014687879,0.0157454,0.062203977,0.0046792286,0.0007576335],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000006817073,0.00006089285,0.00007818398,0.02315397,0.000027184857,0.000025172145,0.000072490206,5.274876e-7,0.0000074651475,0.0120298285,0.12855005,0.83598745],"study_design_scores_gemma":[0.00006921027,0.000054582226,0.00005226693,0.15362123,0.000028922921,0.000046367677,0.00014786942,0.000058390444,0.0000096923795,0.0032515868,0.8424055,0.00025437222],"about_ca_topic_score_codex":0.00069812173,"about_ca_topic_score_gemma":0.000026660537,"teacher_disagreement_score":0.83573306,"about_ca_system_score_codex":0.0009798504,"about_ca_system_score_gemma":0.045016076,"threshold_uncertainty_score":0.99993896},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000011993446,0.9778382,0.01830792,0.0006297086,0.00006907827,0.0015322483,0.0013666175,0.00018930787,0.00005488583],"genre_scores_gemma":[0.0000048990632,0.9028638,0.09574673,0.00007862729,0.0000622487,0.00008474556,0.0010290381,0.00006352933,0.000066403314],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99570316,0.000100109246,0.0015634301,0.0018301739,0.00031144818,0.0004916551],"domain_scores_gemma":[0.9852834,0.0003494006,0.00097485556,0.013015329,0.00023359274,0.00014344465],"candidate_categories":["metaepi_narrow","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.0020048076,0.00055584725,0.0026728462,0.0004768522,0.00027272073,0.00040837383,0.035248965,0.0003739454,0.000004558613],"category_scores_gemma":[0.005766435,0.0004941375,0.00034530644,0.0011736223,0.00020942853,0.0002999952,0.2289358,0.0004459782,0.0000034975085],"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":[0.0000018754929,0.000070461356,0.000026563814,0.14521427,0.0020266266,0.000016542821,0.000019712697,4.2413268e-7,0.0000052425844,0.004290755,0.05271924,0.7956083],"study_design_scores_gemma":[0.000106672356,0.000105242885,0.000026956035,0.016211672,0.0029597376,0.000026826909,0.000004594605,0.0037024114,0.0000016119756,0.0051535,0.97114307,0.0005576868],"about_ca_topic_score_codex":0.000036465575,"about_ca_topic_score_gemma":0.000041052517,"teacher_disagreement_score":0.91842383,"about_ca_system_score_codex":0.000066544366,"about_ca_system_score_gemma":0.00026018266,"threshold_uncertainty_score":0.99975103},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.96385854,0.017012915,0.0005190259,0.0122812595,0.0007890886,0.004426777,0.00009521916,0.0002527399,0.0007644233],"genre_scores_gemma":[0.99432284,0.0003536866,0.0007819836,0.0036298593,0.00029838548,0.00029113502,0.00018862062,0.00002006323,0.00011345208],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9901275,0.0038468062,0.0032118831,0.0005201745,0.0012222122,0.0010714353],"domain_scores_gemma":[0.98768497,0.008544801,0.0012020875,0.0008425145,0.0012730625,0.00045259428],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.014220525,0.00020660339,0.0006960557,0.000580692,0.0012979634,0.000018005245,0.00031809328,0.00020068104,0.0005349738],"category_scores_gemma":[0.031726193,0.00015233361,0.000049121823,0.001577327,0.00041477065,0.0009636683,0.0002803349,0.0010174176,0.00005815681],"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":[0.00077151164,0.000059525773,0.23361802,0.039786402,0.000013216849,0.00001092264,0.37883702,0.000011136977,0.0011345492,0.00019682845,0.015993847,0.32956702],"study_design_scores_gemma":[0.0037773978,0.0016884826,0.14142285,0.033519305,0.000021580407,0.0000010629266,0.8020487,0.004830457,0.0014932471,0.00025443104,0.010587525,0.00035494994],"about_ca_topic_score_codex":0.0054331888,"about_ca_topic_score_gemma":0.00082484435,"teacher_disagreement_score":0.4232117,"about_ca_system_score_codex":0.0005657019,"about_ca_system_score_gemma":0.0035006634,"threshold_uncertainty_score":0.9983017},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.014795715,0.013748121,0.00037509174,0.7996442,0.0023067468,0.0009168023,0.0000809942,0.00033555095,0.16779679],"genre_scores_gemma":[0.48113987,0.0019380482,0.0010374915,0.3225276,0.0010179596,0.00012157189,0.000046241777,0.000048181846,0.19212307],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99394035,0.00019291493,0.00053024734,0.00076622673,0.0042182356,0.00035199337],"domain_scores_gemma":[0.9965791,0.0008105249,0.00016694289,0.0010079049,0.0011995237,0.00023600952],"candidate_categories":["metaresearch","scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.013804995,0.00015632673,0.00034869116,0.00027396803,0.00014816398,0.0013102805,0.0038818747,0.0009209026,0.0016282111],"category_scores_gemma":[0.068732314,0.00010077637,0.00014524042,0.0034856377,0.000030275447,0.00096437923,0.0006609582,0.0025346463,0.00586824],"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":[0.0000055328533,0.000005846236,0.0010440155,0.000022270531,0.0000051630664,0.000012369515,0.0004692981,0.000025003834,0.000025252983,0.0006459699,0.9701299,0.027609358],"study_design_scores_gemma":[0.00007524555,0.000009424807,0.004612467,0.00038890247,0.000005740657,0.0000070509755,0.0004780426,0.00004018925,0.000078297584,0.005516379,0.9886413,0.00014692394],"about_ca_topic_score_codex":0.000016426879,"about_ca_topic_score_gemma":0.000024493838,"teacher_disagreement_score":0.4771166,"about_ca_system_score_codex":0.000038747756,"about_ca_system_score_gemma":0.000078905265,"threshold_uncertainty_score":0.9997665},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8804272,0.0007694216,0.0019211663,0.11318531,0.0014244401,0.00028128334,0.00004584657,0.000051388968,0.0018939436],"genre_scores_gemma":[0.9955338,0.00021370554,0.0013955438,0.0014119432,0.00071952236,0.000009573143,0.000009302721,0.000045385783,0.0006611895],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.96279085,0.011403049,0.004708082,0.000892156,0.018629279,0.0015766096],"domain_scores_gemma":[0.9714782,0.0037389067,0.0034430653,0.002147236,0.01834311,0.00084945315],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","scholarly_communication","research_integrity"],"category_scores_codex":[0.25618094,0.000324327,0.00087045063,0.0037644529,0.0015634674,0.21101531,0.03150591,0.001521913,0.00004062138],"category_scores_gemma":[0.31919596,0.00022505262,0.0004442388,0.014421096,0.00041810743,0.6078136,0.0050633885,0.040030204,0.000023407347],"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":[0.0005092843,0.00040606948,0.20783965,0.000057751302,0.00009562336,0.000054942422,0.02387133,0.033425704,0.00985317,0.006038387,0.6869773,0.030870745],"study_design_scores_gemma":[0.0029992263,0.0002844684,0.36575088,0.0009203899,0.00005438295,0.00015479689,0.0741452,0.08098275,0.0010424851,0.46423894,0.0087900935,0.0006363804],"about_ca_topic_score_codex":0.002171173,"about_ca_topic_score_gemma":0.00024224767,"teacher_disagreement_score":0.67818725,"about_ca_system_score_codex":0.0007072893,"about_ca_system_score_gemma":0.00230483,"threshold_uncertainty_score":0.99977434},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008646494,0.98117113,0.0000076118913,0.0007525513,0.00013627112,0.016679864,0.00032098376,0.00003623331,0.00003069756],"genre_scores_gemma":[0.00011092042,0.99107414,0.00019596698,0.0005067531,0.000037481484,0.0072412784,0.0007087441,0.00005175543,0.00007292991],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.98623186,0.008367386,0.0030991565,0.0005765516,0.0005924772,0.0011325912],"domain_scores_gemma":[0.99148214,0.0056216316,0.0015516795,0.00063833635,0.0001290782,0.00057715323],"candidate_categories":["sts"],"consensus_categories":[],"category_scores_codex":[0.0059446613,0.00036460024,0.0026684767,0.0012561857,0.0021004782,0.000019185702,0.00021993545,0.000078568606,0.000042590756],"category_scores_gemma":[0.00036657733,0.00023994954,0.00026412873,0.003912226,0.00011851971,0.00021111673,0.00018718374,0.0013335646,0.000029026423],"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":[0.000009663092,0.000117514,0.0008424281,0.8401853,0.000341334,0.0000011463655,0.10624584,0.0000015431134,6.184651e-9,0.00033304232,0.0011892472,0.050732967],"study_design_scores_gemma":[0.00035748642,0.000422657,0.00039851927,0.95700455,0.00026913782,0.0000013167498,0.035717174,0.00003618763,7.666015e-9,0.00004465268,0.005519062,0.00022926564],"about_ca_topic_score_codex":0.00624388,"about_ca_topic_score_gemma":0.007882901,"teacher_disagreement_score":0.11681926,"about_ca_system_score_codex":0.0014869502,"about_ca_system_score_gemma":0.0033547569,"threshold_uncertainty_score":0.9991987},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9632019,0.0006186675,0.008781082,0.001411323,0.0015456367,0.018429874,0.003996683,0.00023859058,0.0017762871],"genre_scores_gemma":[0.94490117,0.014063758,0.006206463,0.00031747328,0.0014982524,0.022692235,0.0069937916,0.00014368682,0.0031831544],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99173135,0.0014057376,0.0024228145,0.002222185,0.00074701454,0.0014709088],"domain_scores_gemma":[0.98956263,0.0051452415,0.0007421586,0.0026226335,0.0014894359,0.00043789958],"candidate_categories":["metaepi_narrow","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.027914448,0.00031873854,0.00076360395,0.0011018907,0.0011928737,0.00004339578,0.0015797089,0.0011889728,0.00024545967],"category_scores_gemma":[0.002270825,0.00032737674,0.00007399022,0.0012722415,0.00017913563,0.00027077898,0.0060226973,0.0035690977,0.00013672201],"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":[0.00024077651,0.00009908699,0.97254443,0.011359911,0.000009844689,3.580414e-7,0.0023247101,0.00010792467,0.000002208953,0.0033380117,0.008804583,0.0011681252],"study_design_scores_gemma":[0.0011527425,0.00007225965,0.98153603,0.00021020364,0.000007444048,7.6809835e-7,0.001811753,0.0027769257,1.8540372e-7,0.007829817,0.004405032,0.0001968164],"about_ca_topic_score_codex":0.060007762,"about_ca_topic_score_gemma":0.030737648,"teacher_disagreement_score":0.029270113,"about_ca_system_score_codex":0.00082616357,"about_ca_system_score_gemma":0.004730131,"threshold_uncertainty_score":0.9999178},"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":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.18035269,0.024209984,0.20217086,0.40642425,0.026058173,0.09718378,0.040373508,0.00368935,0.019537412],"genre_scores_gemma":[0.8324711,0.0024971156,0.008994116,0.0090191085,0.003775747,0.033771932,0.024169726,0.00061170827,0.084689476],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.99044526,0.0014435493,0.0008560048,0.0020384667,0.003130378,0.0020863577],"domain_scores_gemma":[0.9914175,0.0021873952,0.00028409832,0.0018306796,0.0019979454,0.002282344],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science"],"category_scores_codex":[0.0061146975,0.0005174813,0.0007532445,0.00080380286,0.0013111103,0.005812483,0.008162516,0.00077754277,0.00062830694],"category_scores_gemma":[0.020582497,0.0005377898,0.00036257986,0.0020494433,0.00027439828,0.0007073369,0.022870217,0.002071776,0.0014224793],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.0019406151,0.00030753494,0.095516585,0.01619943,0.00024299968,0.00029753507,0.00021702149,0.00024971462,0.0000036028273,0.024794402,0.83412874,0.026101826],"study_design_scores_gemma":[0.003185157,0.0003192328,0.0802182,0.003806886,0.000034642373,0.00004122941,0.0070798956,0.0005509635,0.00004482904,0.61648744,0.2869661,0.0012654469],"about_ca_topic_score_codex":0.085038036,"about_ca_topic_score_gemma":0.054308396,"teacher_disagreement_score":0.6521184,"about_ca_system_score_codex":0.011832808,"about_ca_system_score_gemma":0.05734802,"threshold_uncertainty_score":0.99998903},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.053215254,0.003070904,0.24600936,0.028437639,0.005727992,0.002549468,0.000056300323,0.00012123037,0.66081184],"genre_scores_gemma":[0.07453955,0.0040900502,0.29816276,0.0018360262,0.00038970457,0.00025354547,0.000008165085,0.000057444042,0.62066275],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.78477293,0.1642962,0.024081146,0.00819101,0.016585842,0.0020728894],"domain_scores_gemma":[0.89276785,0.08068618,0.007930032,0.015723834,0.0020508354,0.0008412615],"candidate_categories":["metaresearch","metaepi_narrow","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","research_integrity","insufficient_payload"],"category_scores_codex":[0.7154771,0.00092619966,0.0061550518,0.0009416204,0.0003947459,0.003800414,0.017692922,0.0014597226,0.037099995],"category_scores_gemma":[0.38394874,0.00048741058,0.0036231158,0.0064227288,0.0014478266,0.00045213298,0.0032808485,0.002777724,0.05153384],"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":[0.000011266745,0.00025694622,0.08583738,0.00037267926,0.00069076807,0.00017065702,0.0047194962,0.0014924543,0.0010436119,0.08350063,0.52916497,0.29273912],"study_design_scores_gemma":[0.00019068997,0.000049773305,0.13934287,0.00037088987,0.00058676605,0.00008620319,0.0030406686,0.030626956,0.00013879695,0.35955733,0.46509087,0.00091819256],"about_ca_topic_score_codex":0.00023571926,"about_ca_topic_score_gemma":0.00017448897,"teacher_disagreement_score":0.3315284,"about_ca_system_score_codex":0.0011273458,"about_ca_system_score_gemma":0.0026908156,"threshold_uncertainty_score":0.99983656},"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":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007924392,0.00038265844,0.89552104,0.045440495,0.0011632221,0.0012496094,0.04769893,0.00019640458,0.00042322715],"genre_scores_gemma":[0.33050996,0.07910057,0.46854258,0.020395124,0.002579436,0.00031218547,0.09261571,0.00020590032,0.0057385247],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99460465,0.00019630945,0.00045698564,0.0023845981,0.0015051971,0.0008522715],"domain_scores_gemma":[0.9871954,0.0010694292,0.00029233177,0.01085734,0.00011681157,0.0004686951],"candidate_categories":["sts","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.018114448,0.00020920412,0.00020698682,0.00029965286,0.0014741666,0.01234649,0.0605602,0.000033724227,0.000011258881],"category_scores_gemma":[0.0030227124,0.00014068585,0.000015441365,0.0019156693,0.00052056194,0.057475705,0.058726847,0.0004146619,0.00006275514],"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":[0.000038122475,0.00004640682,0.0015623563,0.000025442028,0.000098170545,0.00008164605,0.00046501172,0.00004189485,0.0005124847,0.0082652075,0.92726743,0.0615958],"study_design_scores_gemma":[0.00047026496,0.000054778648,0.0037388315,0.00008534134,0.000055969584,0.00006942753,0.003096505,0.15674263,0.000019872927,0.0012562539,0.8341447,0.00026538182],"about_ca_topic_score_codex":0.0002910757,"about_ca_topic_score_gemma":0.00064453273,"teacher_disagreement_score":0.42697847,"about_ca_system_score_codex":0.00008032258,"about_ca_system_score_gemma":0.00053901714,"threshold_uncertainty_score":0.9998258},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9869408,0.0015714776,0.0055779372,0.0007688015,0.00042816225,0.0008057077,0.00038918667,0.00008438435,0.0034335344],"genre_scores_gemma":[0.99547625,0.00044450123,0.0031008993,0.000014516861,0.00023437006,0.0000058219794,0.0002169215,0.000019084491,0.00048762892],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9977133,0.00028814955,0.0005232011,0.0008150859,0.00037439485,0.00028584572],"domain_scores_gemma":[0.9986504,0.00033704966,0.0002368691,0.00055050023,0.000059160775,0.0001660588],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0015349183,0.00021017088,0.00058432634,0.00005781982,0.00031469128,0.00031384607,0.0011421655,0.00042212204,0.00073224015],"category_scores_gemma":[0.0009697347,0.00018342855,0.000038692735,0.00011827985,0.00055821013,0.00044014916,0.0048431,0.0007020194,4.8646666e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000066666275,0.000050188286,0.99136966,0.000075590826,0.000028565499,0.0000056945782,0.0039774315,0.000006332548,0.0000034246298,0.00060589827,0.000043869226,0.0037666834],"study_design_scores_gemma":[0.00035702798,0.000039082697,0.942606,0.0010552519,0.00014883012,1.5401864e-7,0.0028300397,0.041462176,0.000013759624,0.009538013,0.0014785285,0.00047117233],"about_ca_topic_score_codex":0.7993952,"about_ca_topic_score_gemma":0.81377625,"teacher_disagreement_score":0.04876369,"about_ca_system_score_codex":0.00008521623,"about_ca_system_score_gemma":0.00040632542,"threshold_uncertainty_score":0.8017516},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00000156842,0.95274144,0.0008328381,0.00096773286,0.00022736451,0.039943073,0.00040532995,0.00003166507,0.0048489557],"genre_scores_gemma":[7.541087e-7,0.88892174,0.0021716943,0.00012443298,0.00012582159,0.004082905,0.0001179304,0.000121199846,0.104333535],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.84950775,0.065495804,0.032349646,0.0063419193,0.043247197,0.0030576554],"domain_scores_gemma":[0.91501856,0.031384517,0.010397979,0.024995157,0.014842099,0.0033616798],"candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","open_science","insufficient_payload"],"category_scores_codex":[0.6238078,0.0010101903,0.017815633,0.0058735134,0.0011286838,0.01555474,0.052227702,0.00021593316,0.11683375],"category_scores_gemma":[0.43057942,0.00041716316,0.008058749,0.04485823,0.0005094063,0.0011709919,0.025142452,0.001346237,0.06750111],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000006378436,0.0000796295,0.000010736847,0.15493038,0.00036424558,0.00005070751,0.000030175703,5.011193e-7,7.906348e-7,0.00018960814,0.2834445,0.56089234],"study_design_scores_gemma":[0.000084524574,0.00017187916,0.000016108683,0.4166807,0.00044075947,0.000024343737,0.00006230485,0.000036328882,3.7126563e-7,0.00017922072,0.58197564,0.00032779505],"about_ca_topic_score_codex":0.000252922,"about_ca_topic_score_gemma":0.00075221085,"teacher_disagreement_score":0.5605646,"about_ca_system_score_codex":0.0009931418,"about_ca_system_score_gemma":0.015876397,"threshold_uncertainty_score":0.99982804},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.6348772,0.004634578,0.0019395038,0.3453059,0.0007597984,0.0073545105,0.0012140219,0.00026641798,0.0036480546],"genre_scores_gemma":[0.95612174,0.0068169404,0.005957204,0.011189121,0.006387849,0.00028994295,0.0010163292,0.00013025301,0.01209063],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99663264,0.0002335472,0.00086398976,0.0008647543,0.0010302071,0.00037486758],"domain_scores_gemma":[0.9670718,0.030188294,0.00013647467,0.0017311187,0.0004366022,0.00043571528],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.015229194,0.00017360273,0.0006393567,0.00018861203,0.00010997968,0.000033255044,0.00092450937,0.0002473411,0.00022180784],"category_scores_gemma":[0.12886167,0.00009836863,0.00009509899,0.00061124476,0.0008247851,0.000039688264,0.0007499264,0.001209731,0.000035315286],"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":[0.3426383,0.002082869,0.017861983,0.0044017397,0.0006765768,0.000046574525,0.011433434,0.00006978598,0.001315419,0.06778744,0.546333,0.005352831],"study_design_scores_gemma":[0.20600972,0.02462385,0.031039847,0.009112635,0.0009221425,0.000016834094,0.010412694,0.1645164,0.00040646546,0.0042251707,0.5480989,0.0006153298],"about_ca_topic_score_codex":0.00001442144,"about_ca_topic_score_gemma":0.00003588255,"teacher_disagreement_score":0.3341168,"about_ca_system_score_codex":0.00003127406,"about_ca_system_score_gemma":0.00030826786,"threshold_uncertainty_score":0.8784763},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8143109,0.00044920808,0.13529609,0.026228024,0.0021587112,0.012266756,0.00094550906,0.00059724605,0.0077475524],"genre_scores_gemma":[0.9835308,0.00006238756,0.00013024203,0.0010202709,0.0009923306,0.0024926045,0.00018756404,0.00003946229,0.011544327],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.7840853,0.00015056967,0.00041496794,0.00029926604,0.00015213771,0.21489778],"domain_scores_gemma":[0.99842924,0.0005412186,0.00006417599,0.00017374837,0.0006440939,0.00014753686],"candidate_categories":["metaresearch","sts","open_science"],"consensus_categories":[],"category_scores_codex":[0.23448296,0.00013053558,0.00028821506,0.00017619475,0.39244196,0.000049775514,0.00013413624,0.00014266824,0.00009246337],"category_scores_gemma":[0.00037421336,0.00010720598,0.00005272264,0.0005560982,0.00003153662,0.00018688264,0.090484366,0.0003792227,0.00028911929],"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":[0.00021667691,0.000038785227,0.047039304,0.0013524197,0.0001016522,0.000028436176,0.1524824,0.00019202399,0.00012885635,0.72036487,0.07725184,0.00080276007],"study_design_scores_gemma":[0.001202136,0.00019974596,0.0014066503,0.00005785876,0.000010523936,0.0000014646421,0.7550906,0.0017323715,0.000012816821,0.0021171009,0.23801957,0.00014917804],"about_ca_topic_score_codex":0.00084694446,"about_ca_topic_score_gemma":0.014727971,"teacher_disagreement_score":0.7182478,"about_ca_system_score_codex":0.00030392635,"about_ca_system_score_gemma":0.00012553336,"threshold_uncertainty_score":0.9168716},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19657588,0.0017471187,0.0022292768,0.7961205,0.0006217373,0.0022376925,0.00038392338,0.00007954325,0.0000043013974],"genre_scores_gemma":[0.6790181,0.0019764483,0.008004658,0.2989389,0.0049105557,0.0000760517,0.005305809,0.00007395882,0.0016955558],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99700123,0.0002792371,0.000960347,0.00068544934,0.000708918,0.00036484355],"domain_scores_gemma":[0.97519714,0.022003612,0.00047027573,0.0015424707,0.0002576788,0.00052881683],"candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.003728153,0.00024143711,0.0006208716,0.00016503193,0.00017821595,0.000044454795,0.00034855536,0.00006892782,0.0001138981],"category_scores_gemma":[0.0633927,0.00014054097,0.00004574471,0.0006062299,0.00008416343,0.00031136005,0.00015626024,0.00026516587,0.000006321041],"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":[0.0009862694,0.000016115957,0.0056848936,0.00011218702,0.000087238135,0.00001914026,0.0001235041,0.00003725702,0.0006102802,0.000018241586,0.9261884,0.06611644],"study_design_scores_gemma":[0.007819463,0.0034867413,0.42812535,0.0038032427,0.0008906423,0.000014813733,0.007780022,0.0058542844,0.0005178239,0.00044791025,0.54084516,0.00041453028],"about_ca_topic_score_codex":0.43911657,"about_ca_topic_score_gemma":0.34300613,"teacher_disagreement_score":0.49718162,"about_ca_system_score_codex":0.00019979775,"about_ca_system_score_gemma":0.004597661,"threshold_uncertainty_score":0.94449675},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.019717168,0.00013421202,0.00214716,0.0070394776,0.005163345,0.002938975,0.00037309685,0.00027243973,0.9622141],"genre_scores_gemma":[0.98618895,0.00022945106,0.0010206364,0.00013613891,0.00037602967,0.00010005207,0.00013291022,0.000012012328,0.011803838],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"qualitative","domain_scores_codex":[0.9982119,0.00023482503,0.00047812995,0.0002543983,0.0006196506,0.00020110315],"domain_scores_gemma":[0.9976794,0.00043424257,0.00043748546,0.0012164487,0.00020301886,0.000029373683],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0058662468,0.000105132494,0.00027232297,0.00007544745,0.00058540184,0.00027998956,0.004988407,0.00013710867,0.00011971529],"category_scores_gemma":[0.0007541616,0.00007012802,0.000038575377,0.00027102526,0.00030377862,0.000257498,0.018491395,0.00021361983,0.00009049736],"study_design_candidate":"qualitative","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010965928,0.000034726174,0.07537649,0.00031882952,0.00081099506,0.000002168992,0.33312845,0.0007689102,0.0000010705003,0.42602807,0.1537986,0.009720724],"study_design_scores_gemma":[0.0003902245,0.000031693136,0.16918747,0.0005303027,0.00014100438,5.126521e-7,0.5344711,0.0021696647,0.0000059035083,0.025170414,0.26728263,0.0006191063],"about_ca_topic_score_codex":0.19782299,"about_ca_topic_score_gemma":0.040750694,"teacher_disagreement_score":0.9664718,"about_ca_system_score_codex":0.000020934536,"about_ca_system_score_gemma":0.00034767407,"threshold_uncertainty_score":0.9894469},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9966744,0.0001297199,0.00005994243,0.00041326636,0.00015527986,0.0020326271,0.000063797415,0.0000029662494,0.000467959],"genre_scores_gemma":[0.99704474,0.0000018821487,0.00073836715,0.000023629282,0.000022785438,0.000043494278,0.000010940266,0.000003901791,0.0021102862],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9534682,0.012951975,0.026694076,0.0011672289,0.0054857777,0.0002327634],"domain_scores_gemma":[0.9751249,0.005596789,0.016722215,0.001479586,0.0009927254,0.00008378129],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.23104759,0.00012369426,0.0029002053,0.0002150163,0.00005097611,0.00016566583,0.00031791584,0.000026923883,0.00045236072],"category_scores_gemma":[0.40043446,0.00006629458,0.00015699501,0.0016552755,0.000008208607,0.00009559101,0.000071270435,0.00007361476,0.000077480414],"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":[0.000035662797,0.000037263315,0.8762422,0.00001523228,0.00018771477,0.00002471798,0.005958741,0.0019770437,0.010013146,0.0000228496,0.09384729,0.011638113],"study_design_scores_gemma":[0.00044230357,0.00007397306,0.9830979,0.000049862676,0.00008253963,7.9847877e-7,0.009498992,0.0050881263,0.00035155617,0.00093377254,0.00029093528,0.00008923657],"about_ca_topic_score_codex":0.51817936,"about_ca_topic_score_gemma":0.7627005,"teacher_disagreement_score":0.24452113,"about_ca_system_score_codex":0.000080033,"about_ca_system_score_gemma":0.0005118911,"threshold_uncertainty_score":0.79179835},"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":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.2922622,0.00019572368,0.69416356,0.0025018933,0.0027064434,0.0012356299,0.00015882395,0.001304415,0.0054713255],"genre_scores_gemma":[0.9969276,0.000006197041,0.0017156892,0.00041448022,0.0003204725,0.00003989243,0.00007273964,0.000018150997,0.00048473422],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983521,0.000012056344,0.00020804413,0.0007593851,0.00037333072,0.00029505853],"domain_scores_gemma":[0.9987037,0.00004111231,0.000026706186,0.0010982121,0.00010519431,0.000025053876],"candidate_categories":["scholarly_communication","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0012326638,0.00013586572,0.00013100111,0.00024847142,0.00060286804,0.0032386363,0.003029381,0.00003923771,0.0019986862],"category_scores_gemma":[0.000085288826,0.00010972333,0.000019518613,0.0018279913,0.00024061259,0.009114571,0.00046204808,0.00021186502,0.00080887275],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00011426795,0.004171017,0.008218434,0.00027599777,0.00009741064,0.00025477711,0.00061228726,0.00059560576,0.0038609856,0.022170462,0.012740713,0.946888],"study_design_scores_gemma":[0.00078122667,0.00019162364,0.08519994,0.00015030554,0.000551682,0.00006167538,0.004090074,0.078177676,0.00057211315,0.015049728,0.813847,0.0013269426],"about_ca_topic_score_codex":0.001694106,"about_ca_topic_score_gemma":0.0011585996,"teacher_disagreement_score":0.9455611,"about_ca_system_score_codex":0.000021932952,"about_ca_system_score_gemma":0.000080217185,"threshold_uncertainty_score":0.9999691},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8130016,0.00032656497,0.029291391,0.13790385,0.015618308,0.0018506801,0.00015617463,0.00017804859,0.0016733861],"genre_scores_gemma":[0.9846488,0.000066972796,0.0011929079,0.0009045196,0.00076061476,0.000033823348,0.000014099173,0.00004525595,0.012333009],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99293315,0.0005509937,0.001786036,0.00070688874,0.003411568,0.0006113863],"domain_scores_gemma":[0.98856246,0.007655109,0.0014116801,0.0015018709,0.00055502536,0.00031383938],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.023282766,0.00028694622,0.0006132101,0.001037182,0.0008863417,0.013810027,0.013246387,0.00010195591,0.00010393669],"category_scores_gemma":[0.005928689,0.00015665236,0.00045990798,0.0036137195,0.00019694683,0.003818621,0.003771629,0.00064441294,0.00013962579],"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":[0.00021268065,0.0000928205,0.0055091535,0.000011734326,0.00018755108,0.000077954755,0.0029313331,0.0017414198,0.000023394037,0.002390896,0.8929519,0.09386918],"study_design_scores_gemma":[0.002511778,0.0003271252,0.55556643,0.00013372237,0.00022814995,0.000046228055,0.017408831,0.012905924,0.0005760596,0.153087,0.2566928,0.00051594066],"about_ca_topic_score_codex":0.00003737644,"about_ca_topic_score_gemma":0.000015419708,"teacher_disagreement_score":0.6362591,"about_ca_system_score_codex":0.000049639537,"about_ca_system_score_gemma":0.00028110752,"threshold_uncertainty_score":0.99209243},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.015351371,0.00008334284,0.00068973744,0.023876827,0.0018425903,0.0033091882,0.11089238,0.00028402187,0.84367055],"genre_scores_gemma":[0.094960235,0.00003196608,0.006271542,0.000989598,0.00076550385,0.00007937108,0.20054471,0.00018048605,0.6961766],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.992746,0.000490769,0.0003860187,0.0027294613,0.002509832,0.0011379273],"domain_scores_gemma":[0.9864356,0.00045778556,0.000029692716,0.012486014,0.000038297738,0.0005525932],"candidate_categories":["sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.021170313,0.00020158009,0.00017400441,0.0001856641,0.0019423502,0.009136424,0.031770524,0.00005669388,0.029096266],"category_scores_gemma":[0.0009803518,0.00014196066,0.000023812343,0.0028145155,0.002154493,0.0032849305,0.04894717,0.0006484291,0.0022895355],"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":[0.000045162855,0.000093067196,0.00028843596,0.000018950677,0.000007979878,0.00006483953,0.00022210414,0.000006194464,0.001159422,0.000366085,0.909635,0.08809274],"study_design_scores_gemma":[0.00012393505,0.00014648159,0.001489544,0.00001584734,0.000008335947,0.000007333971,0.0009942851,0.012136628,0.0001036312,0.00018200703,0.98458904,0.00020292783],"about_ca_topic_score_codex":0.002008597,"about_ca_topic_score_gemma":0.009129399,"teacher_disagreement_score":0.14749397,"about_ca_system_score_codex":0.00013176027,"about_ca_system_score_gemma":0.00041365906,"threshold_uncertainty_score":0.999357},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0040463884,0.001211692,0.018829426,0.8711597,0.003030565,0.004972358,0.0012036709,0.00013650054,0.09540972],"genre_scores_gemma":[0.87571234,0.0008637267,0.00486337,0.06277444,0.00032108344,0.0004286651,0.000060838964,0.000040355055,0.05493518],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99497384,0.00026439992,0.00028505217,0.0019411108,0.0015630644,0.0009725178],"domain_scores_gemma":[0.98998135,0.0010682158,0.00012399879,0.008368916,0.000074833086,0.00038268362],"candidate_categories":["sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.008468287,0.00021934345,0.00016421649,0.000052831354,0.0031167287,0.06111741,0.14124198,0.000032745633,0.00012861643],"category_scores_gemma":[0.002536186,0.000102260616,0.000032513228,0.0025177274,0.0006635047,0.041486796,0.09295143,0.0005089907,0.0008167952],"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":[0.000014720804,0.000024621431,0.00035325467,0.0000067383653,0.000027524758,0.00016908131,0.0019292682,0.00033259336,0.00034193028,0.25115937,0.7153663,0.030274585],"study_design_scores_gemma":[0.00007678283,0.000038114937,0.019667637,0.000026900812,0.000011645646,0.000013642121,0.00011630071,0.04833529,0.00006620401,0.0008141437,0.9306569,0.00017641272],"about_ca_topic_score_codex":0.3141188,"about_ca_topic_score_gemma":0.7715022,"teacher_disagreement_score":0.87166595,"about_ca_system_score_codex":0.00033308277,"about_ca_system_score_gemma":0.0020412463,"threshold_uncertainty_score":0.9999612},"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":"methods","domain_gemma":"methods","model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":"methods","domain_consensus":"methods","prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.963336,0.0019613488,0.02054943,0.002425386,0.009168908,0.0007745321,0.000003412445,0.000004752177,0.0017762262],"genre_scores_gemma":[0.99175614,0.00006569527,0.0040569953,0.00048210187,0.0029487526,0.000011066897,0.0000031893944,0.000010738575,0.0006653353],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.13605101,0.43876162,0.4036369,0.0042311703,0.016520971,0.00079833367],"domain_scores_gemma":[0.108483255,0.66604775,0.20321195,0.0055671292,0.015749713,0.0009402136],"candidate_categories":["metaresearch","metaepi_broad","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.9635504,0.0003014415,0.018595431,0.000504372,0.000084900494,0.00015563246,0.001735704,0.00048784277,0.013152716],"category_scores_gemma":[0.9669375,0.00012448896,0.0075487136,0.00123745,0.0007037486,0.0004512868,0.00022836706,0.001573726,0.0001250367],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00009213033,0.00038895884,0.913009,0.000056271037,0.0007694789,0.00006950376,0.00008270089,0.0002731462,0.0000019038658,0.00027776003,0.017416183,0.06756296],"study_design_scores_gemma":[0.00054879294,0.0010881764,0.885282,0.00008720697,0.00066989235,0.00025466917,0.00023744584,0.02403478,3.198103e-7,0.08023911,0.007475642,0.00008196315],"about_ca_topic_score_codex":0.000020451973,"about_ca_topic_score_gemma":0.000016210613,"teacher_disagreement_score":0.22728613,"about_ca_system_score_codex":0.00004220455,"about_ca_system_score_gemma":0.0012506628,"threshold_uncertainty_score":0.99597454},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11193184,0.000012923625,0.000077571225,0.002441619,0.00035419854,0.00026198794,0.000002220355,0.000057358648,0.8848603],"genre_scores_gemma":[0.9967105,0.000039174523,0.0005793432,0.0001796404,0.000020058831,0.000008705192,0.0000036538383,0.000007688424,0.0024512857],"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9981676,0.000011190614,0.00016541028,0.0006943001,0.00069171493,0.00026982406],"domain_scores_gemma":[0.9995882,0.000009851079,0.0000329972,0.00027280406,0.000003224589,0.00009293868],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0008859925,0.0001759979,0.00014856993,0.00018442581,0.00063108164,0.00017565789,0.0011192444,0.000050119786,0.0008701873],"category_scores_gemma":[0.0000070454907,0.0001529274,0.000035502653,0.00027565382,0.0003513776,0.00012596627,0.9996964,0.00033139618,0.0000859287],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000092460605,0.000033995628,0.0011710929,0.00013852264,0.0000027036722,0.000063175125,0.0013275426,0.00022452716,0.00035504848,0.00040183187,0.0003518323,0.9959205],"study_design_scores_gemma":[0.000086877015,0.00003229277,0.00971341,0.00008755594,0.0000102868025,0.0000043060036,0.000095985735,0.00396105,0.00006475683,0.9857155,0.00003847431,0.00018947404],"about_ca_topic_score_codex":0.0033247345,"about_ca_topic_score_gemma":0.005041017,"teacher_disagreement_score":0.9985771,"about_ca_system_score_codex":0.00046280053,"about_ca_system_score_gemma":0.0000123260015,"threshold_uncertainty_score":0.9527941},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[2.0676224e-7,0.9811065,0.000047295664,0.010041269,0.000509563,0.0059649227,0.0018351015,0.00010665454,0.00038847848],"genre_scores_gemma":[5.67793e-7,0.97671914,0.0042514848,0.0075560706,0.0008746603,0.00010186259,0.0077737193,0.00017350273,0.002549015],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.9840281,0.0009995457,0.008571987,0.0031742726,0.0025531924,0.00067288906],"domain_scores_gemma":[0.92852044,0.04595831,0.0029393532,0.020561524,0.00060621515,0.0014141651],"candidate_categories":["metaresearch","metaepi_narrow","metaepi_broad","open_science","research_integrity"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.051309884,0.00076128315,0.015090791,0.00036731837,0.00007372333,0.000075613876,0.0040788013,0.0010127542,0.00017885567],"category_scores_gemma":[0.13160126,0.00045769112,0.0005106792,0.0009040512,0.0010684485,0.00038634345,0.014122595,0.008906658,0.00014070536],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013005053,0.00016570362,0.000011988277,0.78599334,0.0012625365,0.00018381035,0.000009364134,1.0013006e-9,1.1233256e-8,0.00021727821,0.027190492,0.18495245],"study_design_scores_gemma":[0.00033962863,0.00042220918,0.000019794787,0.670038,0.01201654,0.00031787486,0.0000031961874,0.000018335024,2.6597335e-9,0.00011315975,0.3165144,0.00019687464],"about_ca_topic_score_codex":0.00014836855,"about_ca_topic_score_gemma":0.00007686352,"teacher_disagreement_score":0.2893239,"about_ca_system_score_codex":0.00022652467,"about_ca_system_score_gemma":0.008982263,"threshold_uncertainty_score":0.9997875},"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":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0024302867,0.00005751791,0.000092630005,0.027244078,0.00014827472,0.0053145294,0.95389473,0.0000970665,0.0107208565],"genre_scores_gemma":[0.2110494,0.00003866496,0.000013772535,0.0013560194,0.000046576,0.0000034812883,0.78692126,0.00035342344,0.00021739809],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9914386,0.0036906337,0.0015791695,0.0009844801,0.0019246954,0.0003824071],"domain_scores_gemma":[0.9945707,0.0013107491,0.0012138119,0.0018333996,0.00081760035,0.0002537913],"candidate_categories":["metaresearch","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.01791987,0.0002334817,0.0005866844,0.0014203305,0.0007572228,0.007188517,0.035260946,0.00023257989,0.009860677],"category_scores_gemma":[0.06043715,0.00017458505,0.00007604608,0.008641908,0.00028222965,0.002299774,0.011966516,0.0012089644,0.0009564131],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002512527,0.00071040064,0.000663527,0.00006225192,0.000024474444,0.000018944516,0.0021609177,0.00019498495,3.153666e-7,0.00004368704,0.99318236,0.0026868687],"study_design_scores_gemma":[0.0011000473,0.00026334278,0.04434029,0.000037054964,0.00001938506,0.000010859174,0.0025619138,0.0002480281,8.7749964e-7,0.0002867326,0.950966,0.00016548405],"about_ca_topic_score_codex":0.0013462107,"about_ca_topic_score_gemma":0.0001749114,"teacher_disagreement_score":0.2086191,"about_ca_system_score_codex":0.00020513653,"about_ca_system_score_gemma":0.000118426484,"threshold_uncertainty_score":0.9998215},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"reporting","study_design":"observational","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"design_other","genre":"dataset","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00011867884,0.0000018896641,0.00096787716,0.00042562457,0.00047373783,0.0018319633,0.9960969,0.00005928353,0.000024039291],"genre_scores_gemma":[0.0028307473,0.00012850841,0.00035776556,0.00008049131,0.0010251508,0.00013894596,0.9953818,0.00002547006,0.000031121388],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9980481,0.0001404717,0.00042315276,0.00064230757,0.00031197982,0.000433984],"domain_scores_gemma":[0.99379647,0.002981387,0.000433343,0.002512048,0.000112106594,0.0001646348],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0016902385,0.0001707311,0.00029294126,0.00011072528,0.0008211847,0.00015235972,0.0018333382,0.00026529116,0.000026809626],"category_scores_gemma":[0.0078159785,0.00016189457,0.00007544269,0.0002806088,0.000362181,0.00024813425,0.00036712145,0.00018198679,0.00038336267],"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":[0.000012207741,0.000020800546,0.003798363,0.000031568783,0.000115824914,4.1883904e-8,0.00019491251,1.3433342e-7,8.5279504e-7,0.0021709218,0.9924006,0.0012537827],"study_design_scores_gemma":[0.00026424375,0.000020039022,0.060535904,0.00007559146,0.00030953516,1.936474e-7,0.0005790418,0.0000027468254,5.65711e-7,0.0017657545,0.9362947,0.00015171058],"about_ca_topic_score_codex":0.0075678765,"about_ca_topic_score_gemma":0.0029966687,"teacher_disagreement_score":0.056737542,"about_ca_system_score_codex":0.00012758923,"about_ca_system_score_gemma":0.0002744049,"threshold_uncertainty_score":0.99904084},"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00049005105,0.0000049654527,0.00001148893,0.0009574673,0.00052106695,0.00042691556,0.997417,0.000029885137,0.00014118064],"genre_scores_gemma":[0.0004905223,0.0011024283,0.000038072925,0.00025585172,0.0019729086,0.000013765088,0.9941451,0.0000024658025,0.0019788756],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99788237,0.00027573507,0.00029651864,0.00087668234,0.0002884567,0.0003802462],"domain_scores_gemma":[0.996459,0.0003483611,0.0003111053,0.0026183936,0.00008783531,0.00017530282],"candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0014231871,0.0002559686,0.0002509099,0.000008471961,0.00081300514,0.00044238495,0.0033753228,0.00015118078,0.003225975],"category_scores_gemma":[0.00035739754,0.00004788278,0.00008248817,0.00004629173,0.00020812382,0.0004037328,0.0015669514,0.00027337577,0.007510677],"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":[0.000025329895,0.000029250235,0.00013096079,0.000014392958,0.000033148903,0.000015353795,0.0000035772678,2.9979705e-7,0.000043012642,0.0000115600105,0.99329305,0.0064000497],"study_design_scores_gemma":[0.00013887235,0.000055837703,0.011976695,0.00014070586,0.00010559546,0.0000117311,0.0000156098,0.000027750293,0.0000039496135,0.00021075782,0.9870738,0.00023874025],"about_ca_topic_score_codex":0.010669476,"about_ca_topic_score_gemma":0.016393077,"teacher_disagreement_score":0.011845734,"about_ca_system_score_codex":0.000028070377,"about_ca_system_score_gemma":0.000032171185,"threshold_uncertainty_score":0.9976852},"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":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012613918,0.0002387344,0.92222667,0.053168505,0.005261929,0.00090116047,0.0030745175,0.00038874277,0.002125855],"genre_scores_gemma":[0.29750136,0.000011070296,0.6698908,0.0049290024,0.0009479158,0.00014830445,0.0024846476,0.00009471123,0.023992201],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9963288,0.000118272976,0.00033632302,0.0018438186,0.0007133027,0.0006595083],"domain_scores_gemma":[0.9957167,0.00051916693,0.000037912683,0.0030375936,0.00004774689,0.00064086],"candidate_categories":["scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science"],"category_scores_codex":[0.0029368105,0.00020133937,0.00016564633,0.0001659364,0.000640145,0.008770937,0.01697648,0.000085767206,0.000100273915],"category_scores_gemma":[0.00053094234,0.00016551367,0.000040830455,0.0024227554,0.00011724587,0.003298231,0.018442383,0.00041192604,0.00086887437],"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":[0.000008673074,0.000011482642,0.0000018762956,0.000012047744,0.000029678647,0.000036015605,0.00054676534,0.000554123,0.011185032,0.23251404,0.52751696,0.22758332],"study_design_scores_gemma":[0.00008850065,0.000015150297,0.000051820345,0.00013148114,0.000008671461,0.000022009894,0.00007974558,0.17645729,0.019771375,0.04400361,0.75902724,0.00034310072],"about_ca_topic_score_codex":0.0013052608,"about_ca_topic_score_gemma":0.00036135115,"teacher_disagreement_score":0.28488743,"about_ca_system_score_codex":0.0001088179,"about_ca_system_score_gemma":0.0016249714,"threshold_uncertainty_score":0.99990904},"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":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.023907922,0.06036882,0.0034393712,0.0028202797,0.00034549335,0.00075142557,0.0005673854,0.0007093606,0.90708995],"genre_scores_gemma":[0.17953499,0.010285213,0.0056118453,0.00095047196,0.0007667127,0.000014929696,0.00020924253,0.00024936706,0.8023772],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9950938,0.0002879831,0.00057867717,0.0013044808,0.0008837277,0.001851292],"domain_scores_gemma":[0.9952307,0.00057184696,0.0006103667,0.0012967247,0.0005216897,0.0017686826],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow"],"category_scores_codex":[0.0002589662,0.0013420262,0.0010506252,0.00034104838,0.0013896362,0.0014078205,0.0031151173,0.0013230635,0.0020212987],"category_scores_gemma":[0.00014296811,0.0012973144,0.00069114135,0.00010550044,0.000627648,0.0019391052,0.005650396,0.001004094,0.00035541985],"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":[0.0019247523,0.00032273325,0.0020576743,0.0006771481,0.0017073756,0.022951737,0.038848046,0.0084846355,0.008499605,0.13926002,0.5282058,0.24706045],"study_design_scores_gemma":[0.0017401931,0.00049865927,0.00509045,0.00071460265,0.0006800383,0.0022009895,0.000661833,0.006038842,0.0048281634,0.016006164,0.9599319,0.001608128],"about_ca_topic_score_codex":0.085653074,"about_ca_topic_score_gemma":0.071998,"teacher_disagreement_score":0.43172613,"about_ca_system_score_codex":0.004261522,"about_ca_system_score_gemma":0.0009129744,"threshold_uncertainty_score":0.9999734},"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":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19710302,0.0026523217,0.0000062821878,0.75814533,0.003413004,0.00044061049,0.00006350738,0.00004546049,0.038130447],"genre_scores_gemma":[0.6438277,0.008710234,0.000053132986,0.31597936,0.0048202476,0.000019098143,0.000040955852,0.000052411644,0.026496897],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98988515,0.0041483296,0.0010258863,0.00019033239,0.004261311,0.00048900815],"domain_scores_gemma":[0.9873257,0.011113919,0.00096617045,0.00020416715,0.00006665422,0.00032335898],"candidate_categories":["sts","open_science","research_integrity","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.009201007,0.00016815992,0.0004260039,0.000152201,0.0028617228,0.000108094464,0.001621247,0.00059072056,0.0057951002],"category_scores_gemma":[0.0037953514,0.00008397715,0.00024401005,0.00067115703,0.000113525115,0.0015136452,0.039418384,0.0030967542,0.000128473],"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":[0.000055836954,0.0000635785,0.3796105,0.000116513605,0.00011239148,0.00012881102,0.017145125,6.0210783e-7,6.8537344e-7,0.0012507371,0.59888893,0.002626297],"study_design_scores_gemma":[0.0015721845,0.0001056606,0.48065633,0.0009744057,0.00009956115,0.000036624057,0.0021880209,0.00008252673,0.000003846247,0.019063685,0.4950714,0.00014575683],"about_ca_topic_score_codex":0.000026169466,"about_ca_topic_score_gemma":0.00012111982,"teacher_disagreement_score":0.44672465,"about_ca_system_score_codex":0.0010496291,"about_ca_system_score_gemma":0.003148188,"threshold_uncertainty_score":0.99920315},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.97176385,0.003221606,0.000027614853,0.003474311,0.00063294295,0.0033231024,0.00024089857,0.00004371156,0.017271962],"genre_scores_gemma":[0.9966103,0.0006195153,0.00024486126,0.00047133968,0.00022180178,0.00058005843,0.00011777777,0.000032875378,0.0011015129],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.94789094,0.04041365,0.0039289105,0.00072420965,0.0052270377,0.001815271],"domain_scores_gemma":[0.97951555,0.01692411,0.00033009978,0.0010092445,0.0013689997,0.0008519808],"candidate_categories":["metaresearch","research_integrity","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.115895,0.00019482005,0.00080374157,0.001180975,0.0007523909,0.000018435278,0.00055660185,0.00030475427,0.0016895059],"category_scores_gemma":[0.0064951945,0.00017163943,0.00015048732,0.001973936,0.00052119733,0.00014710317,0.0009498409,0.0036111563,0.0006536989],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0031009435,0.002963644,0.6340804,0.09708354,0.00013698242,0.000010085177,0.08429321,0.000016137428,0.0023839858,0.115700096,0.043336347,0.016894605],"study_design_scores_gemma":[0.0068968176,0.005922224,0.49236146,0.017314574,0.000011983503,1.7058804e-7,0.27872235,0.0008347744,0.0022524197,0.028665649,0.16602369,0.0009939083],"about_ca_topic_score_codex":0.035067245,"about_ca_topic_score_gemma":0.002822925,"teacher_disagreement_score":0.19442913,"about_ca_system_score_codex":0.0008921096,"about_ca_system_score_gemma":0.0042881584,"threshold_uncertainty_score":0.9992231},"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":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000072555895,0.66922426,0.000049358747,0.0007807373,0.00007331307,0.32880786,0.000036590845,0.00006511147,0.000955513],"genre_scores_gemma":[7.141995e-7,0.55772704,0.00025625172,0.00079991797,0.00007194012,0.44081646,0.000021682652,0.00004822041,0.00025777315],"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","domain_scores_codex":[0.9881011,0.00811902,0.00190473,0.0004439125,0.00046758848,0.0009636344],"domain_scores_gemma":[0.99050415,0.007527561,0.0005322866,0.0006548681,0.0005400526,0.00024111068],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.023158757,0.00030938626,0.0014175648,0.0008855587,0.00088006206,0.00006123422,0.00060753163,0.00022753885,0.00009416416],"category_scores_gemma":[0.0037116706,0.0002135144,0.00006190013,0.0024965773,0.00024364323,0.00023316254,0.0009915386,0.0015806304,0.00011648059],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0000034703785,0.000011057539,0.00020012217,0.9466365,0.0000072926005,9.272185e-7,0.0046940884,2.7481887e-8,8.436142e-8,0.00046387763,0.0024442386,0.04553834],"study_design_scores_gemma":[0.00019458773,0.0000625004,0.0000055343057,0.50496846,0.000039117815,0.000001812634,0.00055099133,0.000012904109,4.6445404e-8,0.00010277651,0.49395257,0.00010868948],"about_ca_topic_score_codex":0.00023607087,"about_ca_topic_score_gemma":0.0104845995,"teacher_disagreement_score":0.49150833,"about_ca_system_score_codex":0.00055040076,"about_ca_system_score_gemma":0.007199152,"threshold_uncertainty_score":0.9984291},"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":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000141731025,0.98489964,0.006902746,0.0010346987,0.0034624226,0.0013143608,0.00084681367,0.00009778477,0.0014273411],"genre_scores_gemma":[0.0000315139,0.96748525,0.0059218383,0.0003566993,0.0006869445,0.0006541131,0.00057372183,0.00005340038,0.024236498],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9950184,0.0002801719,0.0012703217,0.002600428,0.00021860603,0.00061202265],"domain_scores_gemma":[0.9954284,0.002197884,0.00030077944,0.0018376864,0.00011722443,0.00011803218],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0057730107,0.00035619078,0.0017103135,0.0017499406,0.00007727605,0.00054383295,0.0034278142,0.00032075337,0.00005380268],"category_scores_gemma":[0.00912316,0.00021819219,0.00040066993,0.0017396886,0.000100656136,0.000053550353,0.0027816459,0.000266103,0.0014189914],"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":[0.0000023522305,0.000024784804,0.000022011907,0.00069610093,0.000061161205,0.000005690756,0.0000375204,7.7228594e-7,0.000002983239,0.009365949,0.07546997,0.9143107],"study_design_scores_gemma":[0.0000666556,0.00007271356,0.0000012639705,0.0012408355,0.00012789128,0.000004658655,0.00003282069,0.000396504,3.3949084e-7,0.019278938,0.97853833,0.00023903286],"about_ca_topic_score_codex":0.000028574968,"about_ca_topic_score_gemma":0.0002221075,"teacher_disagreement_score":0.9140717,"about_ca_system_score_codex":0.0001406336,"about_ca_system_score_gemma":0.00014033694,"threshold_uncertainty_score":0.99935853},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8835284,0.07866279,0.0005629791,0.031779062,0.00084930763,0.00247097,0.0008722056,0.000049491802,0.0012247949],"genre_scores_gemma":[0.9947151,0.0016540741,0.0004937467,0.00006920869,0.0021784161,0.000074230025,0.000082647945,0.000042383937,0.000690205],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9950641,0.00028645332,0.000797853,0.00086769275,0.0021070344,0.0008768811],"domain_scores_gemma":[0.9950607,0.001079742,0.00008548097,0.0010718163,0.0020569132,0.0006453811],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009400472,0.0001814825,0.0005176596,0.0009145503,0.00025020386,0.00087648793,0.0010541241,0.00015285755,0.000081288505],"category_scores_gemma":[0.003928725,0.00014637112,0.000108630935,0.0022856726,0.00028462987,0.00078141957,0.0008783362,0.0008665876,0.00001194299],"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":[0.02024388,0.0033182097,0.13017562,0.006922542,0.0011601386,0.0061095324,0.0012200031,0.00019795922,0.014168714,0.05206787,0.48670733,0.2777082],"study_design_scores_gemma":[0.022271587,0.013900915,0.30877683,0.009980941,0.00097171945,0.0012186348,0.0140349725,0.19963479,0.00072187936,0.08897926,0.338437,0.0010714668],"about_ca_topic_score_codex":0.00012097467,"about_ca_topic_score_gemma":0.00020516865,"teacher_disagreement_score":0.27663672,"about_ca_system_score_codex":0.000597572,"about_ca_system_score_gemma":0.0017118059,"threshold_uncertainty_score":0.8451993},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":"commentary","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.007807208,0.02333981,0.00026220438,0.94890505,0.017354272,0.00091574463,0.00004247493,0.00003742287,0.0013358179],"genre_scores_gemma":[0.026556399,0.01701762,0.0018114017,0.6420503,0.30897725,0.00007318837,0.0006113057,0.00012876223,0.0027737233],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9914783,0.00047011007,0.0035337321,0.00073369604,0.0033267243,0.00045739577],"domain_scores_gemma":[0.99083865,0.004445339,0.002296064,0.00032136322,0.0016536025,0.00044498878],"candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.004346788,0.0007732354,0.0026032638,0.0012263026,0.00015423188,0.000027747059,0.0005545154,0.00029516025,0.0025259904],"category_scores_gemma":[0.0073765586,0.0004275079,0.00042422832,0.00042397794,0.00095699955,0.0003040286,0.0001333862,0.005090207,0.000017105238],"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":[0.013204784,0.00016427034,0.00054998085,0.0010521719,0.0026162844,0.01405617,0.0010232371,3.7861506e-7,0.000025000214,0.000083831335,0.95519036,0.012033539],"study_design_scores_gemma":[0.04801162,0.016324088,0.0018500992,0.04101427,0.0025420869,0.002506207,0.0010509696,0.00003182052,0.000033180735,0.0003572931,0.8859346,0.00034375722],"about_ca_topic_score_codex":0.00000752453,"about_ca_topic_score_gemma":0.000003798077,"teacher_disagreement_score":0.3068547,"about_ca_system_score_codex":0.00048041533,"about_ca_system_score_gemma":0.0020784272,"threshold_uncertainty_score":0.99981767},"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":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.25194478,0.17860623,0.000021285941,0.56114507,0.00070827093,0.0020376495,0.00005132662,0.000011172611,0.0054742023],"genre_scores_gemma":[0.96693337,0.024570104,0.0000075232506,0.00037709362,0.00009763451,0.00008601362,0.000005410987,0.000007308281,0.007915563],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.97343946,0.012882228,0.0010726986,0.00067510817,0.011537444,0.00039308984],"domain_scores_gemma":[0.9911002,0.0014184363,0.00036256164,0.001926132,0.005047713,0.00014493622],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.27691504,0.000091963106,0.00028150508,0.000107547276,0.0006118762,0.00039432498,0.0035494682,0.000081182334,0.0022122783],"category_scores_gemma":[0.09536604,0.000033767552,0.00014888734,0.002954163,0.00039213328,0.0007960654,0.0013702792,0.0020538315,0.000042808286],"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":[0.000004422249,0.000043415337,0.018306876,0.0007724628,0.000023767887,1.0065443e-7,0.0024204575,0.0000040863893,0.00013422474,0.004659719,0.4699679,0.5036626],"study_design_scores_gemma":[0.000083700616,0.00003932952,0.048607174,0.0030909663,0.000029511488,0.000012976288,0.0012070299,0.0030266584,0.00046100206,0.028220449,0.9151257,0.000095480464],"about_ca_topic_score_codex":0.0007908361,"about_ca_topic_score_gemma":0.0003110161,"teacher_disagreement_score":0.7149886,"about_ca_system_score_codex":0.00012809917,"about_ca_system_score_gemma":0.0016826412,"threshold_uncertainty_score":0.99869984},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.3834761,0.004326119,0.116546266,0.21897282,0.0030381752,0.002077099,0.00007419782,0.00013754678,0.27135167],"genre_scores_gemma":[0.9988323,0.00037529352,0.00032006405,0.00008609577,0.00005374622,0.000013095848,0.0000010170428,0.000011325547,0.00030705193],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9859081,0.001418442,0.001538302,0.0013485334,0.008651773,0.0011348337],"domain_scores_gemma":[0.98680604,0.009604,0.00075706147,0.0014645549,0.0010963434,0.0002720249],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.10307218,0.0002191284,0.000620451,0.0043774927,0.0007633868,0.004341419,0.007400257,0.00015533518,0.00013592732],"category_scores_gemma":[0.03439563,0.00013396659,0.00018097056,0.02655848,0.010060229,0.006524354,0.0013618349,0.00093182345,0.000013554297],"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":[0.000014812614,0.00008943688,0.018611152,0.00016005697,0.000019234612,0.000011680931,0.009973641,0.000027036773,0.0012617958,0.95650387,0.0065730033,0.0067542996],"study_design_scores_gemma":[0.00033125884,0.00012000975,0.0028176683,0.00089914026,0.000013438459,0.00004474296,0.27611342,0.0063596456,0.0008198417,0.70391,0.008272677,0.00029813655],"about_ca_topic_score_codex":0.00023563232,"about_ca_topic_score_gemma":0.00032930536,"teacher_disagreement_score":0.6153562,"about_ca_system_score_codex":0.0009776802,"about_ca_system_score_gemma":0.0022460704,"threshold_uncertainty_score":0.99797016},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.029672869,0.0013181252,0.30392832,0.6544495,0.002518043,0.001488014,0.0049627027,0.00023136745,0.0014310718],"genre_scores_gemma":[0.7881018,0.011252982,0.17938638,0.017072806,0.0006042679,0.000018498886,0.0015604626,0.000035248675,0.001967556],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9978226,0.000029498618,0.00047631792,0.00053127616,0.00087814033,0.00026218267],"domain_scores_gemma":[0.9954103,0.00016517422,0.0006810476,0.0030445198,0.0006224802,0.00007643733],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.009123488,0.00010011317,0.00015271592,0.00063782255,0.00056066137,0.008154058,0.013631844,0.00008118479,0.0000029443381],"category_scores_gemma":[0.004509317,0.000069555936,0.000009651119,0.00080158835,0.00021413527,0.16732746,0.017758515,0.00030260475,0.0000054825664],"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":[0.000028008655,0.000059509886,0.012100167,0.0002442985,0.00031512888,0.0000040334867,0.009692818,0.000006235429,0.0006065609,0.028784137,0.6627182,0.28544092],"study_design_scores_gemma":[0.00024028482,0.00003976846,0.00077038055,0.00006137732,0.00003660932,0.00002788959,0.00065881776,0.40650296,0.000116646406,0.0023852298,0.5890806,0.000079439655],"about_ca_topic_score_codex":0.000017659135,"about_ca_topic_score_gemma":0.00004869014,"teacher_disagreement_score":0.75842893,"about_ca_system_score_codex":0.00008638751,"about_ca_system_score_gemma":0.0002610438,"threshold_uncertainty_score":0.9928756},"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":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19357145,0.25114435,0.007763782,0.3250867,0.19389606,0.017212056,0.002977851,0.000070458016,0.008277277],"genre_scores_gemma":[0.54643154,0.11612208,0.010389937,0.0059297723,0.31830025,0.00024390317,0.0011921171,0.0003342156,0.0010561922],"study_design_codex":"not_applicable","study_design_gemma":"qualitative","domain_scores_codex":[0.98573273,0.0016554402,0.0046773944,0.001262445,0.005474217,0.0011977613],"domain_scores_gemma":[0.9822575,0.009479414,0.002069694,0.0005561098,0.0045943535,0.0010429586],"candidate_categories":["metaepi_narrow","sts","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow"],"category_scores_codex":[0.018475898,0.0015858253,0.005056886,0.0029412487,0.0010394019,0.00018660586,0.0020135706,0.00013629312,0.0029789456],"category_scores_gemma":[0.0037164646,0.00072625553,0.00029324242,0.0017255764,0.012075461,0.0015217214,0.0023570757,0.0059171985,0.0000013257209],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0040862192,0.00035036693,0.0020249682,0.0006107477,0.007861092,0.0005668789,0.109219864,0.0000012858486,0.0019768816,0.017589798,0.8396036,0.016108284],"study_design_scores_gemma":[0.009367268,0.045931317,0.0025991546,0.01789309,0.006960485,0.00068670104,0.4602187,0.00008930594,0.0005709579,0.062764294,0.39215115,0.0007675741],"about_ca_topic_score_codex":0.0181713,"about_ca_topic_score_gemma":0.011297946,"teacher_disagreement_score":0.44745246,"about_ca_system_score_codex":0.0008641642,"about_ca_system_score_gemma":0.001321416,"threshold_uncertainty_score":0.999689},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9848421,0.0003296843,0.00033209208,0.0026205238,0.00046524234,0.00006869284,0.00002096069,0.0001650259,0.011155662],"genre_scores_gemma":[0.9991693,0.00015555538,0.00025242593,0.00015273987,0.00002351866,0.000004525029,0.0000016444427,0.000005879769,0.0002344313],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.99856824,0.000006630984,0.00008177029,0.0003472127,0.000653988,0.00034218622],"domain_scores_gemma":[0.9995431,0.00009452352,0.000008573084,0.0001154802,0.000025276233,0.00021308607],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0024383294,0.000088553126,0.00006270203,0.00009559607,0.00042465024,0.00044524667,0.00021818947,0.00006349651,0.0006148808],"category_scores_gemma":[0.000744106,0.00007658186,0.000010621779,0.00091580447,0.001912566,0.00041498637,0.0002559549,0.00032302458,0.00013556743],"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":[0.000007482853,0.000042519532,0.0051091616,0.0005917557,0.000010944054,0.00007514331,0.027190622,0.0005874352,0.31764266,0.6214191,0.0047691627,0.02255405],"study_design_scores_gemma":[0.00037775497,0.00021158629,0.5352676,0.000452345,0.000053688287,0.00036040775,0.004072126,0.15477344,0.01115879,0.01967937,0.27199093,0.0016019865],"about_ca_topic_score_codex":0.000041766347,"about_ca_topic_score_gemma":0.000008485812,"teacher_disagreement_score":0.6017397,"about_ca_system_score_codex":0.0002382223,"about_ca_system_score_gemma":0.00008322357,"threshold_uncertainty_score":0.7046929},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9368279,0.0007065671,0.0009343499,0.030127633,0.005980539,0.00040502657,0.00006340189,0.000041033265,0.024913533],"genre_scores_gemma":[0.99514586,0.00018831456,0.00022179235,0.0009117601,0.0009632884,0.000021448432,0.0000863767,0.000014341148,0.0024468235],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9913218,0.000971074,0.0015732282,0.0015843804,0.0041467166,0.00040281427],"domain_scores_gemma":[0.9911137,0.005128814,0.0010469554,0.0005909354,0.0016409894,0.00047857716],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","scholarly_communication"],"category_scores_codex":[0.08978585,0.0002608344,0.0005568701,0.0023530007,0.0004142485,0.009740722,0.010478885,0.00010532907,0.0071004513],"category_scores_gemma":[0.11878299,0.00014150351,0.000027178008,0.0041582137,0.0028061816,0.028649816,0.002267223,0.0012457244,0.000017945811],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00007366244,0.000033311782,0.9598632,0.0000031143056,0.000029051967,0.000037114616,0.0043167267,0.000019832962,0.0000763482,0.0022997663,0.024849927,0.008397978],"study_design_scores_gemma":[0.0008755445,0.00017405959,0.93273824,0.0002918078,0.0000064201836,0.00005297261,0.004831029,0.017504485,0.00002439587,0.021724608,0.021623988,0.00015246645],"about_ca_topic_score_codex":0.023015046,"about_ca_topic_score_gemma":0.006126396,"teacher_disagreement_score":0.05831794,"about_ca_system_score_codex":0.00028071072,"about_ca_system_score_gemma":0.00043037272,"threshold_uncertainty_score":0.9999076},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000003371255,0.9920424,0.000033572367,0.00072198414,0.00042286864,0.00038169368,0.00092563767,0.000035386514,0.0054330593],"genre_scores_gemma":[0.00051704666,0.98068035,0.00064087013,0.0003218713,0.0004139518,0.000055205914,0.015538409,0.0000822241,0.001750067],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9983644,0.00026060877,0.00040923135,0.0006044896,0.00015944467,0.00020185302],"domain_scores_gemma":[0.990527,0.00016098541,0.00015213777,0.009020133,0.000088569075,0.000051219657],"candidate_categories":["open_science","research_integrity"],"consensus_categories":["research_integrity"],"category_scores_codex":[0.00072017626,0.0003221783,0.00041140942,0.000053664087,0.0005693927,0.00018218473,0.009464479,0.0015710477,0.0000068207432],"category_scores_gemma":[0.00018193894,0.00021758334,0.00029668285,0.00023444186,0.0002027505,0.000004490692,0.004002824,0.0051869247,0.000056653753],"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":[0.000010960889,0.0010046049,0.00005284628,0.02825852,0.0025552416,0.00001320726,0.00078288035,0.000004272907,0.0042253113,0.03609251,0.5660515,0.36094815],"study_design_scores_gemma":[0.00004407991,0.000014400261,7.421703e-7,0.0009934758,0.0008217245,0.000010878628,0.00001238434,0.00001943293,0.000020611667,0.00006503183,0.99774975,0.00024750447],"about_ca_topic_score_codex":0.000025057901,"about_ca_topic_score_gemma":0.0009139146,"teacher_disagreement_score":0.43169823,"about_ca_system_score_codex":0.00002752955,"about_ca_system_score_gemma":0.00031542714,"threshold_uncertainty_score":0.9997251},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9267093,0.031970613,0.002050782,0.007074372,0.00095738296,0.0014266744,0.00017006874,0.0000055193836,0.029635284],"genre_scores_gemma":[0.99463594,0.0011376122,0.0003427307,0.00006782851,0.0001423345,0.00007318984,0.00003427319,0.000009983183,0.003556111],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.98320854,0.002815996,0.009625195,0.001089416,0.0030714748,0.00018935897],"domain_scores_gemma":[0.98976225,0.0013808748,0.0027141136,0.0014822698,0.004442728,0.00021774501],"candidate_categories":["metaresearch","scholarly_communication","insufficient_payload"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.060841683,0.00030527235,0.0027339596,0.0017672299,0.00009315343,0.008508168,0.002421628,0.0002153135,0.037781466],"category_scores_gemma":[0.055918798,0.00014009241,0.00055187393,0.0015494418,0.00018897478,0.0011442659,0.0040080533,0.00042493216,0.000040140214],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000008362115,0.000021350408,0.9578884,0.00066903635,0.00011380304,7.28777e-7,0.0010833113,0.00001965913,0.00007760779,0.0052840095,0.03206183,0.0027718479],"study_design_scores_gemma":[0.000106670865,0.000033041095,0.9179659,0.0002809399,0.00013667047,0.000019408963,0.0011206607,0.0017498806,0.0000046763507,0.077377856,0.0010698718,0.0001344616],"about_ca_topic_score_codex":0.0036692305,"about_ca_topic_score_gemma":0.00094998226,"teacher_disagreement_score":0.072093844,"about_ca_system_score_codex":0.00012904151,"about_ca_system_score_gemma":0.00020645797,"threshold_uncertainty_score":0.9925211},"labels":[{"model":"gemma","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"medium"},{"model":"gpt","categories":["bibliometrics","scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"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":"review","genre_gemma":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005557813,0.9894591,0.00090723427,0.00011849789,0.00031250875,0.0014461199,0.0018063539,0.000114339404,0.00027806146],"genre_scores_gemma":[0.008720779,0.98758835,0.0009299477,0.00045462465,0.00036247735,0.000024977406,0.0011906077,0.0002505068,0.00047773577],"study_design_codex":"design_other","study_design_gemma":"systematic_review","domain_scores_codex":[0.99619913,0.00014563007,0.0011091955,0.0011092649,0.0007954795,0.0006413139],"domain_scores_gemma":[0.9952344,0.00055025454,0.0006637819,0.0030170106,0.00007128252,0.00046328135],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.001896523,0.00065013906,0.002555017,0.00037165993,0.00015441919,0.00015811056,0.0011323192,0.00022297908,0.000023146911],"category_scores_gemma":[0.0006169428,0.00040263007,0.0003549635,0.0008495927,0.00008619946,0.0003430647,0.0013198297,0.0013464809,0.00006422408],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000072020644,0.00032517852,0.0006549087,0.06688598,0.00076178304,0.000025208656,0.0049925246,0.0001830985,0.00001916807,0.0005548922,0.0017296952,0.9237955],"study_design_scores_gemma":[0.002361848,0.0010623242,0.00048218836,0.5476752,0.019596737,0.00036406462,0.00761714,0.28596896,0.0000020641025,0.00063683256,0.13161653,0.0026161128],"about_ca_topic_score_codex":0.0009886633,"about_ca_topic_score_gemma":0.00002323985,"teacher_disagreement_score":0.9211794,"about_ca_system_score_codex":0.00072621816,"about_ca_system_score_gemma":0.0045075375,"threshold_uncertainty_score":0.9998426},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.38126272,0.08106502,0.0005365802,0.14015286,0.011539414,0.0035263912,0.0007520716,0.00015520187,0.38100976],"genre_scores_gemma":[0.99402237,0.00033473206,0.00004831774,0.00067090255,0.0015274669,0.00013149128,0.00007393097,0.000057109322,0.0031337026],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.996614,0.00036727023,0.0006931521,0.0003725307,0.0002819691,0.0016711183],"domain_scores_gemma":[0.9980667,0.00042306754,0.00057114847,0.0006104749,0.00027251805,0.00005607204],"candidate_categories":["sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.00827619,0.00027695126,0.00033894027,0.0002609686,0.0014872847,0.06746285,0.0057595083,0.00012090742,0.00016980128],"category_scores_gemma":[0.0005337469,0.0001992397,0.00014791213,0.000098940014,0.000050512186,0.006346461,0.0049025337,0.015820304,0.0000016026099],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.000032299344,0.000023386829,0.000160582,0.00013035059,0.0003782471,0.00022174105,0.011396827,0.0002063196,5.044301e-7,0.8913128,0.034639496,0.061497483],"study_design_scores_gemma":[0.000328424,0.000022416338,0.000017279415,0.00046583373,0.00006996821,0.0011020596,0.028818874,0.0020145301,5.909069e-7,0.8304466,0.13641769,0.00029575807],"about_ca_topic_score_codex":0.94896847,"about_ca_topic_score_gemma":0.9994404,"teacher_disagreement_score":0.61275965,"about_ca_system_score_codex":0.0061197258,"about_ca_system_score_gemma":0.046392594,"threshold_uncertainty_score":0.99981266},"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":"commentary","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006140576,0.078123845,0.043668434,0.6965059,0.002559029,0.002053656,0.000102431324,0.00045030925,0.1703958],"genre_scores_gemma":[0.7027314,0.10452271,0.015106554,0.16420856,0.00040715333,0.00018251479,0.0000504338,0.000036580055,0.012754121],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9931467,0.00064663216,0.0009102245,0.00052976434,0.0043982835,0.00036839867],"domain_scores_gemma":[0.94385815,0.053792283,0.0003621192,0.00082274794,0.00074933964,0.0004153908],"candidate_categories":["metaresearch","bibliometrics","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["bibliometrics","scholarly_communication","insufficient_payload"],"category_scores_codex":[0.021488162,0.00016450441,0.00020767504,0.017100506,0.000367546,0.0945832,0.0058103083,0.00010183448,0.0033844481],"category_scores_gemma":[0.1182745,0.000111713096,0.00007360465,0.08565622,0.00013435267,0.78238827,0.0044644773,0.00043485843,0.0011851377],"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":[0.0002779959,0.000035877023,0.0032042519,0.000095497264,0.000014519748,0.000026461421,0.0001037066,0.00006846697,0.000041154308,0.10917548,0.27065358,0.616303],"study_design_scores_gemma":[0.00011505995,0.00013365505,0.010684756,0.00074751064,0.0000118397,0.000021345832,0.00029238596,0.03061116,0.0008441566,0.0025803167,0.9537746,0.00018326138],"about_ca_topic_score_codex":0.00003062737,"about_ca_topic_score_gemma":6.698412e-8,"teacher_disagreement_score":0.6965908,"about_ca_system_score_codex":0.000027646916,"about_ca_system_score_gemma":0.0011157849,"threshold_uncertainty_score":0.99959254},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005035307,0.0025922735,0.0015983792,0.379195,0.000103011516,0.00039917065,0.0000041672183,0.000003565456,0.6110691],"genre_scores_gemma":[0.9693894,0.011661313,0.0015503964,0.0036360547,0.00002744185,0.0000012796072,0.000002169428,0.000002137194,0.0137298405],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985023,0.0005782316,0.0001691349,0.000078419784,0.00046108288,0.00021080201],"domain_scores_gemma":[0.9975331,0.0013933257,0.00019553376,0.00038364157,0.00027408006,0.00022034749],"candidate_categories":["sts","scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.009601862,0.00003837573,0.00006272627,0.00021763187,0.0017260931,0.006220408,0.0041044187,0.000021161206,0.000049621663],"category_scores_gemma":[0.0018529387,0.00002262684,0.000012479818,0.00072001596,0.0009194759,0.31543872,0.0020993003,0.00029666795,0.0000036256833],"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":[0.000080124395,0.000004979056,0.0004603383,0.000024183748,0.00002392119,0.0000072419143,0.0024526974,0.000008428542,0.000005906074,0.9656919,0.01931546,0.011924777],"study_design_scores_gemma":[0.00017053647,0.000072766496,0.0050637783,0.00002990652,0.000006327498,0.000029098857,0.0077878046,0.00048584293,0.000117736156,0.0009758486,0.9852328,0.00002756842],"about_ca_topic_score_codex":0.0054596523,"about_ca_topic_score_gemma":0.0029246,"teacher_disagreement_score":0.96591735,"about_ca_system_score_codex":0.000056515684,"about_ca_system_score_gemma":0.002930137,"threshold_uncertainty_score":0.9995735},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.09324275,0.00051193166,0.00076999364,0.14765683,0.0011087839,0.0008119773,0.00014723379,0.000022484319,0.755728],"genre_scores_gemma":[0.9848299,0.00012930532,0.0002650993,0.008077651,0.00012102804,0.000007877646,0.000034360153,0.0000072989824,0.0065274714],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9990371,0.000063560394,0.00044241748,0.00007910152,0.00029426804,0.00008357931],"domain_scores_gemma":[0.99911916,0.00037530012,0.00032416414,0.00011999374,0.000037783957,0.000023586803],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00014469563,0.000082081424,0.00016835085,0.00004544284,0.00011121803,0.00087932707,0.0009945013,0.000005372223,0.00047835484],"category_scores_gemma":[0.000014518381,0.000046443536,0.00024512142,0.00013686826,0.00037321562,0.0043157185,0.00031533258,0.0001304905,0.00002476321],"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":[0.000060555172,0.0001275338,0.0023591644,0.00008470178,0.00020985238,0.0000035261562,0.01828379,0.0000959456,0.0000031782736,0.06354254,0.91077554,0.004453657],"study_design_scores_gemma":[0.00022667156,0.0002167305,0.014357237,0.00014221352,0.00002773618,0.0000032151327,0.0036572968,0.000044367185,0.00005189932,0.03491909,0.94628865,0.0000649182],"about_ca_topic_score_codex":0.000027493372,"about_ca_topic_score_gemma":0.000016800635,"teacher_disagreement_score":0.89158714,"about_ca_system_score_codex":0.0000075508838,"about_ca_system_score_gemma":0.000100556084,"threshold_uncertainty_score":0.8479371},"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":"review","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"commentary","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.008349449,0.03987318,0.0006225951,0.83483344,0.06654534,0.0024008006,0.0012250737,0.0017951945,0.044354916],"genre_scores_gemma":[0.31751996,0.3598473,0.017120086,0.01900837,0.012659076,0.005478108,0.028701873,0.001510969,0.23815425],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99649346,0.000053707743,0.0006299605,0.0019073255,0.00038489397,0.0005306388],"domain_scores_gemma":[0.9978404,0.00013756982,0.000051943167,0.0018193112,0.000055183882,0.00009560977],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0072748833,0.0002935974,0.00038457225,0.00033060057,0.00008374252,0.00021563789,0.000587511,0.00019652428,0.0015095741],"category_scores_gemma":[0.0011359636,0.00021219754,0.00008439477,0.0019766507,0.00017483499,0.00016871549,0.00028163329,0.0010440858,0.00014579335],"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":[0.0000031289542,0.000089495625,0.000030710216,0.00977319,0.000017603212,0.0000063120283,0.00077507977,0.00034467704,0.0022780746,0.00012539898,0.24646541,0.7400909],"study_design_scores_gemma":[0.000076943594,0.0000069994708,0.0013751029,0.0021150447,0.00003148724,0.0000022083082,0.00007950435,0.018166667,0.00014935524,0.00025400167,0.97741175,0.0003309606],"about_ca_topic_score_codex":0.0011760266,"about_ca_topic_score_gemma":0.0018512309,"teacher_disagreement_score":0.8158251,"about_ca_system_score_codex":0.0005246016,"about_ca_system_score_gemma":0.00022420286,"threshold_uncertainty_score":0.9994032},"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":"codex-gemma-dda1882f352a","genre_candidate":"review","genre_consensus":"review","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[3.2060413e-7,0.9660246,0.00013004807,0.00021506251,0.00014368258,0.0046277875,0.00008348509,0.000012167307,0.028762866],"genre_scores_gemma":[0.0000058017363,0.9964747,0.00016544422,0.0005439337,0.00001939148,0.00081410125,0.00022610517,0.000005423802,0.001745089],"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","domain_scores_codex":[0.9979845,0.00030455625,0.0007125945,0.0006165141,0.000111507805,0.00027030616],"domain_scores_gemma":[0.999078,0.00016191517,0.00022621782,0.0005111807,0.0000046510936,0.000018039165],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0017957484,0.00023814372,0.0018358021,0.000027333677,0.00013492684,0.000032149695,0.002445876,0.00014412706,0.00030154118],"category_scores_gemma":[0.00009273894,0.00015045037,0.00012743233,0.00034079145,0.0001267915,0.00017774421,0.010414371,0.000095761134,0.000058668356],"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":[0.000006505838,0.00032910757,0.0006099954,0.34720838,0.00032364976,0.000008062693,0.000012496204,0.0000017865731,3.285622e-9,0.0016456386,0.31713933,0.33271503],"study_design_scores_gemma":[0.000133483,0.000031670334,0.000064016946,0.027298018,0.0005108973,8.81989e-7,0.000014749461,0.000005552025,1.1684889e-9,0.00013219651,0.9716603,0.0001482025],"about_ca_topic_score_codex":0.00043141653,"about_ca_topic_score_gemma":0.00017706651,"teacher_disagreement_score":0.654521,"about_ca_system_score_codex":0.0002078759,"about_ca_system_score_gemma":0.000022825465,"threshold_uncertainty_score":0.99758923},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06502955,0.00053088553,0.013811293,0.042458307,0.00013266531,0.004083797,0.00011141567,0.00003136667,0.8738107],"genre_scores_gemma":[0.9898293,0.0005231325,0.004134754,0.00014481427,0.000012033606,0.00009733163,0.0000011301179,0.0000025149268,0.0052549983],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9926869,0.0009295061,0.0005218373,0.0009873763,0.0037818637,0.0010925167],"domain_scores_gemma":[0.9897206,0.0029186194,0.00038139054,0.003981002,0.0027607626,0.00023765926],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","sts","scholarly_communication","open_science"],"category_scores_codex":[0.04059833,0.00011607334,0.0004126214,0.0025266574,0.0016928572,0.012519553,0.07453593,0.000017755136,0.000015084993],"category_scores_gemma":[0.015842032,0.00008152277,0.00006199402,0.05410739,0.008922188,0.027452525,0.06631497,0.0003751805,0.000013222568],"study_design_candidate":"theoretical_or_conceptual","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":[0.000033745633,0.0001060673,0.001591085,0.000014399283,0.0001854516,0.0000020884056,0.00034227138,0.00012842374,0.0045664404,0.98514575,0.0020426214,0.0058416757],"study_design_scores_gemma":[0.002769872,0.00069541996,0.39257473,0.00035467796,0.00015272375,0.0000023877371,0.0026265166,0.21060967,0.11412397,0.07878693,0.19671902,0.0005840657],"about_ca_topic_score_codex":0.0278466,"about_ca_topic_score_gemma":0.001343848,"teacher_disagreement_score":0.92479974,"about_ca_system_score_codex":0.0004778864,"about_ca_system_score_gemma":0.01262704,"threshold_uncertainty_score":0.9996068},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.041721366,0.106397234,0.004360729,0.22051018,0.0010659499,0.0032391376,0.000003174535,0.000040305094,0.62266195],"genre_scores_gemma":[0.94679254,0.048200976,0.0004562695,0.0011042326,0.00003522327,0.00018126758,0.000053594078,0.0000041902617,0.0031716835],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976952,0.0009313518,0.0003514834,0.000284108,0.00055895624,0.00017891715],"domain_scores_gemma":[0.9975426,0.0012500352,0.00014687485,0.00037692138,0.0006294935,0.00005406774],"candidate_categories":["bibliometrics","scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.0066763232,0.00008329768,0.00018214982,0.0029249408,0.00045235784,0.0015138438,0.0057097003,0.000029050489,0.000120166944],"category_scores_gemma":[0.0035816133,0.00004907637,0.00003088095,0.02344408,0.0001039159,0.0007702126,0.0040318505,0.0006664048,0.00002741363],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000028564165,0.00022647543,0.029550487,0.00050017383,0.000044672113,0.000001640353,0.0010336931,0.000084393105,0.000028578479,0.5325326,0.03547949,0.40048927],"study_design_scores_gemma":[0.00045587856,0.00006182705,0.058298066,0.005803462,0.0000064603464,0.000006018622,0.008818882,0.005237357,0.000063473846,0.013366728,0.90776485,0.00011698715],"about_ca_topic_score_codex":0.0011928416,"about_ca_topic_score_gemma":0.000038552862,"teacher_disagreement_score":0.9050712,"about_ca_system_score_codex":0.00009074391,"about_ca_system_score_gemma":0.0008943188,"threshold_uncertainty_score":0.9996699},"labels":[{"model":"gemma","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics"],"domain":null,"study_design":"design_other","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"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":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00050923385,0.0006664832,0.88327044,0.112665534,0.0000628349,0.0019614305,0.000017944756,0.0002234696,0.00062265096],"genre_scores_gemma":[0.022228621,0.00049186114,0.91316223,0.061609287,0.000010863704,0.0012585036,0.00016349775,0.0000121762305,0.001062971],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99820966,0.0001847849,0.00080012705,0.00028198314,0.00022065737,0.0003027768],"domain_scores_gemma":[0.99067056,0.0075824084,0.0003747431,0.0007248424,0.00053394237,0.0001135162],"candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.00274093,0.00016895444,0.00022134697,0.0009236661,0.00096115173,0.011562093,0.0018666887,0.00012141709,0.0000074178256],"category_scores_gemma":[0.015056381,0.00018422806,0.000025774483,0.0018181126,0.00008868534,0.4152171,0.0029915934,0.00027343386,0.0000034324248],"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":[0.00014959896,0.000051907737,0.001165204,0.00033484434,0.00001618532,0.0000012090785,0.0009753588,0.000031002797,0.0000020873404,0.84657985,0.00664156,0.14405118],"study_design_scores_gemma":[0.0006494365,0.00017441648,0.0052347914,0.00041821634,0.000018248536,0.0000054849497,0.0067484425,0.0017035286,0.00033560282,0.004010278,0.980439,0.0002625674],"about_ca_topic_score_codex":0.000015435327,"about_ca_topic_score_gemma":3.67376e-7,"teacher_disagreement_score":0.97379744,"about_ca_system_score_codex":0.00006706129,"about_ca_system_score_gemma":0.0018557007,"threshold_uncertainty_score":0.99324024},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005519388,0.00006605681,0.00032804106,0.022387164,0.00021635032,0.00051203504,0.00020257676,0.00007089247,0.9706975],"genre_scores_gemma":[0.9626108,0.000082791754,0.00036576536,0.0031980176,0.00038221874,0.000016106775,0.0007730287,0.000007228524,0.032564078],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984024,0.00027210673,0.00019396916,0.0005449092,0.00030210408,0.00028452097],"domain_scores_gemma":[0.9981308,0.00011545486,0.00006932709,0.0013292235,0.00008611172,0.00026907714],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0018983462,0.00012059769,0.00021337446,0.00017424335,0.0008627835,0.0009662716,0.003755249,0.00020795023,0.0055630705],"category_scores_gemma":[0.0003714277,0.0001218729,0.000052126285,0.00015085032,0.00011342112,0.00017005292,0.00084569573,0.0003175132,0.00016819773],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000013099892,0.00019823667,0.0066399807,0.00008306868,0.00029741987,0.000025660827,0.032250293,0.0058413814,7.1674623e-7,0.40062845,0.22404352,0.3299782],"study_design_scores_gemma":[0.0001626199,0.000033839293,0.008048279,0.00029628194,0.00016756538,8.710454e-8,0.026907247,0.006735341,0.0000059101712,0.119918175,0.83669317,0.0010314577],"about_ca_topic_score_codex":0.99446595,"about_ca_topic_score_gemma":0.99907494,"teacher_disagreement_score":0.9570914,"about_ca_system_score_codex":0.00061036675,"about_ca_system_score_gemma":0.004987058,"threshold_uncertainty_score":0.99534595},"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":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.052345224,0.00057864224,0.62297237,0.031120619,0.0056506027,0.0016654223,0.0004323041,0.003567835,0.28166696],"genre_scores_gemma":[0.8490533,0.000100809695,0.09289783,0.04453002,0.002222147,0.00006522153,0.0015597909,0.00003267634,0.009538175],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976364,0.0001700602,0.0005798108,0.0006528771,0.00060886174,0.00035199334],"domain_scores_gemma":[0.99702734,0.00007898857,0.0002957331,0.002332736,0.000049847677,0.00021534806],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.00032622396,0.00030429926,0.000282007,0.00009366393,0.00023517555,0.004438512,0.009016497,0.00024322594,0.0003309185],"category_scores_gemma":[0.000011606589,0.00017564752,0.00015864441,0.0003977573,0.00006247631,0.0127213895,0.0023623942,0.00048385852,0.0007649052],"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":[0.000008573095,0.000112184585,0.005114392,0.000031900432,0.00004580261,0.00001640364,0.023603115,0.0018114796,0.000031464224,0.063034594,0.78649735,0.11969276],"study_design_scores_gemma":[0.00045836414,0.0001815419,0.08025243,0.0001670535,0.0000066279895,0.000015733263,0.0018666539,0.37808537,0.0016238925,0.055814166,0.48020956,0.0013186056],"about_ca_topic_score_codex":0.00019997924,"about_ca_topic_score_gemma":0.000003714021,"teacher_disagreement_score":0.7967081,"about_ca_system_score_codex":0.000018450794,"about_ca_system_score_gemma":0.00039838266,"threshold_uncertainty_score":0.99659497},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15803678,0.0013689524,0.017685167,0.031112112,0.005151328,0.0004868249,0.000008376648,0.000016915998,0.7861335],"genre_scores_gemma":[0.9715212,0.000033622873,0.0013861356,0.00041924205,0.00062699354,0.000002298284,4.0561105e-7,0.0000058571295,0.026004247],"study_design_codex":"not_applicable","study_design_gemma":"theoretical_or_conceptual","domain_scores_codex":[0.9955549,0.00013777411,0.00049396046,0.00039111805,0.0032039315,0.00021827545],"domain_scores_gemma":[0.99685633,0.0006437595,0.0003425447,0.0006461218,0.0012886487,0.00022259228],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.009766743,0.00011457232,0.00019252184,0.0007836206,0.00032060492,0.0003756262,0.0017643315,0.00008216481,0.002002022],"category_scores_gemma":[0.0017629195,0.000055984743,0.00008063893,0.00060043344,0.0011729853,0.0002370066,0.0013299857,0.00064745947,0.000071782786],"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":[0.00037710165,0.00007118667,0.00041161792,0.000011350043,0.00002139859,0.000023100856,0.0028720407,0.00014562636,0.0015206804,0.29278713,0.6451657,0.056593046],"study_design_scores_gemma":[0.0001687837,0.00043753075,0.0023336613,0.000082908125,0.0000051267994,0.000023050417,0.0013534883,0.0012278046,0.0015697355,0.958885,0.03376037,0.00015252737],"about_ca_topic_score_codex":0.000014058843,"about_ca_topic_score_gemma":0.000029592462,"teacher_disagreement_score":0.81348443,"about_ca_system_score_codex":0.00026921253,"about_ca_system_score_gemma":0.002337519,"threshold_uncertainty_score":0.9989103},"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":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0034719957,0.00018526257,0.0000042952247,0.15596874,0.0010357852,0.00022459733,0.000058621645,0.00010690927,0.8389438],"genre_scores_gemma":[0.10302722,0.00024863528,0.00017926292,0.023895033,0.0042769094,0.000027117349,0.000043044347,0.000019009387,0.86828375],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99801093,0.00028076128,0.00022217102,0.0002970018,0.0006114944,0.0005776647],"domain_scores_gemma":[0.9984785,0.0007091615,0.00012290504,0.00027514802,0.00007557137,0.0003387222],"candidate_categories":["sts","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0008077517,0.00018014104,0.00021382733,0.0002079446,0.0019006545,0.00038791614,0.0006078328,0.00043978944,0.003909263],"category_scores_gemma":[0.00050497625,0.0001230285,0.00015454764,0.00046193993,0.00014775587,0.00012830808,0.00023690412,0.00068899256,0.0010121523],"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":[0.000001848234,0.000016213873,0.0030635572,0.0000042903457,0.00004889748,0.0000022731895,0.015853038,0.0000014112265,6.00766e-8,0.45215032,0.5280687,0.0007894126],"study_design_scores_gemma":[0.00002327722,0.000005573766,0.0035370302,0.000013180026,0.0000074069712,3.650999e-8,0.00064097234,4.440511e-7,4.1433165e-7,0.012938859,0.98266745,0.00016535711],"about_ca_topic_score_codex":0.8461668,"about_ca_topic_score_gemma":0.89821297,"teacher_disagreement_score":0.45459875,"about_ca_system_score_codex":0.0035426512,"about_ca_system_score_gemma":0.0025550372,"threshold_uncertainty_score":0.9997657},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9778116,0.00076629827,0.0008032302,0.005638955,0.00009716747,0.0016765232,0.00000691587,0.000026444717,0.013172866],"genre_scores_gemma":[0.99348843,0.000511287,0.0022287585,0.00010084578,0.0000877477,0.00010858946,0.000009794044,0.00002645843,0.0034381156],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.9900614,0.003990131,0.0012037061,0.0007891332,0.0032653594,0.0006902562],"domain_scores_gemma":[0.94201875,0.05349746,0.00014486186,0.0016180016,0.0019717412,0.0007492019],"candidate_categories":["metaresearch","research_integrity"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.03104547,0.00014975012,0.00060458644,0.0011843443,0.000147895,0.000024654944,0.0007169732,0.00050393236,0.00035850555],"category_scores_gemma":[0.15521082,0.00013609215,0.00012677214,0.001990288,0.0015048843,0.000045064644,0.0009618827,0.0043667615,0.00003842859],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0046342188,0.015067716,0.23147608,0.0014955355,0.00053221313,0.00026185406,0.47232667,0.00015353704,0.0015251855,0.26866764,0.0013551152,0.002504237],"study_design_scores_gemma":[0.005126772,0.0009798857,0.17584504,0.0005625062,0.000028870216,0.0000031169936,0.7210073,0.0018002968,0.00004927003,0.094364434,0.000116480325,0.000116050294],"about_ca_topic_score_codex":0.0026883152,"about_ca_topic_score_gemma":0.023553759,"teacher_disagreement_score":0.2486806,"about_ca_system_score_codex":0.0009536514,"about_ca_system_score_gemma":0.00744076,"threshold_uncertainty_score":0.9981861},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0139865605,0.0003663353,0.0026017276,0.2617511,0.0010839907,0.00079257943,0.000063942054,0.00066059147,0.7186932],"genre_scores_gemma":[0.8272645,0.00026826997,0.008635833,0.013446119,0.0062110717,0.000027845877,0.0042010727,0.00009440388,0.13985087],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982782,0.000034206285,0.0004318378,0.0004392317,0.0005290068,0.00028752402],"domain_scores_gemma":[0.9989656,0.000035837198,0.00012181909,0.0005350849,0.000054447555,0.000287209],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00018467895,0.00024568604,0.00042575534,0.00031368085,0.000046039106,0.000052922904,0.00015803943,0.0009201417,0.0025488464],"category_scores_gemma":[0.000021254244,0.00017706903,0.00023420599,0.00020956623,0.00044561055,0.000058121273,0.0006241457,0.0011442098,0.0001306144],"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":[0.0001512573,0.0008393112,0.0076115313,0.0024226198,0.00035455704,0.0003680852,0.00044157918,0.000040007744,0.0008101571,0.37897488,0.50104445,0.106941566],"study_design_scores_gemma":[0.0036271159,0.0003436797,0.12354845,0.0008484375,0.0003554998,0.0001918648,0.000048221922,0.003158302,0.0015279602,0.30289227,0.5624731,0.0009850515],"about_ca_topic_score_codex":0.0010946272,"about_ca_topic_score_gemma":0.000023651768,"teacher_disagreement_score":0.81327796,"about_ca_system_score_codex":0.00006424999,"about_ca_system_score_gemma":0.0015195097,"threshold_uncertainty_score":0.99836296},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0032777335,0.00012570665,0.018105853,0.04106198,0.0038821402,0.00040302498,0.00004373915,0.00085437746,0.93224543],"genre_scores_gemma":[0.70517534,0.00014723204,0.039091524,0.03459584,0.0016667078,0.00007537892,0.00021352398,0.000024657282,0.21900983],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99839747,0.000056791167,0.00038834603,0.00033067403,0.0004463266,0.00038041806],"domain_scores_gemma":[0.99867743,0.000046009798,0.00025943542,0.0006554614,0.00004564628,0.0003159955],"candidate_categories":["scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00028445973,0.0001632515,0.00018197838,0.0003805985,0.00012585025,0.002528491,0.002539063,0.00025876495,0.0006532422],"category_scores_gemma":[0.000024469555,0.00014267187,0.00009563454,0.0003492383,0.000007781519,0.00977477,0.0010448578,0.00027737123,0.001778495],"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":[4.2180613e-7,0.000013476795,0.024268497,0.000035130266,0.00003422661,0.000006762244,0.0038336464,0.0009034299,0.0000039544075,0.056811653,0.8826817,0.031407088],"study_design_scores_gemma":[0.00007893947,0.000011490422,0.003700328,0.000031084477,0.0000011445169,0.0000046731043,0.00009958011,0.054998156,0.0006428383,0.0027280925,0.9373284,0.0003752829],"about_ca_topic_score_codex":0.017844643,"about_ca_topic_score_gemma":0.0036776124,"teacher_disagreement_score":0.7132356,"about_ca_system_score_codex":0.00030866763,"about_ca_system_score_gemma":0.0019386897,"threshold_uncertainty_score":0.99899876},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.19362408,0.0038890508,0.12508462,0.085023195,0.0069793854,0.002560486,0.00020240877,0.000021174626,0.5826156],"genre_scores_gemma":[0.8996482,0.000050003666,0.0854501,0.004717996,0.000037573947,0.000014422666,0.000011429781,0.0000034466955,0.010066854],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9982086,0.00006733631,0.0005585899,0.00015846184,0.00057595206,0.00043109216],"domain_scores_gemma":[0.99477494,0.00048444438,0.000292294,0.00018040411,0.0025563594,0.001711537],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication"],"consensus_categories":["bibliometrics","scholarly_communication"],"category_scores_codex":[0.009494694,0.00007753376,0.000161258,0.040606193,0.0016601903,0.020929761,0.0029339592,0.00005433787,0.00026693047],"category_scores_gemma":[0.014024946,0.000058923015,0.000026713036,0.055569842,0.00043741643,0.027289229,0.0006101858,0.0001241365,0.0000030692931],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000010296631,0.0000051991474,0.021980844,0.000019599962,0.000005287919,0.0000024598014,0.000255452,0.00010119572,0.0000077408395,0.22137366,0.07058112,0.68565714],"study_design_scores_gemma":[0.00029486822,0.000048325648,0.11048402,0.000037810318,0.0000042031247,0.000028891996,0.0009757962,0.01026248,0.00012960623,0.016093694,0.8615426,0.000097709075],"about_ca_topic_score_codex":0.06665031,"about_ca_topic_score_gemma":0.009360983,"teacher_disagreement_score":0.7909615,"about_ca_system_score_codex":0.0005924775,"about_ca_system_score_gemma":0.041599613,"threshold_uncertainty_score":0.9996395},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"incentives","study_design":"not_applicable","genre":"empirical","about_ca_system":true,"about_ca_topic":true,"confidence":"high"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":true,"about_ca_topic":true,"confidence":"high"}],"label_agreement":"split"},{"id":"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.95366204,0.0001810167,0.0010649617,0.009391448,0.00023904679,0.001028144,0.0000118018415,0.000010462751,0.03441111],"genre_scores_gemma":[0.99436206,0.00019465691,0.0017470097,0.0020266955,0.000033202854,0.00006440726,8.372703e-7,0.0000053660024,0.0015657492],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.98846227,0.00046256572,0.0007017952,0.0017939897,0.007081799,0.0014975733],"domain_scores_gemma":[0.9928499,0.0018875556,0.00031169786,0.0011449787,0.0030826656,0.00072321796],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","sts","open_science"],"category_scores_codex":[0.11282765,0.00017962912,0.00039731548,0.0058943834,0.004490667,0.04906435,0.018838493,0.0000619434,0.000049680475],"category_scores_gemma":[0.0644308,0.00010765649,0.000038494054,0.24454923,0.009819855,0.006694152,0.049590588,0.00032802785,0.000014477336],"study_design_candidate":"observational","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00005709914,0.00023540718,0.31244734,0.000005177899,0.000009480303,0.0000034865022,0.012811306,0.000110287096,0.0007868477,0.032277845,0.0015110392,0.6397447],"study_design_scores_gemma":[0.0019187176,0.00015577264,0.8958852,0.000017434588,0.0000033459035,0.0000021533733,0.012942603,0.073874906,0.0002568165,0.0095821265,0.005146863,0.00021403468],"about_ca_topic_score_codex":0.010294846,"about_ca_topic_score_gemma":0.00026011968,"teacher_disagreement_score":0.63953066,"about_ca_system_score_codex":0.00041139234,"about_ca_system_score_gemma":0.0040828427,"threshold_uncertainty_score":0.99680537},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008979428,0.0033897248,0.00007555297,0.29941,0.013546313,0.00027548685,0.00006292792,0.000118332355,0.68222374],"genre_scores_gemma":[0.17628217,0.0009254476,0.00008559328,0.009794976,0.0014285125,0.00028214412,0.00031095324,0.00003057984,0.8108596],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9974221,0.00011574979,0.0006729531,0.0006878393,0.00070729986,0.0003940694],"domain_scores_gemma":[0.99776584,0.00014571004,0.00035956647,0.00038711636,0.0011552128,0.0001865369],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00047419925,0.00032404004,0.00046703892,0.0005119119,0.0013181702,0.0025190252,0.008430641,0.00011102589,0.065055154],"category_scores_gemma":[0.00041571565,0.0002810205,0.00030466687,0.00011867927,0.0005549044,0.005130646,0.003970005,0.0003780137,0.00033783854],"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":[0.000016260801,0.0002028232,0.0002999786,0.000008203939,0.00043467025,0.000006794991,0.006707444,9.452023e-7,0.000004323007,0.37815458,0.6089554,0.0052085933],"study_design_scores_gemma":[0.00058667664,0.000015833819,0.00013076437,0.0000334325,0.00004370307,0.0000010594862,0.001726512,0.000106231775,6.924193e-7,0.0050619836,0.9919496,0.000343517],"about_ca_topic_score_codex":0.0064441427,"about_ca_topic_score_gemma":0.018450722,"teacher_disagreement_score":0.3829942,"about_ca_system_score_codex":0.03015908,"about_ca_system_score_gemma":0.0006886694,"threshold_uncertainty_score":0.999982},"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":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0031943268,0.00007779185,0.98328835,0.0008858062,0.0010766136,0.0058971546,0.001126001,0.000014638194,0.00443932],"genre_scores_gemma":[0.5011929,0.00594837,0.42353037,0.0037108238,0.0022827613,0.006893348,0.007051557,0.000103105755,0.04928672],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9935165,0.0032392573,0.0016885429,0.0008599247,0.00042853237,0.00026724132],"domain_scores_gemma":[0.9882935,0.0072704735,0.00083144906,0.0031151946,0.0004189393,0.00007041054],"candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["metaresearch","scholarly_communication","open_science"],"category_scores_codex":[0.1993732,0.000081201135,0.0005458704,0.00015507746,0.00013726582,0.004524844,0.015429646,0.000055717675,0.000021182566],"category_scores_gemma":[0.06758762,0.000068614754,0.000072056995,0.0003880903,0.00006285141,0.020373102,0.022127934,0.00016759535,0.000009268802],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00044090653,0.000048975304,0.9477964,0.000056396497,0.0004907018,0.000004953268,0.0000028863178,0.000084835796,0.000007250888,0.010773008,0.033426143,0.0068675335],"study_design_scores_gemma":[0.001417245,0.00007455738,0.9317436,0.000035557932,0.00000984922,0.0000014625801,0.000004988499,0.011600867,0.0000066229854,0.0018566566,0.05317705,0.00007150314],"about_ca_topic_score_codex":0.00042379467,"about_ca_topic_score_gemma":0.00015575677,"teacher_disagreement_score":0.55975795,"about_ca_system_score_codex":0.000026516265,"about_ca_system_score_gemma":0.0005114297,"threshold_uncertainty_score":0.99650854},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0677459,0.10252166,0.0008683935,0.80140984,0.0057763616,0.0005092878,0.00055596454,0.000031179734,0.020581445],"genre_scores_gemma":[0.90973705,0.051154666,0.0024262425,0.023024658,0.0024710628,0.000013102505,0.000026446402,0.00008114405,0.0110656],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","domain_scores_codex":[0.9936207,0.0020725247,0.0014663665,0.00045954913,0.0012368656,0.0011439987],"domain_scores_gemma":[0.9741439,0.016695315,0.00053177815,0.00058884756,0.005780858,0.0022592796],"candidate_categories":["metaresearch","metaepi_narrow","sts","research_integrity","insufficient_payload"],"consensus_categories":["research_integrity"],"category_scores_codex":[0.015370016,0.00041692142,0.00085447496,0.0008975694,0.000454527,0.00035883224,0.0009947392,0.0033800635,0.0011649105],"category_scores_gemma":[0.053981505,0.0004088728,0.00042213063,0.0007818595,0.0028136692,0.00056647306,0.00016944851,0.016835794,0.000037639835],"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":[0.00048392903,0.00031053636,0.21560049,0.006557584,0.001361502,0.0033663386,0.043108914,0.0001506564,0.0010871142,0.5560081,0.04448322,0.1274816],"study_design_scores_gemma":[0.0037145994,0.0032300735,0.40248382,0.088247515,0.0014354547,0.000590204,0.008171353,0.00031527254,0.01180642,0.20401725,0.27478793,0.0012000869],"about_ca_topic_score_codex":0.45565128,"about_ca_topic_score_gemma":0.90217817,"teacher_disagreement_score":0.8419912,"about_ca_system_score_codex":0.0030190058,"about_ca_system_score_gemma":0.08682572,"threshold_uncertainty_score":0.9999001},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.22363074,0.0022542314,0.03989788,0.25417992,0.00170849,0.00062757556,0.00040300668,0.00005266807,0.47724548],"genre_scores_gemma":[0.9751857,0.0018365722,0.020671682,0.0021218194,0.0000876833,0.0000011199095,0.000024725652,0.000003957075,0.000066749504],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984514,0.00016854079,0.0006463222,0.00019140104,0.00042858883,0.00011373989],"domain_scores_gemma":[0.99429643,0.002482686,0.0014744966,0.00031376796,0.0013287022,0.00010390196],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0030921919,0.000093380964,0.0003603432,0.0006077801,0.000040439652,0.00019051113,0.0019972702,0.00003668417,0.000011081369],"category_scores_gemma":[0.0013987279,0.0000752244,0.000036060042,0.0009854535,0.00003501413,0.005409753,0.001963087,0.00012443107,0.0000036259237],"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":[0.010352495,0.0024237547,0.012226231,0.0005181837,0.05678336,0.00076677365,0.14323498,0.002924981,0.0044193505,0.61239296,0.008663846,0.1452931],"study_design_scores_gemma":[0.0073056654,0.001117422,0.17875122,0.00059783255,0.0029515594,0.0048589166,0.052144647,0.11625002,0.0016340117,0.016730214,0.61661375,0.0010447685],"about_ca_topic_score_codex":0.0015560889,"about_ca_topic_score_gemma":0.0006919388,"teacher_disagreement_score":0.75155497,"about_ca_system_score_codex":0.00008798521,"about_ca_system_score_gemma":0.00043716384,"threshold_uncertainty_score":0.39219403},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9761334,0.0006810075,0.0022134243,0.010416481,0.00063030823,0.002545111,0.0004861572,0.000050081697,0.0068440232],"genre_scores_gemma":[0.9916394,0.0005919557,0.0036628004,0.00043957002,0.0012134183,0.00003351125,0.001054146,0.000020180514,0.0013450227],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99280137,0.0003750131,0.0031821707,0.0020923917,0.0012306954,0.00031834745],"domain_scores_gemma":[0.9216621,0.06725367,0.0010363016,0.008804794,0.0007871251,0.0004559908],"candidate_categories":["metaresearch","research_integrity"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.04824465,0.000183999,0.0009154758,0.0002680921,0.00012189376,0.00013507376,0.001996161,0.00036519405,0.000038233873],"category_scores_gemma":[0.33076406,0.00015248249,0.000027757582,0.00052141736,0.0011260547,0.0008399502,0.0035388365,0.0024632602,0.0000011013151],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.012311712,0.00048092543,0.9828653,0.0005437162,0.00017406531,0.000047309408,0.00016558696,5.624705e-7,0.000010770409,0.0019354405,0.0009746246,0.00049000053],"study_design_scores_gemma":[0.05663314,0.0016178532,0.9340742,0.0012789042,0.0002573614,0.000023293058,0.000948666,0.0014288212,0.000001284958,0.002320827,0.0013093833,0.00010626164],"about_ca_topic_score_codex":0.0016837695,"about_ca_topic_score_gemma":0.008668402,"teacher_disagreement_score":0.2825194,"about_ca_system_score_codex":0.000051563155,"about_ca_system_score_gemma":0.0009885204,"threshold_uncertainty_score":0.9998381},"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003021729,0.000009884984,0.0012736629,0.00008001685,0.0005738722,0.0012551884,0.9930752,0.00046594188,0.00024449744],"genre_scores_gemma":[0.0004981443,0.000044105873,0.008631863,0.0027404062,0.0012835253,0.00009394886,0.9865317,0.000042609645,0.00013370362],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9968919,0.000013617967,0.00061185856,0.0017834089,0.00022313063,0.0004760447],"domain_scores_gemma":[0.99514943,0.00012115295,0.00049921515,0.0040770434,0.000113785165,0.00003938555],"candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00032425683,0.0005498718,0.0006120383,0.00085739075,0.00040125076,0.0007653183,0.003931434,0.0003427065,0.00026816514],"category_scores_gemma":[0.0022168562,0.00054424285,0.000096821546,0.00036435146,0.000037845864,0.0015422347,0.006861573,0.00070419855,0.0018653049],"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":[0.00009574407,0.00003776544,0.000086222484,0.0004846133,0.0000989555,0.0000052943124,0.000018803796,0.00021766534,0.00007990352,0.00030534688,0.99567646,0.0028932448],"study_design_scores_gemma":[0.00040950172,0.000015979758,0.0001471069,0.00027429502,0.0002461769,0.0000012714848,0.00024256813,0.0045011803,0.000021583815,0.00013211579,0.99339706,0.0006111338],"about_ca_topic_score_codex":0.0051963506,"about_ca_topic_score_gemma":0.001991532,"teacher_disagreement_score":0.0073582,"about_ca_system_score_codex":0.00017111044,"about_ca_system_score_gemma":0.000072489565,"threshold_uncertainty_score":0.9997009},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.11386886,0.00019956088,0.38953507,0.013293408,0.0016454301,0.0016545062,0.00002233887,0.00022929622,0.47955152],"genre_scores_gemma":[0.92787665,0.0006456673,0.06525891,0.0015663545,0.00009877172,0.000092408205,0.00014545168,0.000009442162,0.004306335],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99908966,0.000036323436,0.0003015613,0.0003622722,0.00011664098,0.00009353486],"domain_scores_gemma":[0.99908525,0.000038524595,0.00028161352,0.00037503376,0.00008549217,0.00013407246],"candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.0007858199,0.00010843169,0.00015882611,0.00012039281,0.00008411596,0.012741205,0.0018245535,0.00008465875,0.000008875371],"category_scores_gemma":[0.000028431292,0.000092363145,0.000011591263,0.000105262385,0.000030940795,0.02449254,0.008462462,0.00011077806,0.0000057758425],"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":[0.000004213612,0.000033691445,0.013726475,0.0002116875,0.000032301377,0.0000016699377,0.009744909,0.0009060574,0.000008136215,0.47563833,0.07279881,0.4268937],"study_design_scores_gemma":[0.00020476508,0.000033731332,0.019871246,0.00007608221,0.0000022439212,0.00001556799,0.00038324954,0.8555197,0.0000824991,0.09548684,0.027999904,0.00032418265],"about_ca_topic_score_codex":0.00008187692,"about_ca_topic_score_gemma":0.0000043507775,"teacher_disagreement_score":0.8546136,"about_ca_system_score_codex":0.000039343555,"about_ca_system_score_gemma":0.00042163886,"threshold_uncertainty_score":0.9995569},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.99747056,0.00011467154,0.000029143794,0.0002470041,0.00016006194,0.00019809039,4.77089e-7,0.0000014849254,0.0017784922],"genre_scores_gemma":[0.9995122,0.000019568859,0.00025618408,0.00010983049,0.00005068557,0.000002520962,6.7477345e-8,0.0000026496537,0.00004630875],"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","domain_scores_codex":[0.9986838,0.00007100717,0.00037031414,0.0002546476,0.0004150428,0.00020516569],"domain_scores_gemma":[0.99894786,0.000020796253,0.00028741112,0.00020305598,0.000501622,0.000039231174],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002817892,0.00008000745,0.00015957936,0.00026334813,0.00036683673,0.0002158738,0.0017878592,0.000015036695,0.0000034534517],"category_scores_gemma":[0.00027213208,0.000052396757,0.000025972373,0.0010283937,0.00061795727,0.00010490524,0.0010680037,0.00015501089,4.274738e-7],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000059726197,0.00022905847,0.03648615,0.0000036390813,0.00000883047,0.0000033794513,0.001372015,0.000037209244,0.9610605,0.00015446231,0.00009606026,0.00048894907],"study_design_scores_gemma":[0.0006716737,0.0028648034,0.73823637,0.00010803828,0.000016786431,0.000040468618,0.01822637,0.0002729591,0.23853648,0.000044608947,0.0008661237,0.00011534731],"about_ca_topic_score_codex":0.00037217123,"about_ca_topic_score_gemma":0.00031816176,"teacher_disagreement_score":0.72252405,"about_ca_system_score_codex":0.00006422381,"about_ca_system_score_gemma":0.0005198058,"threshold_uncertainty_score":0.3322317},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.833047,0.00065815786,0.000013353654,0.09472745,0.00045811568,0.012172874,0.000006320469,0.000028626582,0.058888134],"genre_scores_gemma":[0.99284184,0.0014051625,0.0002775233,0.0011313098,0.00028297,0.0018856646,0.000005533956,0.00004119447,0.0021288001],"study_design_codex":"qualitative","study_design_gemma":"qualitative","domain_scores_codex":[0.80222285,0.16833238,0.0051456937,0.0011329913,0.018832754,0.0043333094],"domain_scores_gemma":[0.64449644,0.3162845,0.0007787134,0.0025663394,0.03455572,0.0013182532],"candidate_categories":["metaresearch","sts","open_science","research_integrity"],"consensus_categories":["metaresearch","sts","open_science"],"category_scores_codex":[0.3506349,0.0002799371,0.001298269,0.01059274,0.012409662,0.0007783928,0.00901304,0.0004966386,0.00018753177],"category_scores_gemma":[0.5439509,0.00021979936,0.00012707325,0.0119728865,0.0071631023,0.0011682868,0.009146985,0.042358574,0.00020518135],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":true,"study_design_scores_codex":[0.20912422,0.0117445355,0.09682607,0.0038425827,0.00035278784,0.00043351908,0.42579204,0.0003860043,0.0023782803,0.18764381,0.056926876,0.0045492724],"study_design_scores_gemma":[0.21342054,0.022907468,0.07092129,0.023348412,0.000029433993,0.0000025835623,0.4569714,0.0028570194,0.00025703543,0.18020214,0.028397672,0.0006849799],"about_ca_topic_score_codex":0.007236953,"about_ca_topic_score_gemma":0.0122073265,"teacher_disagreement_score":0.19331601,"about_ca_system_score_codex":0.007318554,"about_ca_system_score_gemma":0.070285805,"threshold_uncertainty_score":0.9993739},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.82490367,0.0017004292,0.000131877,0.010384888,0.0012536494,0.01186535,0.00035320787,0.00008828737,0.14931864],"genre_scores_gemma":[0.92301613,0.0008841775,0.003288467,0.00016445143,0.00024195881,0.0012657759,0.000010456378,0.000055942357,0.071072645],"study_design_codex":"design_other","study_design_gemma":"observational","domain_scores_codex":[0.86400604,0.01368693,0.002856184,0.009224446,0.10529793,0.0049284883],"domain_scores_gemma":[0.7970627,0.08301832,0.0014385254,0.01188999,0.10362444,0.0029660505],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaresearch","bibliometrics","sts","open_science"],"category_scores_codex":[0.53597844,0.000531726,0.0011701378,0.2145123,0.0039040085,0.07919924,0.059312083,0.00052049925,0.0007975684],"category_scores_gemma":[0.68403435,0.00040505172,0.00014425954,0.34488538,0.0038071803,0.0032377625,0.41908514,0.0063845855,0.00048301698],"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":[0.0031625025,0.0042019864,0.15253454,0.000468707,0.0002228725,0.0006633521,0.008665497,0.00011991629,0.008371126,0.004502578,0.10086114,0.7162258],"study_design_scores_gemma":[0.0038970741,0.0059634713,0.64756376,0.00082694885,0.00005345464,0.00004273815,0.03784709,0.03349748,0.0033727626,0.15386422,0.11116617,0.0019048474],"about_ca_topic_score_codex":0.012526058,"about_ca_topic_score_gemma":0.001095179,"teacher_disagreement_score":0.71432096,"about_ca_system_score_codex":0.0014796141,"about_ca_system_score_gemma":0.033559326,"threshold_uncertainty_score":0.99984014},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.92119753,0.0006513706,0.00019348062,0.070280075,0.00008561328,0.0034745487,0.0019780947,0.000007787919,0.002131486],"genre_scores_gemma":[0.7557173,0.0019101355,0.0065694964,0.0003739949,0.00011321363,0.0005389572,0.005245259,0.000038666018,0.229493],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9725335,0.0041600266,0.0037951425,0.0025788345,0.016471125,0.0004613344],"domain_scores_gemma":[0.91158324,0.018572196,0.008615717,0.012218806,0.04861324,0.0003967872],"candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["metaresearch","open_science"],"category_scores_codex":[0.4841322,0.00017846684,0.0012623752,0.00031054177,0.0010189619,0.0017557387,0.008491457,0.00007513189,0.00020469549],"category_scores_gemma":[0.57430595,0.00007659453,0.00012021669,0.003362536,0.001623015,0.0018795934,0.009729375,0.00032664297,0.000044821994],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00018977406,0.000055190125,0.50750715,0.00026080277,0.000069736874,0.000008178632,0.00008301158,0.000018646311,0.00018350175,0.00019321052,0.4901874,0.0012433754],"study_design_scores_gemma":[0.00022885727,0.00016431569,0.737583,0.00007877628,0.000027824924,0.000019279194,0.00010216459,0.006380669,0.000016316237,0.00019732381,0.2551275,0.00007401618],"about_ca_topic_score_codex":0.004087217,"about_ca_topic_score_gemma":0.0065157157,"teacher_disagreement_score":0.23505992,"about_ca_system_score_codex":0.000045533565,"about_ca_system_score_gemma":0.0006022954,"threshold_uncertainty_score":0.9992805},"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":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.020015858,0.0005628796,0.4114211,0.3936777,0.0027803245,0.005140528,0.001168953,0.00009042655,0.16514224],"genre_scores_gemma":[0.95667154,0.00013738323,0.019167766,0.014977749,0.0005480205,0.00026660995,0.00020449606,0.000029684225,0.007996775],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9924188,0.0003758667,0.002858773,0.0014368166,0.0022939215,0.00061577867],"domain_scores_gemma":[0.98816293,0.00737268,0.0015921299,0.0009928269,0.0013117541,0.00056770083],"candidate_categories":["metaresearch","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.010107327,0.00036477429,0.00085721473,0.00016109266,0.0009658678,0.026977994,0.03694253,0.0003445949,0.02093419],"category_scores_gemma":[0.015298838,0.00021541369,0.00022083463,0.0020951247,0.00031779954,0.020470668,0.008906057,0.0005398765,0.00019171782],"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":[0.0003395609,0.0000805677,0.004373802,0.00001734247,0.00015189726,4.3064875e-7,0.0076114186,0.00002105473,0.00009501291,0.0027584466,0.67455846,0.309992],"study_design_scores_gemma":[0.0013162115,0.00009890844,0.026415812,0.00005752828,0.00007835382,6.5117587e-7,0.0152308,0.0058999076,0.010237493,0.012890184,0.9273116,0.0004625702],"about_ca_topic_score_codex":0.0007631348,"about_ca_topic_score_gemma":0.0002532315,"teacher_disagreement_score":0.93665564,"about_ca_system_score_codex":0.00007150822,"about_ca_system_score_gemma":0.0034112048,"threshold_uncertainty_score":0.99910975},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.94839615,0.0004006656,0.04782677,0.0016738363,0.0003842346,0.00032929625,0.00001829024,0.0000067260817,0.0009640504],"genre_scores_gemma":[0.9746092,0.00007405841,0.024936503,0.00018842745,0.00006112873,0.000040622734,8.5417633e-7,0.0000053977333,0.000083794665],"study_design_codex":"qualitative","study_design_gemma":"observational","domain_scores_codex":[0.9674442,0.030055718,0.0011231394,0.0002046817,0.00091052015,0.00026176055],"domain_scores_gemma":[0.9874983,0.010430503,0.0006297087,0.00015257896,0.0011874571,0.00010148941],"candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.049864344,0.00009907633,0.00035169462,0.0003276352,0.00015369712,0.000069894624,0.0005012278,0.00006024919,0.00007468863],"category_scores_gemma":[0.037473004,0.000075493524,0.000072348776,0.00087117177,0.0006745094,0.00033797708,0.00011344415,0.0003181329,5.0378424e-7],"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":[0.00053282187,0.00068852317,0.3469118,0.000008046944,0.00066700665,0.000010931044,0.61198986,0.00008517226,0.0042351275,0.02249812,0.000058278925,0.012314327],"study_design_scores_gemma":[0.0015881782,0.00027533303,0.5236318,0.00019196898,0.00012757546,5.5276024e-7,0.28934455,0.0006148168,0.007839242,0.17300957,0.0032385353,0.00013784312],"about_ca_topic_score_codex":0.0028427597,"about_ca_topic_score_gemma":0.007168301,"teacher_disagreement_score":0.3226453,"about_ca_system_score_codex":0.0002414276,"about_ca_system_score_gemma":0.0005483015,"threshold_uncertainty_score":0.9783646},"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":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000039361556,0.0006767945,0.7015411,0.22215125,0.0010522817,0.002688651,0.0002662936,0.00009615306,0.07148807],"genre_scores_gemma":[0.66172487,0.014081157,0.14794052,0.016431699,0.012916507,0.00078078493,0.00029730183,0.00013241368,0.14569478],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99528533,0.00013567955,0.0003880576,0.0020282492,0.00096811424,0.0011945897],"domain_scores_gemma":[0.97403026,0.0003206483,0.0007317628,0.024525112,0.00023082753,0.00016139768],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.006098706,0.0003204364,0.00024217149,0.00016669102,0.0049060294,0.0682991,0.07585825,0.000106644846,0.000043295593],"category_scores_gemma":[0.010478454,0.0002066611,0.00008983333,0.000334563,0.00031675096,0.13276555,0.08312254,0.00049961975,0.000117913],"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":[0.000023081828,0.00006199193,0.0004990508,0.00018009677,0.00015020974,0.000018963976,0.00091709004,0.000003256822,0.00002345398,0.72426796,0.025089012,0.24876583],"study_design_scores_gemma":[0.00039217158,0.00007372379,0.0035232292,0.00012807494,0.000047911337,0.000009406481,0.00024861595,0.37477538,0.000029352821,0.0043732254,0.6161058,0.0002931164],"about_ca_topic_score_codex":0.0023079284,"about_ca_topic_score_gemma":0.0010190282,"teacher_disagreement_score":0.71989477,"about_ca_system_score_codex":0.00007355689,"about_ca_system_score_gemma":0.00019777841,"threshold_uncertainty_score":0.9978567},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8499712,0.011476533,0.02942251,0.027471289,0.0035569577,0.0015767743,0.000040650248,0.00006116386,0.07642296],"genre_scores_gemma":[0.96876734,0.0002532851,0.025943622,0.003932218,0.00009078666,0.00004697108,0.00003373758,0.000017036822,0.0009150077],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99582374,0.00016163127,0.00074889464,0.0020943722,0.0001835125,0.0009878366],"domain_scores_gemma":[0.99154425,0.0000854173,0.00049918285,0.006269754,0.00024382435,0.0013575805],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.00464302,0.00026910176,0.0003379799,0.0013013252,0.00077398185,0.003244557,0.027186487,0.00012450706,0.000049215832],"category_scores_gemma":[0.0011235549,0.00028197403,0.000045101297,0.0014688508,0.0011231008,0.017179647,0.0034319484,0.00046133346,0.000009971629],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017099027,0.0007222431,0.030700428,0.000063847445,0.00040703558,0.0012102835,0.011831138,0.000013821456,0.13717876,0.6445756,0.1497409,0.02338496],"study_design_scores_gemma":[0.008161003,0.0025050018,0.13761765,0.00061401894,0.00010956735,0.0025946214,0.015157553,0.0022492348,0.047628988,0.071765006,0.7097178,0.0018795491],"about_ca_topic_score_codex":0.01004226,"about_ca_topic_score_gemma":0.030789541,"teacher_disagreement_score":0.5728106,"about_ca_system_score_codex":0.0011079808,"about_ca_system_score_gemma":0.002004551,"threshold_uncertainty_score":0.9999632},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9878289,0.0006240045,0.00019336704,0.001026735,0.00004509424,0.0010601495,0.00025798372,0.000046668614,0.008917086],"genre_scores_gemma":[0.99805564,0.00060997816,0.00010240397,0.00001722565,0.0000310352,1.4532588e-7,0.00004916282,0.000017027118,0.0011174113],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9985287,0.00031622118,0.00012591545,0.0002032133,0.00038495086,0.00044099297],"domain_scores_gemma":[0.99813735,0.00030956388,0.00012573082,0.0010895119,0.00007940373,0.00025843596],"candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.0022447864,0.00019068176,0.0003429299,0.00009374211,0.000989328,0.00021288692,0.0053779352,0.00011370199,0.00015029601],"category_scores_gemma":[0.00007771481,0.00016224275,0.000059198777,0.00045009566,0.00065611186,0.002666471,0.011633699,0.0003681222,0.000017231076],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.002103133,0.00095824234,0.63187015,0.00044948328,0.0023474006,0.000019294199,0.17474727,0.0025150958,0.0020419115,0.006495962,0.11461392,0.061838124],"study_design_scores_gemma":[0.004699703,0.0005170871,0.5086923,0.00024141866,0.0005386598,0.000035028635,0.33573604,0.027399918,0.00009556109,0.00035307367,0.120871924,0.00081929093],"about_ca_topic_score_codex":0.010673007,"about_ca_topic_score_gemma":0.0050658365,"teacher_disagreement_score":0.16098876,"about_ca_system_score_codex":0.00030975684,"about_ca_system_score_gemma":0.00010080774,"threshold_uncertainty_score":0.99936306},"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":"review","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.069666825,0.44400662,0.01640344,0.22815974,0.0024122514,0.00093760074,0.00066220475,0.00068866153,0.23706265],"genre_scores_gemma":[0.33063403,0.07638933,0.0059937793,0.001473402,0.0018779079,0.000058439935,0.00009657026,0.00010662359,0.5833699],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9874735,0.004426701,0.00093656476,0.0014291817,0.0031698716,0.0025641765],"domain_scores_gemma":[0.98517275,0.010632445,0.00016138842,0.00085917127,0.002412153,0.00076209626],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["bibliometrics","insufficient_payload"],"category_scores_codex":[0.027783608,0.00048246697,0.00050980743,0.012781419,0.00071229023,0.009977939,0.0042379727,0.00096678507,0.0044155866],"category_scores_gemma":[0.03935568,0.0003695089,0.00039614917,0.10008521,0.000358988,0.0022646266,0.0012493561,0.0037012163,0.0033851776],"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":[0.000020924823,0.00022394501,0.00015566683,0.000043527263,0.00010523538,0.00004718778,0.0015207077,0.00002621916,0.0016700445,0.23988584,0.3796219,0.3766788],"study_design_scores_gemma":[0.0003328529,0.00038781372,0.0013688875,0.00012677012,0.00004128244,0.00015382281,0.0019725757,0.0055215945,0.0035822713,0.09599838,0.8900621,0.0004516881],"about_ca_topic_score_codex":0.0027967426,"about_ca_topic_score_gemma":0.001495494,"teacher_disagreement_score":0.5104402,"about_ca_system_score_codex":0.001956593,"about_ca_system_score_gemma":0.015012199,"threshold_uncertainty_score":0.99987566},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.23101072,0.0012486938,0.00035958705,0.0004816597,0.011742654,0.0009540289,0.0011332748,0.00045556572,0.75261384],"genre_scores_gemma":[0.78909785,0.00036329075,0.00034697074,0.00014739252,0.0014183128,0.00012780006,0.00010630506,0.00041423604,0.20797786],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.97638667,0.0010588353,0.0023145673,0.0042152507,0.012608634,0.0034160581],"domain_scores_gemma":[0.98500407,0.0017009784,0.0024490126,0.0026955183,0.0054823495,0.0026680857],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication"],"consensus_categories":["metaepi_narrow","sts"],"category_scores_codex":[0.007577396,0.0020287165,0.0017299907,0.0012698346,0.003251614,0.0075816284,0.0045906524,0.0009445454,0.00007125021],"category_scores_gemma":[0.009727593,0.0022668212,0.0007557751,0.0035216818,0.006425399,0.004790249,0.0032671865,0.0021869366,0.0003547363],"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":[0.00065226905,0.0038259886,0.1389215,0.0012844866,0.0012986162,0.07465299,0.002051796,0.0028598616,0.1235824,0.0017148308,0.57299054,0.07616473],"study_design_scores_gemma":[0.0013981861,0.00015100796,0.06718807,0.001991088,0.0005536172,0.042863864,0.0028231547,0.005446363,0.014603291,0.00006534225,0.86019534,0.0027206622],"about_ca_topic_score_codex":0.42513272,"about_ca_topic_score_gemma":0.31658524,"teacher_disagreement_score":0.5580871,"about_ca_system_score_codex":0.09261425,"about_ca_system_score_gemma":0.35334632,"threshold_uncertainty_score":0.9992455},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017028567,0.0008793411,0.47667482,0.01277496,0.0004514357,0.0010194443,0.00010436675,0.0024295482,0.50396323],"genre_scores_gemma":[0.9067843,0.0022840376,0.04764822,0.0016709534,0.0006371275,5.117295e-7,0.0014404055,0.0024465986,0.037087876],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9967559,0.00037833746,0.0002145271,0.0010466106,0.0010459294,0.00055868336],"domain_scores_gemma":[0.99729604,0.00005459942,0.000059252005,0.0015006376,0.0008250757,0.00026439904],"candidate_categories":["sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.005590617,0.00013407391,0.00010441752,0.0010147614,0.0035044763,0.23793873,0.01994872,0.00003882121,0.0009092214],"category_scores_gemma":[0.0026487939,0.0001283067,0.000033725544,0.0055535594,0.00025424158,0.11629295,0.026907783,0.0005963963,0.0062427223],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000007833456,0.00005565123,8.955279e-7,0.0001105017,0.000029946017,0.0001485796,0.0015698051,0.0000083439745,0.0071550794,0.67395115,0.08772826,0.22923397],"study_design_scores_gemma":[0.0001333176,0.00009097804,0.00024086762,0.00013918233,0.000002260267,0.00012278285,0.00014133395,0.0038394395,0.00029695037,0.0014079635,0.9934298,0.00015513755],"about_ca_topic_score_codex":0.0000136385315,"about_ca_topic_score_gemma":9.1765536e-8,"teacher_disagreement_score":0.9057015,"about_ca_system_score_codex":0.00022551141,"about_ca_system_score_gemma":0.00002677254,"threshold_uncertainty_score":0.99779284},"labels":[{"model":"gemma","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"agree"},{"id":"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010172085,0.000084024345,0.011962836,0.011264914,0.00041857097,0.00057480385,0.0001245933,0.00047330617,0.9649249],"genre_scores_gemma":[0.9262946,0.00005813543,0.00034529908,0.003384498,0.00021882782,1.2793589e-7,0.00048155757,0.00073086587,0.06848608],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99633527,0.0007471006,0.00058448216,0.0008135893,0.0010923163,0.000427214],"domain_scores_gemma":[0.9969308,0.00024750785,0.00022762507,0.0010993944,0.0012848988,0.00020977373],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.0065672467,0.00015124213,0.0002575021,0.00065255084,0.0032796583,0.04341032,0.02598169,0.00011819539,0.032014463],"category_scores_gemma":[0.015817609,0.00012974074,0.00006670683,0.0032634458,0.00019312432,0.005816657,0.023183985,0.00051293394,0.012645677],"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":[0.000030957686,0.00004095092,0.00008161411,0.000005044897,0.000015441965,0.0000037706593,0.00017569569,0.00002967203,0.00008990903,0.020487292,0.7425623,0.23647736],"study_design_scores_gemma":[0.00040741178,0.0000314487,0.0038570138,0.000027257152,0.0000070362025,0.000016395792,0.00037646623,0.00035422534,0.00022493766,0.015651342,0.97889847,0.0001479792],"about_ca_topic_score_codex":0.00008955951,"about_ca_topic_score_gemma":0.0000012700264,"teacher_disagreement_score":0.9161225,"about_ca_system_score_codex":0.00014749584,"about_ca_system_score_gemma":0.000022884484,"threshold_uncertainty_score":0.9980179},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.087522075,0.00043589203,0.004818726,0.022586837,0.00020896939,0.004922088,0.0002423927,0.000118404736,0.8791446],"genre_scores_gemma":[0.99339175,0.000048877617,0.0016807489,0.0009823303,0.000072371295,6.962742e-7,0.0000134287175,0.0004081767,0.003401612],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9820644,0.008503643,0.003318047,0.0018576332,0.003762311,0.0004939263],"domain_scores_gemma":[0.9892775,0.0021901093,0.0017586132,0.0038485967,0.0023100725,0.0006151425],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","open_science","insufficient_payload"],"category_scores_codex":[0.1039862,0.0002518862,0.0015065164,0.000556587,0.0019706807,0.023180665,0.0353072,0.000054415137,0.3238648],"category_scores_gemma":[0.26524538,0.0001675258,0.00026058394,0.0055436557,0.00035264008,0.0027286445,0.017259805,0.00027414638,0.102952056],"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":[0.000102651524,0.00016083897,0.0002549394,0.00011955144,0.0000747388,0.00004328687,0.0031447548,0.00009433355,0.0040672976,0.0072436053,0.6971482,0.2875458],"study_design_scores_gemma":[0.0001782945,0.00007952911,0.004268664,0.00014713196,0.00004083541,0.000029560626,0.0014976297,0.0030112227,0.0007927868,0.00054284587,0.9891703,0.0002411831],"about_ca_topic_score_codex":0.000030369647,"about_ca_topic_score_gemma":0.000004266245,"teacher_disagreement_score":0.90586966,"about_ca_system_score_codex":0.00015822481,"about_ca_system_score_gemma":0.000040179417,"threshold_uncertainty_score":0.9993286},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.15958048,0.00036647104,0.024042707,0.016606377,0.00038840336,0.0011334269,0.000040144303,0.00047145374,0.79737055],"genre_scores_gemma":[0.99512786,0.00025632672,0.0015095792,0.00064316904,0.0000271695,7.796577e-8,0.00009170476,0.0001587083,0.0021854234],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9988197,0.00019553467,0.00016940093,0.0003744141,0.0002932791,0.00014763426],"domain_scores_gemma":[0.997622,0.000047313977,0.000106847096,0.00050558336,0.0016106942,0.000107523694],"candidate_categories":["sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.00097292475,0.000065949804,0.00010020695,0.000058080997,0.001440597,0.0017626854,0.0015790132,0.000014015641,0.00015415504],"category_scores_gemma":[0.0025819058,0.000054081287,0.000013588992,0.0016240097,0.0007321315,0.00055781833,0.0037531077,0.00009442363,0.00018819685],"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":[0.0000516045,0.0005460183,0.0000823343,0.00007041497,0.000040168976,0.000040399136,0.019419353,0.0000534952,0.0009146274,0.31364846,0.13368517,0.53144795],"study_design_scores_gemma":[0.00044909905,0.00010144534,0.005319715,0.0000141404,0.000004004523,0.00017159474,0.00038222605,0.0021615673,0.00028223812,0.00024274217,0.9907843,0.000086933884],"about_ca_topic_score_codex":0.000005541162,"about_ca_topic_score_gemma":3.895947e-7,"teacher_disagreement_score":0.8570991,"about_ca_system_score_codex":0.000043957625,"about_ca_system_score_gemma":0.00005349386,"threshold_uncertainty_score":0.9998594},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[4.92678e-7,0.0060760714,3.287727e-7,0.0005914876,0.00014557366,0.0031583859,0.21452563,0.0004866904,0.77501535],"genre_scores_gemma":[0.00039346557,0.000034902085,0.00006598845,0.004688875,0.0052880603,0.0025270586,0.20761801,0.003338546,0.7760451],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9969145,0.00011721366,0.0003635554,0.0011315284,0.00059091795,0.0008822691],"domain_scores_gemma":[0.9975685,0.000054910746,0.0008301217,0.001302069,0.000047789817,0.00019660662],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["open_science","research_integrity","insufficient_payload"],"category_scores_codex":[0.00021495509,0.00078686996,0.0007648836,0.0004090939,0.00013099055,0.0017087075,0.011523555,0.003103371,0.64697224],"category_scores_gemma":[0.00071749993,0.0005722517,0.00015485578,0.00090058317,0.0000132620335,0.0009160252,0.010591567,0.004534803,0.09300255],"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":[0.000011593389,0.000020514983,0.00014417866,0.00048120023,0.00004952849,0.000037334394,0.000590882,0.0000014910389,0.0000020735997,0.00003263411,0.9959018,0.0027267183],"study_design_scores_gemma":[0.0005504345,0.000014772926,0.0022895786,0.032139227,0.000030826053,0.0000129408845,0.00001674817,0.000101802245,0.00000912887,0.000059242626,0.9640224,0.00075287576],"about_ca_topic_score_codex":0.0016847698,"about_ca_topic_score_gemma":0.011498618,"teacher_disagreement_score":0.5539697,"about_ca_system_score_codex":0.00031212458,"about_ca_system_score_gemma":0.00019118046,"threshold_uncertainty_score":0.9996729},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8365679,0.0045421207,0.00010348497,0.007721123,0.0036066505,0.0033435405,0.014757922,0.0002504547,0.12910683],"genre_scores_gemma":[0.9971076,0.000003839193,0.0011536055,0.00037591084,0.000047309106,0.00015416363,0.000042309028,0.000012717603,0.0011025105],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.99001366,0.00011004084,0.0010015421,0.0022012223,0.0057388525,0.0009346636],"domain_scores_gemma":[0.9923995,0.0020754507,0.0003376989,0.0009686517,0.003761654,0.0004570596],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaresearch","scholarly_communication","open_science"],"category_scores_codex":[0.041217495,0.00021287276,0.00029156092,0.0039121397,0.0022309842,0.09289081,0.03134297,0.00008456129,0.0016210156],"category_scores_gemma":[0.11723636,0.00014865003,0.000036419904,0.047628194,0.0014358092,0.0613134,0.014807114,0.00071416574,0.0002153134],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":true,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000140821,0.00034514032,0.047991488,0.00029570083,0.000012245004,0.00007838095,0.0120802615,0.0017159783,0.08304668,0.2663185,0.43183962,0.1561352],"study_design_scores_gemma":[0.00091182283,0.00010714633,0.41656294,0.007597992,0.000006365075,0.00006805407,0.0041061617,0.15677316,0.03548873,0.17535985,0.20162694,0.0013908383],"about_ca_topic_score_codex":0.00011024761,"about_ca_topic_score_gemma":0.0006875498,"teacher_disagreement_score":0.36857146,"about_ca_system_score_codex":0.000622778,"about_ca_system_score_gemma":0.010873875,"threshold_uncertainty_score":0.99929166},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8637406,0.00013850833,0.00002282129,0.012604281,0.000077398625,0.00070571183,0.00004995273,0.0000012512152,0.12265943],"genre_scores_gemma":[0.8318544,0.00039634996,0.061797213,0.00007247665,0.0001951038,0.000008985757,0.000045708894,0.000020934369,0.1056088],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9979765,0.000044168202,0.00019132023,0.00079771155,0.0006786745,0.0003116194],"domain_scores_gemma":[0.9926194,0.0020718025,0.000049034643,0.004647167,0.0003425913,0.0002700153],"candidate_categories":["metaresearch","open_science","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.017036118,0.00005885894,0.00017615965,0.000053389664,0.0006003391,0.0009950643,0.0033635735,0.00017648867,0.0012685214],"category_scores_gemma":[0.037220966,0.000048498627,0.0000108660115,0.000048240865,0.00068840955,0.00037136086,0.016933024,0.0017485141,0.0011306213],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006420695,0.0006261181,0.6433855,0.0001980092,0.00011380169,0.0007241268,0.0011877412,1.098097e-7,0.018072875,0.00081563427,0.0033184919,0.33091548],"study_design_scores_gemma":[0.008097534,0.00091829116,0.7189452,0.0020720665,0.000056446843,0.00006851925,0.0007315836,0.008867507,0.0072568427,0.009546338,0.24306853,0.00037113202],"about_ca_topic_score_codex":0.00071550364,"about_ca_topic_score_gemma":0.0026730422,"teacher_disagreement_score":0.33054435,"about_ca_system_score_codex":0.000024070883,"about_ca_system_score_gemma":0.00027211368,"threshold_uncertainty_score":0.99964714},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00898847,0.00011998359,0.40412718,0.46388474,0.00018581215,0.0018901062,0.000029259314,0.00022193474,0.12055252],"genre_scores_gemma":[0.99574006,0.000039725193,0.00053695915,0.0032088086,0.000051674528,1.120564e-7,0.0000037136545,0.00007258886,0.0003463784],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981993,0.00039861514,0.0001422674,0.00037874898,0.00063799997,0.00024311047],"domain_scores_gemma":[0.9982199,0.00018214088,0.00013154637,0.0010473321,0.0003114783,0.00010757439],"candidate_categories":["sts","scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.0024304818,0.0000666227,0.00006551425,0.000058264697,0.0027467832,0.0068656225,0.013875592,0.000015557265,0.00031427873],"category_scores_gemma":[0.007235929,0.00003838751,0.000018473198,0.0015938263,0.00022240393,0.0050677033,0.022969961,0.00024183959,0.00022281667],"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":[0.000055718923,0.000064283086,0.00006647714,0.000042543426,0.00003311961,0.0000045018182,0.004244864,0.00021242612,0.003932607,0.51003766,0.28980273,0.19150308],"study_design_scores_gemma":[0.00014483639,0.00025760126,0.0051849084,0.000020572259,0.0000067773035,0.0000107430515,0.0009176091,0.006106481,0.001102254,0.00026210598,0.98590994,0.00007618259],"about_ca_topic_score_codex":0.00015562157,"about_ca_topic_score_gemma":0.0000032216549,"teacher_disagreement_score":0.98675156,"about_ca_system_score_codex":0.00005189575,"about_ca_system_score_gemma":0.000021243055,"threshold_uncertainty_score":0.9985515},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006238146,0.000030119398,0.12589565,0.004495705,0.0006323162,0.000498116,0.000016225678,0.0017950154,0.8603987],"genre_scores_gemma":[0.9570203,0.000011442748,0.038085368,0.00095563126,0.0002283785,8.696306e-8,0.000109902336,0.0006763174,0.0029125784],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9976993,0.00024977815,0.00030988673,0.0005864718,0.0006130218,0.0005415016],"domain_scores_gemma":[0.99657214,0.000051442606,0.00016819817,0.0017628146,0.0012535125,0.00019188461],"candidate_categories":["sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.0029014314,0.00014383922,0.00014206069,0.0003628416,0.0068057515,0.0076647573,0.014094254,0.00003707455,0.0015952995],"category_scores_gemma":[0.002047373,0.00015378426,0.000025728208,0.002022287,0.0012760698,0.0014046325,0.023024637,0.00029110914,0.004761706],"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":[0.00004744255,0.00022985425,0.00002766356,0.000047046604,0.0000493497,0.0003612318,0.04429191,0.000018165354,0.0053735687,0.27853853,0.44955948,0.22145575],"study_design_scores_gemma":[0.00021520168,0.00012175018,0.0003159134,0.00006314565,0.0000025659147,0.001312787,0.0007898042,0.0025133183,0.001946085,0.0013161644,0.9911703,0.00023298324],"about_ca_topic_score_codex":0.00003955293,"about_ca_topic_score_gemma":5.8328857e-7,"teacher_disagreement_score":0.9507821,"about_ca_system_score_codex":0.0002033377,"about_ca_system_score_gemma":0.000043868487,"threshold_uncertainty_score":0.99931735},"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":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000123102245,0.0001997555,2.5723034e-7,0.000014364473,0.000075164906,0.0017605583,0.7000508,0.00020734903,0.29767948],"genre_scores_gemma":[0.018787993,0.0000023146747,0.0037568666,0.0011082424,0.0008516196,0.002111533,0.8214914,0.0045968695,0.14729314],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984033,0.00011171527,0.00018926138,0.00044100994,0.00035572107,0.0004989651],"domain_scores_gemma":[0.99861306,0.000071682414,0.00029117696,0.0007987945,0.000048865648,0.00017641418],"candidate_categories":["metaepi_narrow","open_science","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00029038504,0.00027978065,0.0003944987,0.0004888844,0.00017409584,0.0005001316,0.0059540966,0.00022807704,0.24343844],"category_scores_gemma":[0.003540884,0.00031523212,0.000057265082,0.00131139,0.000055496585,0.00024921336,0.0039295363,0.0008194931,0.013108581],"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":[0.000009728232,0.000012446075,0.000021005553,0.00035429143,0.000009323157,0.0000075269027,0.00023664541,0.0000043148725,0.000090775065,0.000015315187,0.9984284,0.00081022846],"study_design_scores_gemma":[0.0005270122,0.000029891733,0.0061509884,0.006918513,0.0000055511546,0.0000022915738,0.00007086471,0.00006732389,0.000065493085,0.00009800958,0.98568904,0.00037504273],"about_ca_topic_score_codex":0.00043048637,"about_ca_topic_score_gemma":0.0015147805,"teacher_disagreement_score":0.23032986,"about_ca_system_score_codex":0.00035762435,"about_ca_system_score_gemma":0.0006443969,"threshold_uncertainty_score":0.99992996},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0055837017,0.00029881764,0.0004075653,0.0008927871,0.0009881747,0.0040391125,0.00001557756,0.00008774951,0.9876865],"genre_scores_gemma":[0.22937004,0.000013626376,0.002724353,0.0002917411,0.0003108096,0.0002235501,0.00002191744,0.00007841683,0.76696557],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99791,0.000101690515,0.00016456343,0.0011090515,0.00047385241,0.0002408372],"domain_scores_gemma":[0.9976504,0.00003199995,0.00015857327,0.0020058842,0.000048722002,0.00010446108],"candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0015994569,0.00016060256,0.00033126702,0.00012408766,0.00034345596,0.00010434813,0.0015426874,0.0001778767,0.020077603],"category_scores_gemma":[0.0013793687,0.000127203,0.000036361365,0.00074304745,0.0006480611,0.00011253982,0.00084455806,0.00061602524,0.000186836],"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":[0.0019400355,0.0043748165,0.25352895,0.00010012973,0.0002396497,0.00013011863,0.003942391,0.0000055111987,0.00047052692,0.003873126,0.707897,0.023497732],"study_design_scores_gemma":[0.00028247182,0.0006754405,0.18193154,0.00007172952,0.000051963063,0.0000128155225,0.0015689844,0.0000040376885,0.00005698282,0.00077620737,0.8143623,0.00020555551],"about_ca_topic_score_codex":0.00022485753,"about_ca_topic_score_gemma":0.000049689625,"teacher_disagreement_score":0.22378634,"about_ca_system_score_codex":0.00012241917,"about_ca_system_score_gemma":0.00043178326,"threshold_uncertainty_score":0.98081815},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.31360453,0.0000045087436,0.00000814148,0.009144114,0.00016608425,0.0009899151,0.000014142558,0.000009388989,0.6760592],"genre_scores_gemma":[0.9281459,0.00025794597,0.00023124882,0.0005260317,0.00003705072,0.00011391823,0.000015323223,0.0000067319565,0.07066585],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99346066,0.0028667853,0.00032406655,0.0013960656,0.0015487627,0.00040368928],"domain_scores_gemma":[0.9936738,0.0014671788,0.00032991517,0.0021280996,0.0021200033,0.00028099763],"candidate_categories":["metaresearch","insufficient_payload"],"consensus_categories":["metaresearch","insufficient_payload"],"category_scores_codex":[0.09860799,0.000085013984,0.00018803612,0.00062112225,0.00082682114,0.00053957105,0.003418722,0.00014983724,0.044059508],"category_scores_gemma":[0.041103277,0.00007456056,0.000031411822,0.002525876,0.0007785217,0.00070717465,0.0011745602,0.00043048014,0.039265838],"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":[0.00006324161,0.00019175322,0.8682222,0.00001635759,0.000091753,0.0000014487463,0.017850058,0.00001864494,0.0008967074,0.044588234,0.0661334,0.0019262005],"study_design_scores_gemma":[0.00021880349,0.0000015449026,0.69903857,0.000020611571,0.000017377115,0.0000027622489,0.002058186,0.00003090409,0.00020938473,0.009773446,0.28853118,0.00009725736],"about_ca_topic_score_codex":0.673807,"about_ca_topic_score_gemma":0.37551847,"teacher_disagreement_score":0.61454135,"about_ca_system_score_codex":0.00033711037,"about_ca_system_score_gemma":0.0034007907,"threshold_uncertainty_score":0.9669739},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06084329,0.0012871693,0.00017923485,0.0070513096,0.0005023972,0.0022550228,0.00057087373,0.00012515235,0.92718554],"genre_scores_gemma":[0.6367292,0.0009530318,0.0014149287,0.00010594687,0.00017764204,0.0000044885446,0.0007544423,0.00012566712,0.35973468],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99203897,0.0002456328,0.00056056335,0.001144602,0.004956547,0.0010537084],"domain_scores_gemma":[0.9925059,0.00026128333,0.00026162356,0.0020070784,0.0036657986,0.001298283],"candidate_categories":["metaepi_narrow","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.011329134,0.00040681483,0.0011963886,0.003080725,0.0012262042,0.00036050138,0.003891896,0.00034851712,0.00026748818],"category_scores_gemma":[0.0013066413,0.00045995752,0.00056940014,0.005244086,0.0026486327,0.0009362684,0.0038803325,0.0031817078,0.00021613957],"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":[0.0009041226,0.0010252849,0.00076439715,0.003361393,0.0003568446,0.01683333,0.0011161929,0.0000016465403,0.002644665,0.00008767361,0.796888,0.17601646],"study_design_scores_gemma":[0.0008536245,0.00080283376,0.009420056,0.0063794125,0.0003184058,0.0051309383,0.006276291,0.000036690457,0.00034831118,0.00065728807,0.9691707,0.0006054373],"about_ca_topic_score_codex":0.059838008,"about_ca_topic_score_gemma":0.0016833504,"teacher_disagreement_score":0.5758859,"about_ca_system_score_codex":0.0021253168,"about_ca_system_score_gemma":0.01219289,"threshold_uncertainty_score":0.99978524},"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":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0015842478,0.000019338617,0.00013714531,0.0020426393,0.0008583981,0.00090017874,0.94965637,0.0001543122,0.044647355],"genre_scores_gemma":[0.01082214,0.00032948854,0.00048634174,0.0016593529,0.0012596695,4.389135e-7,0.98159766,0.0017950883,0.0020498051],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9984059,0.00012710337,0.00028332483,0.0005154921,0.0003990436,0.00026916142],"domain_scores_gemma":[0.99906594,0.000013537419,0.00014909526,0.00048594968,0.00007272457,0.00021277393],"candidate_categories":["scholarly_communication","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00050816266,0.00016530245,0.00019193228,0.00036645064,0.0011187734,0.005076902,0.004464091,0.000054125332,0.11621769],"category_scores_gemma":[0.000096921336,0.00013979687,0.00004573647,0.0013545599,0.0003079631,0.0018884065,0.0073694563,0.0004438172,0.027116839],"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":[0.00005849262,0.00009230048,1.286755e-7,0.00004899495,0.000016910159,0.000013052774,0.00021478006,0.00007360519,0.0069836825,0.0000066916696,0.9831129,0.009378467],"study_design_scores_gemma":[0.00009898011,0.00035206258,0.00017377968,0.0004924835,0.000021946349,0.00006588728,0.00007003388,0.0000048757342,0.0015253111,0.00001184459,0.99702567,0.00015712492],"about_ca_topic_score_codex":0.00012210189,"about_ca_topic_score_gemma":0.000004100759,"teacher_disagreement_score":0.089100845,"about_ca_system_score_codex":0.00018010796,"about_ca_system_score_gemma":0.000009672119,"threshold_uncertainty_score":0.99595594},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.9541819,0.00015834121,0.000059671052,0.0045868997,0.00007490452,0.00014827389,0.00002896923,0.00023021908,0.0405308],"genre_scores_gemma":[0.9940393,0.00021081578,0.000020214713,0.00022165764,0.000049896564,1.6215502e-8,0.00006284136,0.000105197956,0.005290058],"study_design_codex":"qualitative","study_design_gemma":"not_applicable","domain_scores_codex":[0.9987491,0.0004572234,0.00006627777,0.00018865785,0.00033372006,0.00020507269],"domain_scores_gemma":[0.9996612,0.000023204339,0.000036184552,0.00010059373,0.00009661023,0.000082184226],"candidate_categories":["sts","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0008497332,0.000047781727,0.00006281099,0.00007174778,0.9515576,0.0000024138062,0.0005064551,0.000022368202,0.009593019],"category_scores_gemma":[0.00021155228,0.00005213038,0.00001826528,0.00021633385,0.0004174209,0.00009021316,0.83320284,0.00014546352,0.000048518046],"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":[0.000034810175,0.000038193684,0.00023438719,0.000003655652,0.000011860771,0.0000056203135,0.981838,0.000009627537,0.00007885035,0.007207417,0.008780012,0.001757551],"study_design_scores_gemma":[0.00016870971,0.000121349345,0.00054305524,0.0000013729156,0.000005133145,0.000010604752,0.004131726,0.00001856598,0.000010122488,0.0014309675,0.9934962,0.000062206105],"about_ca_topic_score_codex":0.00015499485,"about_ca_topic_score_gemma":0.000014539039,"teacher_disagreement_score":0.9847162,"about_ca_system_score_codex":0.00016463737,"about_ca_system_score_gemma":0.0000019019009,"threshold_uncertainty_score":0.9913123},"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":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000007761817,0.00033643312,7.1951978e-9,0.00008436706,0.00016669347,0.0003067059,0.40908924,0.000082364524,0.5899264],"genre_scores_gemma":[0.00007245312,0.0000399794,0.000030772262,0.00022994682,0.00065110903,0.00015031305,0.30037546,0.0000066997027,0.6984433],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9989506,0.000020023379,0.00010980309,0.00040954063,0.00020988665,0.00030015074],"domain_scores_gemma":[0.99922556,0.00032700482,0.00012784051,0.00011584668,0.000055061602,0.00014866778],"candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00004024035,0.00018309923,0.00017564638,0.000028623399,0.00017038782,0.0005649064,0.0020465537,0.00017574325,0.98667216],"category_scores_gemma":[0.000580651,0.00006493919,0.00006255964,0.000507065,0.00000300999,0.00011107216,0.00080697,0.00018889585,0.22908671],"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":[0.0000017755499,0.000013969426,0.000041514857,0.0000083807845,0.0000043782998,0.0000035093801,0.0000033883955,5.7363415e-8,0.00022023474,0.0000356111,0.9808423,0.018824905],"study_design_scores_gemma":[0.00000837392,0.00002279668,0.0003447103,0.0020598408,0.0000032083465,0.000007086238,0.000011936366,0.0000030290437,0.00004126526,0.000014549298,0.9972674,0.0002158033],"about_ca_topic_score_codex":0.0010637284,"about_ca_topic_score_gemma":0.0002203358,"teacher_disagreement_score":0.75758547,"about_ca_system_score_codex":0.000019064133,"about_ca_system_score_gemma":0.000027992397,"threshold_uncertainty_score":0.7715135},"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000049598543,0.0008601418,4.4401833e-9,0.000019263563,0.000025657284,0.0034983435,0.9955568,0.000011901503,0.000022938688],"genre_scores_gemma":[0.00038054114,0.000007255752,0.000024059085,0.00008934305,0.00003515654,0.003100289,0.9962167,0.000038204133,0.00010845236],"study_design_codex":"not_applicable","study_design_gemma":"systematic_review","domain_scores_codex":[0.9958079,0.00074010075,0.0012342832,0.00059470074,0.0012841591,0.0003388621],"domain_scores_gemma":[0.98524153,0.009555594,0.0027358283,0.0018708422,0.00042037645,0.00017583206],"candidate_categories":["metaresearch","metaepi_narrow","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00034405675,0.0004968511,0.0014043995,0.00034528927,0.000080322374,0.000207241,0.004989247,0.00022771899,0.9493382],"category_scores_gemma":[0.04743213,0.0003330201,0.00020872349,0.00075417507,0.00008259324,0.0009129758,0.006029836,0.0005436059,0.001130398],"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":[0.00014225297,0.00024219383,0.0000015016694,0.17876326,0.00014779872,0.000014204319,0.000007468562,7.6433406e-7,0.000023745684,4.1677745e-7,0.82065487,0.0000015507635],"study_design_scores_gemma":[0.00039424826,0.0001657853,0.00046502776,0.8057051,0.0006723469,0.000014845919,0.000018215374,0.000076234035,0.00072188745,0.0000031867733,0.19144267,0.00032048408],"about_ca_topic_score_codex":0.00023665473,"about_ca_topic_score_gemma":0.0001790698,"teacher_disagreement_score":0.9482078,"about_ca_system_score_codex":0.00013829504,"about_ca_system_score_gemma":0.0015460815,"threshold_uncertainty_score":0.9999122},"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0036890118,0.0016343709,0.0005279876,0.00044290873,0.00093628943,0.00054833904,0.990617,0.000063232634,0.0015409011],"genre_scores_gemma":[0.0104334215,0.0024025457,0.00055576104,0.0006963638,0.0010333481,0.00011527288,0.9845437,0.00006513457,0.00015446082],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99673057,0.000026234926,0.00079152687,0.0019323305,0.00004133802,0.00047797718],"domain_scores_gemma":[0.9967287,0.00030502447,0.0005806467,0.0022263192,0.000021061425,0.00013824618],"candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.00078435044,0.0004466187,0.00063642,0.000295583,0.0002771593,0.00042435154,0.0017577413,0.00021327502,0.00005839009],"category_scores_gemma":[0.000057031153,0.00051500864,0.00014570086,0.00010938391,0.000073864525,0.00035565806,0.00073317206,0.00044117824,0.0011240148],"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":[0.00010310792,0.0001012354,0.0009252904,0.0007062146,0.00035723587,0.0000053451436,0.0005204334,0.0003074884,0.00000217624,0.0073753903,0.9888944,0.0007017044],"study_design_scores_gemma":[0.0004030645,0.000080206904,0.0016131981,0.00031715396,0.00006858196,0.00001082488,0.00034092684,0.04151267,0.0000020039627,0.015634451,0.9394681,0.0005488466],"about_ca_topic_score_codex":0.0011671073,"about_ca_topic_score_gemma":0.0012768233,"teacher_disagreement_score":0.04942631,"about_ca_system_score_codex":0.0002543682,"about_ca_system_score_gemma":0.000029397494,"threshold_uncertainty_score":0.99973017},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016239721,0.009304049,0.11465653,0.0046223556,0.00026280002,0.002117788,0.0048559355,0.0037850952,0.8441557],"genre_scores_gemma":[0.04734858,0.011502075,0.30568162,0.0009446616,0.0017473478,0.0000020300583,0.03383662,0.07352527,0.5254118],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.95399165,0.041378357,0.00075653405,0.0014003814,0.00089371135,0.0015793404],"domain_scores_gemma":[0.993546,0.0021390251,0.0005767649,0.0011700293,0.0017826792,0.0007854992],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.017909061,0.000855654,0.0007583927,0.00092406577,0.0031073752,0.0027869928,0.0022610594,0.0010385397,0.20123915],"category_scores_gemma":[0.037149996,0.0010057022,0.00026374165,0.0019870817,0.0015336393,0.0004976321,0.0021846015,0.0023509085,0.04602627],"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":[0.0005353191,0.0016404563,0.00005292896,0.0029760809,0.0010272338,0.00025606478,0.12302073,0.0015356407,0.13557333,0.10118873,0.4875985,0.14459498],"study_design_scores_gemma":[0.001031003,0.0004817996,0.0011127702,0.0023367861,0.00018648597,0.00064657844,0.010753474,0.002365491,0.004574678,0.010921659,0.96449506,0.0010942051],"about_ca_topic_score_codex":0.00075490016,"about_ca_topic_score_gemma":0.0000758855,"teacher_disagreement_score":0.47689655,"about_ca_system_score_codex":0.0041140895,"about_ca_system_score_gemma":0.00012223166,"threshold_uncertainty_score":0.9999507},"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.031193405,0.004483806,0.000007617411,0.0003243124,0.00010111814,0.0014284138,0.96242213,0.00003771384,0.0000014927778],"genre_scores_gemma":[0.004097813,0.007291254,0.00008173414,0.00018012883,0.00028056433,0.00034358448,0.9876893,0.000033458666,0.0000021360097],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9965058,0.0011092027,0.00057571183,0.0009384816,0.00053873763,0.000332085],"domain_scores_gemma":[0.98728955,0.007135124,0.00050989893,0.00451913,0.0003824545,0.00016385663],"candidate_categories":["metaepi_narrow","sts"],"consensus_categories":[],"category_scores_codex":[0.0019170363,0.00046688615,0.00057072256,0.00008963324,0.001618666,0.00035461423,0.0017222504,0.0002547621,0.0000023089003],"category_scores_gemma":[0.0035669175,0.00041540977,0.000025597088,0.0002521834,0.000545898,0.000502221,0.00063349376,0.0007223586,9.149239e-8],"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":[0.00022651047,0.000082843646,0.075939916,0.00038378825,0.00016684823,0.0000035190576,0.00040090582,2.0510302e-7,0.0000010428324,0.0000033406345,0.9225486,0.0002424955],"study_design_scores_gemma":[0.0003074027,0.00007358108,0.4491823,0.00032015645,0.00022302766,0.000011349066,0.00020789064,0.000016000868,0.00000267646,0.00008989783,0.5493157,0.00025000997],"about_ca_topic_score_codex":0.9989177,"about_ca_topic_score_gemma":0.999993,"teacher_disagreement_score":0.3732424,"about_ca_system_score_codex":0.00031372375,"about_ca_system_score_gemma":0.0014813746,"threshold_uncertainty_score":0.99982977},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0016567968,0.00025758203,0.00025900986,0.0030386983,0.00030608985,0.0013961303,0.9925756,0.000012465767,0.0004976591],"genre_scores_gemma":[0.0039378344,0.0032216562,0.004273788,0.00006314717,0.00013735947,0.000044874458,0.9882001,0.0000074790887,0.00011373417],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9908901,0.0026391628,0.0007395129,0.0020065208,0.0031646094,0.00056011946],"domain_scores_gemma":[0.9849961,0.0079046115,0.0010115871,0.0037473412,0.0020229642,0.00031743996],"candidate_categories":["metaresearch","sts","open_science"],"consensus_categories":["metaresearch","sts"],"category_scores_codex":[0.16623333,0.000154045,0.00034680587,0.00080065074,0.0021930886,0.0010057887,0.021071846,0.00048480765,0.00013121468],"category_scores_gemma":[0.22134098,0.00013624731,0.000019736575,0.004227165,0.014682801,0.005906085,0.0069825454,0.0021369567,0.000035106266],"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":[0.000047599926,0.000075123964,0.0016845153,0.00020983195,0.000018146575,3.2239504e-7,0.0028647159,5.7987455e-7,0.000056264118,0.0022737833,0.9918859,0.00088321895],"study_design_scores_gemma":[0.00014571114,0.0001897992,0.006178715,0.00037060824,0.00006193314,0.0000019325055,0.086651556,0.0024341068,0.000016193593,0.0022609157,0.9014652,0.00022332187],"about_ca_topic_score_codex":0.37542096,"about_ca_topic_score_gemma":0.2635356,"teacher_disagreement_score":0.11188537,"about_ca_system_score_codex":0.00016316243,"about_ca_system_score_gemma":0.013904553,"threshold_uncertainty_score":0.99910593},"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"methods","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005672075,0.000165443,0.90892655,0.0052687908,0.00022218202,0.0012451775,0.000026830636,0.003730335,0.07474263],"genre_scores_gemma":[0.78278357,0.00030844563,0.21499261,0.00006247043,0.00009688672,5.421284e-7,0.000051111332,0.000738686,0.00096567906],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9968695,0.00024846024,0.00018570972,0.0011763878,0.00093630183,0.00058365846],"domain_scores_gemma":[0.9974506,0.00008132907,0.00007492524,0.0017581554,0.00045846065,0.00017655033],"candidate_categories":["sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.0043031,0.00014684547,0.00012816313,0.0006312993,0.003556107,0.12993963,0.03474823,0.000044560933,0.00024569067],"category_scores_gemma":[0.0021042451,0.0001283563,0.000012703349,0.003759506,0.00042928493,0.069479965,0.054235373,0.0005241362,0.00090173376],"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":[0.000019570823,0.00006516664,0.000004757697,0.00003931645,0.000024549508,0.000077714554,0.00031914277,0.000037219852,0.0034764581,0.6598119,0.0029056866,0.33321851],"study_design_scores_gemma":[0.00017842681,0.00053817406,0.00026557947,0.00025011585,0.0000067434084,0.00009674731,0.0006951624,0.004129609,0.0037628345,0.0069010328,0.98291594,0.00025962866],"about_ca_topic_score_codex":0.000057451103,"about_ca_topic_score_gemma":2.5720945e-7,"teacher_disagreement_score":0.9800103,"about_ca_system_score_codex":0.00023122982,"about_ca_system_score_gemma":0.00003173708,"threshold_uncertainty_score":0.9998762},"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":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0000047340463,0.0008867016,0.00009903072,0.00069303514,0.00021117774,0.0013698001,0.008132305,0.0030100222,0.9855932],"genre_scores_gemma":[0.0013630361,0.00018375543,0.00026052212,0.00012202769,0.0005169969,5.3189535e-8,0.0152579,0.115211375,0.8670843],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9967327,0.00072568253,0.00040328648,0.00097088446,0.0007281845,0.00043925806],"domain_scores_gemma":[0.9953163,0.000026529917,0.00036627782,0.0038266736,0.00030849938,0.00015570829],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.0019624229,0.00031454818,0.00037324792,0.0005350009,0.00056621106,0.0021756415,0.01539089,0.0001677062,0.22906671],"category_scores_gemma":[0.0009411983,0.0002514782,0.000065271684,0.0013198594,0.00048660216,0.00027020168,0.03191491,0.00062311755,0.5243863],"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":[0.000042514665,0.00009157331,1.3362718e-7,0.00020941354,0.0002579117,0.000013127068,0.00034790096,0.0000016883546,0.0004326514,0.0048777196,0.98569417,0.0080311755],"study_design_scores_gemma":[0.00036025536,0.00008615891,0.0000067241076,0.00029380393,0.000107661705,0.000057386158,0.00030741424,0.00011536997,0.000040116818,0.00016714951,0.9981925,0.0002654757],"about_ca_topic_score_codex":0.00030976778,"about_ca_topic_score_gemma":0.000005431555,"teacher_disagreement_score":0.2953196,"about_ca_system_score_codex":0.00012241557,"about_ca_system_score_gemma":0.000011024196,"threshold_uncertainty_score":0.99999374},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0047014127,0.000062193976,0.027581284,0.010985811,0.000348303,0.0019848663,0.000097426615,0.00025353872,0.95398515],"genre_scores_gemma":[0.99456096,0.0004624958,0.00093411584,0.0006786266,0.000040926683,5.6650293e-7,0.00003877214,0.00019529312,0.0030882598],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.9974555,0.00048372787,0.00020380567,0.000513382,0.000937916,0.00040568423],"domain_scores_gemma":[0.99769723,0.0000868243,0.00015680467,0.0017257673,0.00026268128,0.00007067181],"candidate_categories":["sts","scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.003909093,0.00009531567,0.00008690853,0.000159242,0.004240332,0.01105973,0.0222035,0.000018376326,0.0000835008],"category_scores_gemma":[0.0015058162,0.00006166282,0.000033377237,0.0017655067,0.00058456225,0.007599946,0.03859458,0.00024562704,0.0002296234],"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":[0.00004504938,0.00014454579,0.00001093703,0.00003508122,0.000020270312,0.0000013554954,0.00039483167,0.000007687904,0.0010279852,0.8813749,0.016676228,0.100261144],"study_design_scores_gemma":[0.00033578507,0.00015243555,0.00465037,0.0000611851,0.0000055881465,0.000008514041,0.00065491773,0.003222633,0.0016566786,0.00046055656,0.9886945,0.00009684635],"about_ca_topic_score_codex":0.000029344139,"about_ca_topic_score_gemma":5.1563103e-7,"teacher_disagreement_score":0.9898595,"about_ca_system_score_codex":0.00020455006,"about_ca_system_score_gemma":0.000012036759,"threshold_uncertainty_score":0.997056},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.67027265,0.00005577949,0.06707608,0.042306844,0.00021473253,0.0009494769,0.000058189016,0.00030594022,0.2187603],"genre_scores_gemma":[0.991987,0.000020349336,0.0044342615,0.0021505195,0.00006425927,1.3076131e-7,0.00040972844,0.00018955828,0.00074420794],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.99498427,0.0018715567,0.0006209805,0.0005060068,0.0015736311,0.0004435775],"domain_scores_gemma":[0.99715126,0.00037762587,0.00022029084,0.00071524613,0.0014009018,0.00013469216],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.022429941,0.000107246706,0.00018361415,0.00044588532,0.0046128347,0.014363285,0.0096365055,0.00004746244,0.00290687],"category_scores_gemma":[0.013014897,0.000085182466,0.00003237905,0.0040223277,0.00022155001,0.002103067,0.0064396258,0.0008051076,0.0004897765],"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":[0.000058002985,0.00023900224,0.0005369597,0.000022571396,0.000017545271,0.00010641357,0.07291175,0.003188008,0.002053538,0.025791712,0.06829435,0.82678014],"study_design_scores_gemma":[0.00029207682,0.00006845035,0.0102472715,0.00011236687,0.0000038218054,0.00016413064,0.082453124,0.0032270772,0.00019587675,0.0027329738,0.900321,0.00018183482],"about_ca_topic_score_codex":0.00011410471,"about_ca_topic_score_gemma":0.0000143389625,"teacher_disagreement_score":0.83202666,"about_ca_system_score_codex":0.00027258357,"about_ca_system_score_gemma":0.00008158374,"threshold_uncertainty_score":0.9980046},"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02511466,0.000061317354,0.00013092214,0.0065167453,0.0000552364,0.0015624035,0.9665464,0.0000015947205,0.000010713534],"genre_scores_gemma":[0.048670046,0.00038514825,0.00027791972,0.000656793,0.00032749563,0.00016943304,0.94949794,7.4836987e-7,0.000014478571],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9985708,0.0004669763,0.0004428181,0.000270172,0.0001434608,0.000105784115],"domain_scores_gemma":[0.9960761,0.0027041666,0.0006021647,0.0004231865,0.0001691796,0.000025222098],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0021730217,0.000115898096,0.00024113231,0.000020035834,0.00015945979,0.000047790472,0.002047688,0.00007532484,0.0000030538204],"category_scores_gemma":[0.00080454134,0.000027403581,0.000038637605,0.00020500849,0.000088396606,0.000083113635,0.00024542428,0.00022124131,5.1803475e-8],"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":[0.0001393213,0.00039025772,0.00010207219,0.000037182475,0.0001121149,2.6942894e-7,0.00010552486,0.00009289074,0.0028433134,0.0019762034,0.9915167,0.002684145],"study_design_scores_gemma":[0.00023173727,0.0004606666,0.39498878,0.00001770011,0.00017054293,0.000022870629,0.0007464072,0.0005869109,0.000007752154,0.00010065137,0.6025783,0.000087712935],"about_ca_topic_score_codex":0.0064634182,"about_ca_topic_score_gemma":0.0047133914,"teacher_disagreement_score":0.3948867,"about_ca_system_score_codex":0.000013062436,"about_ca_system_score_gemma":0.000055523084,"threshold_uncertainty_score":0.9770798},"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":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00044518017,0.002855108,0.0000028004445,0.000096439544,0.00048642606,0.001318027,0.4386435,0.00033567465,0.5558168],"genre_scores_gemma":[0.01862195,0.000050402465,0.00034333704,0.00011795182,0.0015610602,0.00079901324,0.26406905,0.0016410603,0.71279615],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9952411,0.00036157345,0.0008238847,0.0015230103,0.00061866746,0.0014317828],"domain_scores_gemma":[0.99684024,0.00016737726,0.00064410374,0.0015610277,0.0003712516,0.00041600387],"candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","insufficient_payload"],"category_scores_codex":[0.00015208284,0.0013358212,0.0011949851,0.0005722034,0.00041729512,0.0005932131,0.0023756602,0.0016385439,0.81269425],"category_scores_gemma":[0.0021443493,0.0015241147,0.00042370017,0.001203427,0.0002082419,0.0018279399,0.0012356394,0.0011471,0.6471856],"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":[0.000068712376,0.00085648434,0.0021855044,0.0024748228,0.00018353082,0.000379524,0.0012159357,0.001424782,0.0015479282,0.0009032481,0.9850747,0.0036848302],"study_design_scores_gemma":[0.0021741341,0.00011448489,0.018464003,0.022769712,0.000036260873,0.00007741001,0.00036694104,0.0006640491,0.00064734626,0.00003827139,0.9530365,0.0016109204],"about_ca_topic_score_codex":0.0025507961,"about_ca_topic_score_gemma":0.011904062,"teacher_disagreement_score":0.17457443,"about_ca_system_score_codex":0.0012611444,"about_ca_system_score_gemma":0.0028984859,"threshold_uncertainty_score":0.9999393},"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[1.0645151e-7,0.001477372,1.4689024e-9,0.000073399526,0.00075807516,0.0017234803,0.9746918,0.00019620851,0.021079544],"genre_scores_gemma":[6.6107657e-7,0.000017394228,0.0000060754396,0.00019979423,0.002045885,0.0012991758,0.979433,0.000393035,0.016605007],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9950469,0.00012282234,0.00093381805,0.0019885474,0.0011147604,0.00079313613],"domain_scores_gemma":[0.99433136,0.00011392226,0.00092966936,0.004006443,0.00024930504,0.0003692829],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.00031569536,0.0009125023,0.0010148291,0.0005809122,0.00014885371,0.0072707427,0.012148623,0.0007297049,0.9786863],"category_scores_gemma":[0.0032343545,0.00089429313,0.00031936163,0.001261042,0.000013908903,0.0015811174,0.02293082,0.0017451758,0.96068645],"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":[0.000016249889,0.00006692945,6.913392e-7,0.0018864885,0.00020651928,0.02909827,0.000002844523,0.0000014615238,7.4708e-7,3.3890944e-7,0.968659,0.000060471077],"study_design_scores_gemma":[0.00019017926,0.000026162308,0.000039136205,0.01462461,0.00017688215,0.00089424907,0.000003422876,0.0000029543585,0.00003127705,0.00007913807,0.98285747,0.0010745084],"about_ca_topic_score_codex":0.0005450624,"about_ca_topic_score_gemma":0.0008906592,"teacher_disagreement_score":0.028204022,"about_ca_system_score_codex":0.00042203377,"about_ca_system_score_gemma":0.001028687,"threshold_uncertainty_score":0.9993508},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.06578057,0.00571653,0.006961302,0.029174706,0.0005215613,0.0015950595,0.0010190357,0.00072215573,0.8885091],"genre_scores_gemma":[0.30308434,0.0014507235,0.011156427,0.0024926548,0.0012903194,0.0006743159,0.00006758624,0.0015807226,0.6782029],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9790801,0.007997234,0.0019252921,0.0036097658,0.0050254506,0.0023621698],"domain_scores_gemma":[0.94526803,0.048052423,0.0023439443,0.0022084005,0.0007166608,0.0014105428],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","sts","insufficient_payload"],"category_scores_codex":[0.016139526,0.0022253897,0.0015940772,0.006523968,0.00196508,0.0031099145,0.007256997,0.0011144341,0.04085372],"category_scores_gemma":[0.058520164,0.0023916261,0.0010761082,0.006974915,0.013408065,0.0026508747,0.0037245776,0.0030262785,0.013266599],"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":[0.0005125552,0.0028572448,0.07216646,0.0007972103,0.0010813812,0.00010501405,0.01950474,0.0004087629,0.049388014,0.37831825,0.4715126,0.003347772],"study_design_scores_gemma":[0.0024580308,0.00024214719,0.087181985,0.0016225667,0.00032242006,0.00028105578,0.0005466428,0.0031248517,0.0057671466,0.0021535077,0.89395237,0.0023472852],"about_ca_topic_score_codex":0.7912874,"about_ca_topic_score_gemma":0.99690175,"teacher_disagreement_score":0.52560484,"about_ca_system_score_codex":0.06688026,"about_ca_system_score_gemma":0.59248513,"threshold_uncertainty_score":0.9993342},"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":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0056920527,0.00022443928,4.0669732e-7,0.0008140197,0.000610597,0.0043504513,0.00010925192,0.0010252038,0.98717356],"genre_scores_gemma":[0.20787027,0.00020284115,0.0000063812226,0.00023202592,0.00030195032,0.0000043266173,0.0013270311,0.02312531,0.76692986],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9954088,0.00087262463,0.0004444429,0.0011421809,0.001187333,0.00094466785],"domain_scores_gemma":[0.99632263,0.000035820685,0.0006205326,0.0022302656,0.00040053655,0.00039022948],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.0015316125,0.00047889224,0.0005490558,0.0007822005,0.0006809023,0.0066090906,0.014817986,0.00015180933,0.11217792],"category_scores_gemma":[0.00072278356,0.0005215221,0.000057185072,0.00083397486,0.00017488365,0.0014783745,0.013921135,0.0006336858,0.040906414],"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":[0.000049671344,0.000070246504,0.00059503893,0.00019797393,0.000098543904,0.000097078315,0.00007370632,0.0000137379,0.0000038385624,0.0070901983,0.99146134,0.00024863367],"study_design_scores_gemma":[0.0014911904,0.000041772666,0.0028367597,0.00056816475,0.000022428585,0.000007156664,0.000077044984,0.0000012475567,0.00004779948,0.0006980581,0.99362814,0.00058023504],"about_ca_topic_score_codex":0.87974524,"about_ca_topic_score_gemma":0.83843786,"teacher_disagreement_score":0.22024372,"about_ca_system_score_codex":0.0039832867,"about_ca_system_score_gemma":0.0007143697,"threshold_uncertainty_score":0.99984026},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0034108919,0.0000282134,0.00001145983,0.0015976339,0.0002357119,0.003094532,0.00048681046,0.000054957418,0.9910798],"genre_scores_gemma":[0.03583127,0.0008174374,0.0013347557,0.0003161181,0.00009699734,0.0011138702,0.00044453584,0.00043352623,0.9596115],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98123014,0.007666084,0.00091569684,0.005759079,0.0034781962,0.0009508275],"domain_scores_gemma":[0.98176414,0.0017483246,0.0017651517,0.010240743,0.0039069136,0.00057471613],"candidate_categories":["metaresearch","metaepi_narrow","open_science","insufficient_payload"],"consensus_categories":["metaresearch","open_science","insufficient_payload"],"category_scores_codex":[0.12830313,0.00039517324,0.0006406873,0.003056971,0.0007277069,0.0009807707,0.010044793,0.00057127,0.40518695],"category_scores_gemma":[0.087812394,0.00039135333,0.00006952687,0.004995721,0.0018373517,0.0004400535,0.008708916,0.0013979065,0.30671933],"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":[0.000096279284,0.00044891523,0.10831746,0.000076818455,0.0002749053,0.0000092913815,0.00042843446,0.000017818185,0.0008615859,0.003192844,0.8856041,0.0006715737],"study_design_scores_gemma":[0.00042764843,0.0000014650424,0.41634056,0.00018415581,0.00003600958,0.0000505694,0.00016015471,0.000028270839,0.000217852,0.0020726642,0.580201,0.0002796515],"about_ca_topic_score_codex":0.79959553,"about_ca_topic_score_gemma":0.6365774,"teacher_disagreement_score":0.3080231,"about_ca_system_score_codex":0.0013125531,"about_ca_system_score_gemma":0.008595941,"threshold_uncertainty_score":0.99985385},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.87025726,0.000025131494,0.037989307,0.00075073884,0.000027414764,0.0014250526,0.00011027401,0.000051090883,0.08936374],"genre_scores_gemma":[0.9530567,0.0002493571,0.041966904,0.00004249201,0.00004673182,7.197058e-7,0.00009907632,0.000017055221,0.0045209634],"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.9970722,0.00010669614,0.0001268813,0.0010357394,0.0007441343,0.0009143507],"domain_scores_gemma":[0.9965227,0.000068621826,0.00013078682,0.0009181585,0.0019922745,0.00036742532],"candidate_categories":["sts","open_science"],"consensus_categories":["sts"],"category_scores_codex":[0.0019068245,0.00017984734,0.00026659318,0.0011797192,0.0018384306,0.00011466966,0.006245394,0.00010636019,0.00006324764],"category_scores_gemma":[0.0004065511,0.0002205594,0.00016280971,0.0024920914,0.0102670295,0.00018361934,0.007314348,0.00024419936,0.000004368028],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.001066747,0.00019812642,0.0019328165,0.000028274198,0.00005976489,0.000013333946,0.0003068823,0.0000016945631,0.97454983,0.006283173,0.009712649,0.005846721],"study_design_scores_gemma":[0.001482073,0.0017441706,0.0013137195,0.000047177793,0.00004334522,0.000008911359,0.0046153227,0.001695287,0.5781136,0.002171959,0.40838933,0.0003750903],"about_ca_topic_score_codex":0.000443305,"about_ca_topic_score_gemma":0.0007399517,"teacher_disagreement_score":0.39867666,"about_ca_system_score_codex":0.00019584164,"about_ca_system_score_gemma":0.0018171645,"threshold_uncertainty_score":0.99946105},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.012674059,0.0014072667,0.00054208725,0.000028707189,0.0005100305,0.0013972467,0.0006176625,0.00023669201,0.98258626],"genre_scores_gemma":[0.019663576,0.00019873933,0.0002176951,0.000009867342,0.00022317028,0.000004407223,0.000046395548,0.00043273158,0.9792034],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9963523,0.00064717507,0.0005410114,0.001079876,0.00083007413,0.0005495513],"domain_scores_gemma":[0.9953426,0.0002263706,0.0019348967,0.0017123906,0.0005423261,0.00024138782],"candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.000622221,0.00059122057,0.001030654,0.0034138807,0.0004954716,0.0010039238,0.005198328,0.0004600998,0.00028513063],"category_scores_gemma":[0.00009981288,0.00054161204,0.00029155496,0.0054192236,0.0011347247,0.0012283137,0.0037277641,0.00034520935,0.0003795829],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006960741,0.00014968963,0.25067443,0.00021858339,0.011438475,0.00042940964,0.00017112501,0.00007135873,0.0012936784,0.0033314694,0.73133224,0.00019345786],"study_design_scores_gemma":[0.0013265201,0.00012264616,0.12547901,0.0005077634,0.0068100933,0.000034604614,0.00043757437,0.00018621967,0.0003282493,0.0002177977,0.86354065,0.0010088986],"about_ca_topic_score_codex":0.05033275,"about_ca_topic_score_gemma":0.034716547,"teacher_disagreement_score":0.13220838,"about_ca_system_score_codex":0.0009899228,"about_ca_system_score_gemma":0.0007430059,"threshold_uncertainty_score":0.9997035},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","open_science"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"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":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00026385288,0.000120741766,0.00040770834,0.00006985958,0.00059368525,0.0018091633,0.001629098,0.0009295352,0.9941763],"genre_scores_gemma":[0.0010273447,0.000048657545,0.0005682662,0.00019377103,0.0006795484,6.563966e-7,0.00030657175,0.0010425558,0.9961326],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9943513,0.00044482457,0.0005172407,0.0022248584,0.001120705,0.0013410213],"domain_scores_gemma":[0.99404556,0.00027612457,0.0015264618,0.0019268563,0.00039551107,0.0018294622],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaepi_narrow","scholarly_communication","open_science"],"category_scores_codex":[0.000009271238,0.0014678683,0.001468054,0.003636018,0.0013524917,0.009156565,0.023879232,0.0009792702,0.0003339064],"category_scores_gemma":[0.00035260225,0.0016660824,0.00039169274,0.0032782832,0.00044963483,0.022142407,0.0305724,0.00108559,0.0008140762],"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":[0.000803406,0.0001724994,0.010624657,0.000107988126,0.0007767529,0.0009173897,0.000021142685,0.0000052761575,0.00091382203,0.0039181598,0.98016775,0.0015711879],"study_design_scores_gemma":[0.0031975152,0.00010215341,0.00065829913,0.0014394975,0.0002584656,0.000074922886,0.0024985617,2.5445084e-7,0.000276684,0.00044359048,0.9890658,0.0019842058],"about_ca_topic_score_codex":0.80418766,"about_ca_topic_score_gemma":0.8836169,"teacher_disagreement_score":0.07942924,"about_ca_system_score_codex":0.0073130676,"about_ca_system_score_gemma":0.028123047,"threshold_uncertainty_score":0.9999639},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0008464614,0.012101291,0.000004541992,0.0057014674,0.0005540851,0.005794854,0.0003416034,0.00018636588,0.9744693],"genre_scores_gemma":[0.047873132,0.012039454,0.0003069141,0.00008146688,0.0027182873,0.0006636565,0.00014189033,0.0006590377,0.9355162],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.93530965,0.023344032,0.0011141881,0.008086777,0.02571132,0.0064340592],"domain_scores_gemma":[0.982764,0.007692936,0.0001395534,0.0066074976,0.0012460205,0.0015499893],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","sts","open_science","insufficient_payload"],"category_scores_codex":[0.22549309,0.00070031616,0.0010188189,0.0025384268,0.0023227283,0.0033941437,0.007825333,0.0008982452,0.0087859295],"category_scores_gemma":[0.068935886,0.0006194286,0.0002976648,0.002964389,0.014243465,0.00079135416,0.013684429,0.018619545,0.019030131],"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":[0.00043581397,0.0006256225,0.00014558445,0.00022670953,0.00008271788,0.0012257897,0.007644641,0.000009754066,0.00019562908,0.027527945,0.95553035,0.006349463],"study_design_scores_gemma":[0.00046266275,0.0003489925,0.00009719852,0.00082565047,0.0000068544527,0.000005232742,0.004043396,0.00005433592,0.000008542612,0.0429101,0.9506399,0.00059714256],"about_ca_topic_score_codex":0.025333973,"about_ca_topic_score_gemma":0.02931472,"teacher_disagreement_score":0.1565572,"about_ca_system_score_codex":0.016018663,"about_ca_system_score_gemma":0.0074892957,"threshold_uncertainty_score":0.9996257},"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":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.000013689911,0.00014261673,0.036743324,0.0073994175,0.0001306001,0.0016306705,0.00045593403,0.00059799885,0.95288575],"genre_scores_gemma":[0.0071652965,0.000055169992,0.019675465,0.007366574,0.0011567944,0.00071829057,0.001322297,0.00069004035,0.96185005],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99814093,0.00006383526,0.0002832454,0.0008357238,0.00035893876,0.00031732305],"domain_scores_gemma":[0.9975066,0.00033253702,0.00042039077,0.001546176,0.000034785888,0.00015947597],"candidate_categories":["metaepi_narrow","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.000087268534,0.00034615363,0.00033646473,0.0002389179,0.00064710336,0.06841675,0.023794195,0.00010415225,0.00037195956],"category_scores_gemma":[0.000020698968,0.00025187113,0.00008234972,0.0024457374,0.00008342792,0.015034734,0.017045023,0.00031291277,0.0000697287],"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":[0.000011404469,0.00011870014,0.000053804553,0.000010265421,0.00007386285,0.0000066739635,0.00012444041,0.00010705466,1.2291251e-7,0.07091238,0.9109154,0.017665923],"study_design_scores_gemma":[0.0003491956,0.000052461055,0.00032538656,0.00015888104,0.0000106993275,0.000016243523,0.00004239608,0.0005911218,0.0000047978087,0.08882021,0.9092372,0.0003913771],"about_ca_topic_score_codex":0.00038669608,"about_ca_topic_score_gemma":0.00015035801,"teacher_disagreement_score":0.05338202,"about_ca_system_score_codex":0.000025471081,"about_ca_system_score_gemma":0.00032852302,"threshold_uncertainty_score":0.9999933},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.016075168,0.00006351365,0.00010375958,0.0009849473,0.0014833099,0.002759244,0.00074121763,0.0007469755,0.97704184],"genre_scores_gemma":[0.0052362797,0.0008583361,0.00042038682,0.0005382687,0.00050794875,0.00040569188,0.0014054031,0.0014049899,0.9892227],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9840312,0.0013627106,0.0009995763,0.011405646,0.0010337103,0.001167146],"domain_scores_gemma":[0.97164816,0.00081662234,0.0008971606,0.025971845,0.00012068851,0.0005455572],"candidate_categories":["metaresearch","metaepi_narrow","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.006156449,0.0012583154,0.0013267838,0.00029190377,0.00052380725,0.0007622211,0.011468932,0.00096165284,0.68781054],"category_scores_gemma":[0.008639495,0.0010934573,0.00027486967,0.0005383497,0.0013095449,0.00048725316,0.06920723,0.0019734758,0.75399095],"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":[0.00021188722,0.0006584024,0.034757998,0.00027034382,0.00066119764,0.0009987496,0.00038420325,0.00028607922,0.00046141385,0.0007612616,0.94816023,0.01238821],"study_design_scores_gemma":[0.0010051372,0.0000022904194,0.0403909,0.00026433697,0.00032606925,0.00035049056,0.00003877838,0.0010113232,0.00008664874,0.00042125193,0.95501614,0.0010866598],"about_ca_topic_score_codex":0.004025021,"about_ca_topic_score_gemma":0.011201144,"teacher_disagreement_score":0.066180415,"about_ca_system_score_codex":0.00024973883,"about_ca_system_score_gemma":0.0002649699,"threshold_uncertainty_score":0.99971116},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00024714743,0.000011094719,0.0012008408,0.00017504684,0.0013705045,0.0043398626,0.0002692248,0.0020538506,0.9903324],"genre_scores_gemma":[0.0021180178,0.00017692588,0.0016933307,0.00011596344,0.00067615556,0.0006651041,0.00043382944,0.0029653984,0.99115527],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9881221,0.0013326904,0.0012067525,0.006588197,0.0014922415,0.0012580046],"domain_scores_gemma":[0.9801302,0.0005664274,0.0016506318,0.01690794,0.00022383545,0.00052095397],"candidate_categories":["metaepi_narrow","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","insufficient_payload"],"category_scores_codex":[0.0054528173,0.0013698173,0.00140707,0.0009185921,0.00015642501,0.0004434136,0.0036091942,0.0019392814,0.9189384],"category_scores_gemma":[0.0044854386,0.0016175251,0.0006478499,0.00068940286,0.00042755337,0.00037829444,0.0027429017,0.0018015779,0.9977701],"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":[0.00017435072,0.0005225492,0.0021780957,0.0004708509,0.00070166145,0.00004589063,0.00047601046,0.00040539738,0.0017403904,0.0025529398,0.98476964,0.005962232],"study_design_scores_gemma":[0.0010760125,8.719554e-7,0.00094298786,0.0008616041,0.0002855035,0.00005361615,0.00010083347,0.00025420368,0.0010253063,0.00093939016,0.99300045,0.0014592063],"about_ca_topic_score_codex":0.00083492935,"about_ca_topic_score_gemma":0.00086997624,"teacher_disagreement_score":0.07883171,"about_ca_system_score_codex":0.00090195163,"about_ca_system_score_gemma":0.00071490335,"threshold_uncertainty_score":0.9999052},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.8079669,0.00011496709,0.0000033707458,0.000049615013,0.00062006357,0.0007084246,0.010137813,0.00005046132,0.18034843],"genre_scores_gemma":[0.9760474,0.000043043572,0.00066548417,0.000090904614,0.0002469818,0.0001138211,0.006418841,0.00010691876,0.016266583],"study_design_codex":"design_other","study_design_gemma":"not_applicable","domain_scores_codex":[0.9975361,0.000105630075,0.0005016493,0.0010661037,0.00036095886,0.00042951704],"domain_scores_gemma":[0.997646,0.00014536764,0.0006507334,0.0011464655,0.0001364576,0.00027500102],"candidate_categories":["metaepi_narrow","sts"],"consensus_categories":[],"category_scores_codex":[0.00024237571,0.0005193313,0.00059466047,0.00007646644,0.0024914863,0.000679293,0.0012850899,0.00014404283,0.0002918818],"category_scores_gemma":[0.00009654395,0.000552029,0.00009334958,0.00008748162,0.00009047969,0.0022367912,0.0013032374,0.00095037895,0.00005141719],"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":[0.00023870345,0.0004098519,0.0011987849,0.00015809224,0.0009424965,0.000012893518,0.00004344939,0.00002279216,0.012512779,0.32066107,0.00017333205,0.6636258],"study_design_scores_gemma":[0.0061288546,0.00042323812,0.06254447,0.0025310204,0.0020937698,0.0000052792675,0.006043119,0.0012714879,0.010743409,0.22339806,0.6800559,0.0047613573],"about_ca_topic_score_codex":0.0017587463,"about_ca_topic_score_gemma":0.0016111123,"teacher_disagreement_score":0.6798826,"about_ca_system_score_codex":0.00012263053,"about_ca_system_score_gemma":0.000048552545,"threshold_uncertainty_score":0.9996931},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.025071653,0.0008995841,0.0018362937,0.0643887,0.0020171881,0.0020428072,0.0012988042,0.00014503136,0.90229994],"genre_scores_gemma":[0.98769355,0.00024725875,0.00008600809,0.0025486553,0.00012332536,5.100483e-8,0.0003655538,0.0004388525,0.008496721],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9917728,0.0015664765,0.0010440202,0.0019020591,0.0027203488,0.0009943324],"domain_scores_gemma":[0.99256325,0.00053834677,0.0003938195,0.0044521964,0.0016939553,0.00035846088],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["open_science","insufficient_payload"],"category_scores_codex":[0.02705862,0.00024674868,0.00036755454,0.00073799543,0.009303901,0.03410503,0.11018926,0.00010174819,0.015903559],"category_scores_gemma":[0.037371177,0.00018923363,0.000030943673,0.010614804,0.0013219533,0.006475671,0.11203058,0.0013601787,0.0028523998],"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":[0.00004240036,0.000046441168,0.00016815763,0.000009865002,0.000011913756,0.000012068599,0.00041418156,0.0000895718,0.000044155782,0.0054790173,0.6615917,0.33209053],"study_design_scores_gemma":[0.0005054216,0.00002775254,0.00946036,0.000115915864,0.0000118053085,0.000027325232,0.0066109803,0.009386432,0.000055147422,0.0014117841,0.97217256,0.00021452701],"about_ca_topic_score_codex":0.42761952,"about_ca_topic_score_gemma":0.042418182,"teacher_disagreement_score":0.9626219,"about_ca_system_score_codex":0.0017511636,"about_ca_system_score_gemma":0.0016094833,"threshold_uncertainty_score":0.997924},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0022110245,0.003745128,0.00013495341,0.014023291,0.0053227753,0.0014131578,0.091767535,0.00006196359,0.8813202],"genre_scores_gemma":[0.9792178,0.00009769943,0.00004804891,0.0055899564,0.002688091,0.000101967,0.0032996994,0.00002489075,0.008931836],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9970924,0.00042673075,0.0008893572,0.00047772948,0.0004348737,0.0006789024],"domain_scores_gemma":[0.9975986,0.0005131205,0.00049060443,0.00025254878,0.00041142624,0.0007336834],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00083812996,0.00030682914,0.00070485636,0.00027827028,0.0026399384,0.009225836,0.0047168373,0.00020822149,0.5020847],"category_scores_gemma":[0.0010756869,0.00029977417,0.00027112156,0.0010003335,0.00004424813,0.003531817,0.0014074553,0.00088603154,0.0030048299],"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":[0.00004332638,0.00014270328,0.00031318452,0.000041731684,0.00008564873,0.000013875241,0.0030927225,0.00033146373,0.0000028663903,0.002019857,0.98916906,0.004743579],"study_design_scores_gemma":[0.0004666991,0.000036357906,0.007827732,0.0050063473,0.000018969233,0.000012971105,0.00082137913,0.00060200354,0.000021874943,0.0024969603,0.9823513,0.00033740862],"about_ca_topic_score_codex":0.0001797196,"about_ca_topic_score_gemma":0.00033473875,"teacher_disagreement_score":0.9770068,"about_ca_system_score_codex":0.0015078182,"about_ca_system_score_gemma":0.0016998173,"threshold_uncertainty_score":0.99994546},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.87498283,0.003774993,0.023941794,0.006343271,0.0025542,0.0044453605,0.0005016335,0.00084136997,0.08261454],"genre_scores_gemma":[0.9602322,0.0021269368,0.0013745321,0.0012189747,0.0006726553,7.2086283e-7,0.002905393,0.0009942211,0.03047436],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9981175,0.00017405086,0.00021911957,0.0010450822,0.00014310605,0.00030112793],"domain_scores_gemma":[0.99791706,0.000011032867,0.00006653006,0.0013422151,0.0005827374,0.00008045161],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001251837,0.00013184421,0.00012540947,0.000057006448,0.001203951,0.00051945687,0.0006050103,0.00011912677,0.00039859748],"category_scores_gemma":[0.001185078,0.00012103902,0.0001039591,0.00019194144,0.00020442334,0.00002058522,0.0008602613,0.00011534029,0.00017556899],"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":[0.00036558483,0.00040083565,0.000096369855,0.0003750529,0.00022826665,0.0000022534116,0.0021839738,0.0000560411,0.24278897,0.0016002408,0.6591911,0.09271133],"study_design_scores_gemma":[0.000465344,0.00022897578,0.00034933753,0.00002231364,0.000017209795,0.0000063701614,0.0014059595,0.0000715945,0.041896023,0.00036704628,0.95502174,0.00014806616],"about_ca_topic_score_codex":0.000004764011,"about_ca_topic_score_gemma":0.0000018446477,"teacher_disagreement_score":0.29583067,"about_ca_system_score_codex":0.000048448568,"about_ca_system_score_gemma":0.0000146288,"threshold_uncertainty_score":0.9259939},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.05503981,0.0000032386085,0.0007596334,0.013447867,0.000075861215,0.32918942,0.000002563268,0.0001023822,0.6013792],"genre_scores_gemma":[0.86051977,4.2622574e-7,0.0001680353,0.0011688807,0.00011249721,0.1097641,8.3156056e-7,0.000013960969,0.028251516],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9993343,0.00004100318,0.00013501476,0.00017237301,0.00016500043,0.00015232054],"domain_scores_gemma":[0.9988845,0.00053833844,0.00003798986,0.00041431072,0.000040320294,0.000084557774],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00023325387,0.00008035742,0.00012308019,0.000019066172,0.00008412008,0.00015676454,0.0006627778,0.000025524321,0.00079304446],"category_scores_gemma":[0.00087744446,0.000029354056,0.000023493536,0.00009325395,0.000027277118,0.00030492386,0.0011723865,0.00006119953,0.0001451107],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005318858,0.0008290594,0.86560756,0.000022018974,0.00008733948,0.00005829668,0.00034969338,9.302728e-7,0.00020280774,0.002298536,0.03763939,0.0923725],"study_design_scores_gemma":[0.01283296,0.0017359514,0.8818524,0.00046313333,0.000122681,0.000044799403,0.0006880249,0.000052396455,0.0060607945,0.000748746,0.095098205,0.00029989914],"about_ca_topic_score_codex":0.0046378598,"about_ca_topic_score_gemma":0.0017230678,"teacher_disagreement_score":0.80547994,"about_ca_system_score_codex":0.00001946309,"about_ca_system_score_gemma":0.000059449296,"threshold_uncertainty_score":0.8683281},"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":"codex-gemma-dda1882f352a","genre_candidate":"protocol","genre_consensus":"protocol","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08104081,2.8040498e-8,0.0000012394945,0.00011524346,0.000056140903,0.78819686,8.5937126e-7,0.000052950993,0.13053589],"genre_scores_gemma":[0.26228312,6.2667276e-9,0.000041468764,0.00044272045,0.000049870476,0.7342386,0.000004111238,0.000002843712,0.0029372],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.9983059,0.00035418413,0.00034429898,0.00047289085,0.00014428531,0.00037839936],"domain_scores_gemma":[0.9992105,0.00007586099,0.000102837825,0.00041069955,0.00006495283,0.00013520014],"candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.00040306753,0.00018962486,0.0002902583,0.00006716826,0.00028489195,0.00060442457,0.00195359,0.00005901239,0.21869712],"category_scores_gemma":[0.000026852926,0.000119925084,0.000026556563,0.00024383493,0.00005542903,0.0016388046,0.0005881766,0.00017630348,0.006600087],"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":[0.00014029538,0.0010438608,0.9770841,0.000010872489,0.000036641126,0.000014350258,0.00013581404,0.000049115548,0.0000017673793,0.000010710802,0.01428962,0.0071828454],"study_design_scores_gemma":[0.0026316864,0.002801153,0.9871075,0.0000012720819,0.000005688319,0.0000030319318,0.0009070275,0.00070316374,0.0000145752,0.0015964372,0.0040414934,0.00018695576],"about_ca_topic_score_codex":0.04359805,"about_ca_topic_score_gemma":0.05158401,"teacher_disagreement_score":0.21209703,"about_ca_system_score_codex":0.000002885699,"about_ca_system_score_gemma":0.00008701442,"threshold_uncertainty_score":0.9941734},"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00019434298,0.0016699627,0.00000882744,0.000012128497,0.0001919438,0.0028887342,0.9902888,0.00005630105,0.004688936],"genre_scores_gemma":[0.00060639903,0.000012657245,0.007323301,0.00023065486,0.0012360599,0.0006275894,0.98336667,0.00047232446,0.006124336],"study_design_codex":"not_applicable","study_design_gemma":"observational","domain_scores_codex":[0.9888334,0.00076674577,0.002265737,0.0056720716,0.0014702231,0.0009917727],"domain_scores_gemma":[0.98369277,0.0053692493,0.0027893663,0.0068794442,0.0007544805,0.0005146827],"candidate_categories":["metaresearch","metaepi_narrow","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["metaepi_narrow","open_science","insufficient_payload"],"category_scores_codex":[0.000735399,0.0013398053,0.0023408171,0.00085971033,0.00049559644,0.0030326773,0.016480735,0.0012442436,0.9985393],"category_scores_gemma":[0.009087799,0.0015152405,0.00017137932,0.0024243486,0.00020046294,0.0032958644,0.049432606,0.0010509089,0.0013245895],"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":[0.00056931097,0.0005304967,0.0019637465,0.0020825919,0.001091728,0.00003234844,0.00005749683,0.000103307735,0.000495132,1.16213236e-7,0.9881151,0.0049586375],"study_design_scores_gemma":[0.0018545549,0.00007657441,0.44474182,0.02359273,0.00018731141,0.000009178674,0.000118854754,0.08383757,0.000061274215,0.00003734342,0.44406724,0.0014155501],"about_ca_topic_score_codex":0.016269043,"about_ca_topic_score_gemma":0.032417394,"teacher_disagreement_score":0.99721473,"about_ca_system_score_codex":0.0009083053,"about_ca_system_score_gemma":0.007071874,"threshold_uncertainty_score":0.9999353},"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":"dataset","genre_gemma":"dataset","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0001330039,0.00006367231,0.00038072446,0.0005274795,0.00034515213,0.15142831,0.8324108,0.00035836917,0.0143524865],"genre_scores_gemma":[0.0022452045,0.00021076945,0.00058103766,0.00041893264,0.0005950513,0.00021100295,0.9909676,0.003509423,0.0012610044],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99025697,0.0016492999,0.0012270017,0.0026612491,0.0021417143,0.0020637803],"domain_scores_gemma":[0.99131495,0.00020557707,0.0010919747,0.002955636,0.0034520407,0.0009798304],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["sts","open_science","insufficient_payload"],"category_scores_codex":[0.0066383467,0.0010328894,0.00085045514,0.0007584697,0.028936332,0.01988882,0.016958432,0.000330587,0.0053158724],"category_scores_gemma":[0.0053879186,0.00089211325,0.00017059268,0.0034826472,0.004204145,0.0015690188,0.041171454,0.0011543995,0.0020768265],"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":[0.0032423383,0.00043192704,0.000013231148,0.00084591255,0.0003437147,0.000041086896,0.00030621336,0.000051139687,0.00011961919,0.010795798,0.96384764,0.019961353],"study_design_scores_gemma":[0.006061135,0.002072316,0.0009424569,0.00058099406,0.0002600631,0.00015583124,0.0007754229,0.0005362491,0.000047400696,0.0014240607,0.9862356,0.00090845674],"about_ca_topic_score_codex":0.00026487844,"about_ca_topic_score_gemma":0.000043590917,"teacher_disagreement_score":0.15855676,"about_ca_system_score_codex":0.0038884257,"about_ca_system_score_gemma":0.0005812056,"threshold_uncertainty_score":0.99993545},"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.00041972834,0.0019982788,0.00018382011,0.0005680687,0.0113417255,0.0141895255,0.95346117,0.0009181964,0.016919473],"genre_scores_gemma":[0.0016940014,0.0012475397,0.003622804,0.00076547195,0.00480818,0.0024298574,0.94949263,0.0012824017,0.034657132],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9667905,0.00077284337,0.007630729,0.014653977,0.0034019367,0.0067500146],"domain_scores_gemma":[0.95339763,0.0015678273,0.0064751934,0.034570996,0.00173813,0.0022501983],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"category_scores_codex":[0.00930402,0.0060755447,0.0073825,0.0034308247,0.0035380041,0.006796111,0.077921316,0.0042349687,0.01422063],"category_scores_gemma":[0.00930177,0.0072652427,0.0016586117,0.0036288449,0.0009215123,0.009380021,0.19747843,0.007965903,0.0143552385],"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":[0.0021439327,0.0018265663,0.0060824705,0.0046285163,0.004313561,0.0010433042,0.00018016118,0.000031715444,0.00018675168,0.00042176072,0.9762779,0.0028633615],"study_design_scores_gemma":[0.007891913,0.00075415435,0.0012517862,0.008300977,0.005217747,0.00019864916,0.0008121189,0.0017006764,0.00024297286,0.00035100672,0.9663706,0.006907399],"about_ca_topic_score_codex":0.011894269,"about_ca_topic_score_gemma":0.022009207,"teacher_disagreement_score":0.119557105,"about_ca_system_score_codex":0.0024097862,"about_ca_system_score_gemma":0.006541464,"threshold_uncertainty_score":0.99909055},"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":"codex-gemma-dda1882f352a","genre_candidate":"dataset","genre_consensus":"dataset","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038818344,0.00024599643,0.00016054756,0.00068978005,0.010106829,0.0132777095,0.96423274,0.0006065751,0.006797965],"genre_scores_gemma":[0.0016620388,0.00020898598,0.0031041808,0.0012034386,0.0020975303,0.0014292831,0.9485915,0.0006734727,0.04102958],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.97453505,0.00056697463,0.00627429,0.010227336,0.0025741819,0.0058221533],"domain_scores_gemma":[0.9715718,0.0010609825,0.0053051705,0.018885514,0.0014718467,0.0017047083],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","open_science","research_integrity","insufficient_payload"],"category_scores_codex":[0.0052168225,0.004372567,0.0050685783,0.0018474093,0.0026485512,0.004903985,0.03507089,0.0032303066,0.019613707],"category_scores_gemma":[0.0043432913,0.0054921564,0.0010901663,0.0020129385,0.00067485357,0.008669187,0.07544877,0.006282345,0.0037524663],"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":[0.002315731,0.0013083352,0.16268232,0.0020852766,0.0025413225,0.00037097096,0.0005133006,0.000031560157,0.000048257712,0.00026299615,0.8272608,0.0005791459],"study_design_scores_gemma":[0.0075439652,0.00051816605,0.051820185,0.0052285613,0.0038058774,0.00008945603,0.00040705767,0.0012956985,0.000090313464,0.00015962769,0.9242722,0.0047688913],"about_ca_topic_score_codex":0.2755601,"about_ca_topic_score_gemma":0.98326516,"teacher_disagreement_score":0.707705,"about_ca_system_score_codex":0.004686724,"about_ca_system_score_gemma":0.015602698,"threshold_uncertainty_score":0.9991341},"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":"empirical","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.73307145,0.019325905,0.00021265716,0.24621904,0.00008442899,0.00038837976,0.00018146046,0.000013639269,0.00050304376],"genre_scores_gemma":[0.9788486,0.0135449255,0.001199336,0.004592778,0.000049872855,0.000012536787,0.0013744402,0.000014335864,0.0003632209],"study_design_codex":"observational","study_design_gemma":"observational","domain_scores_codex":[0.99143815,0.0040997937,0.0020929875,0.0006576496,0.0013472799,0.00036415138],"domain_scores_gemma":[0.99019104,0.004466156,0.0019168576,0.0023953717,0.0007034099,0.00032718354],"candidate_categories":["scholarly_communication","open_science","research_integrity"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.013386956,0.00030083043,0.0007205309,0.0021338088,0.0010534901,0.09676865,0.012229824,0.00035675854,0.000018050609],"category_scores_gemma":[0.004771802,0.00020678085,0.00006269177,0.008777972,0.00033280012,0.1169445,0.0030409417,0.003124426,1.568237e-7],"study_design_candidate":"observational","study_design_consensus":"observational","about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00008403161,0.0001435803,0.91500306,0.000049451715,0.00047626553,0.0000025791835,0.008148943,0.000024395056,0.0000084547655,0.0026563832,0.04145041,0.031952444],"study_design_scores_gemma":[0.0012048064,0.00007049566,0.7867856,0.0005107943,0.00080591626,0.000028718734,0.046512395,0.02351917,0.000008184042,0.010336593,0.12990202,0.00031532993],"about_ca_topic_score_codex":0.14242809,"about_ca_topic_score_gemma":0.107416555,"teacher_disagreement_score":0.2457771,"about_ca_system_score_codex":0.00006352636,"about_ca_system_score_gemma":0.0020022104,"threshold_uncertainty_score":0.9991754},"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":"other","genre_gemma":"other","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":"other","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.040476523,0.00028851678,0.000109634464,0.0009375367,0.0014123716,0.006570579,0.0053819856,0.00028286385,0.94453996],"genre_scores_gemma":[0.40988874,0.0008825782,0.00039779552,0.000043598495,0.00045505827,0.00036373752,0.0018713165,0.0013909617,0.58470625],"study_design_codex":"observational","study_design_gemma":"not_applicable","domain_scores_codex":[0.9309423,0.031990044,0.0053466507,0.020644326,0.008014646,0.0030620517],"domain_scores_gemma":[0.86945856,0.0095422305,0.010537086,0.10535477,0.004202618,0.0009047248],"candidate_categories":["metaresearch","metaepi_narrow","bibliometrics","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaresearch","metaepi_narrow","open_science","research_integrity","insufficient_payload"],"category_scores_codex":[0.2310785,0.0020549388,0.0038989473,0.005895393,0.0009912498,0.0066290926,0.052390415,0.0024723497,0.736718],"category_scores_gemma":[0.14042953,0.0014837672,0.0020056618,0.026194332,0.0028959212,0.00394585,0.06558107,0.011078745,0.50798565],"study_design_candidate":"not_applicable","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0029903725,0.0061319787,0.46385026,0.008967838,0.079751134,0.0001816521,0.052598543,0.009943815,0.013619458,0.0070003006,0.349372,0.0055926265],"study_design_scores_gemma":[0.0036815924,0.000024934418,0.22032571,0.01937425,0.049005978,0.0001967998,0.015332482,0.17584202,0.0048754863,0.011725605,0.49425256,0.0053625912],"about_ca_topic_score_codex":0.008930883,"about_ca_topic_score_gemma":0.017505536,"teacher_disagreement_score":0.3694122,"about_ca_system_score_codex":0.0011340666,"about_ca_system_score_gemma":0.002738094,"threshold_uncertainty_score":0.9998176},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01923884,0.0023532282,0.18346038,0.07208792,0.0008791761,0.027719924,0.004352901,0.001063427,0.6888442],"genre_scores_gemma":[0.94566226,0.0060569225,0.04163416,0.0027537288,0.00013185393,0.000012113804,0.0008694563,0.00082662376,0.0020528596],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99630016,0.000557731,0.00030190882,0.0015474006,0.00057548843,0.00071730034],"domain_scores_gemma":[0.9963894,0.0001167273,0.00016620304,0.0011998732,0.0014018145,0.00072601484],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.010817446,0.00020208224,0.00024599218,0.0015174159,0.004315253,0.038211245,0.020617235,0.00012042734,0.00008030182],"category_scores_gemma":[0.02600012,0.00023051637,0.000017839953,0.0017107546,0.0005578555,0.026199372,0.079467766,0.0011685466,0.000037789563],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000044249442,0.000023075867,0.000036434212,0.00016476019,0.000032611206,0.000022278202,0.0021864101,0.00012307467,0.00019748078,0.9563168,0.0050287484,0.03582403],"study_design_scores_gemma":[0.0010702795,0.00016774469,0.0018861465,0.0005176804,0.00001651375,0.000034765973,0.002866671,0.0069228285,0.00009580771,0.029000401,0.9569517,0.0004694563],"about_ca_topic_score_codex":0.016339164,"about_ca_topic_score_gemma":0.0013748074,"teacher_disagreement_score":0.95192295,"about_ca_system_score_codex":0.00197237,"about_ca_system_score_gemma":0.0010158432,"threshold_uncertainty_score":0.99698097},"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":"other","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.01923884,0.0023532282,0.18346038,0.07208792,0.0008791761,0.027719924,0.004352901,0.001063427,0.6888442],"genre_scores_gemma":[0.94566226,0.0060569225,0.04163416,0.0027537288,0.00013185393,0.000012113804,0.0008694563,0.00082662376,0.0020528596],"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","domain_scores_codex":[0.99630016,0.000557731,0.00030190882,0.0015474006,0.00057548843,0.00071730034],"domain_scores_gemma":[0.9963894,0.0001167273,0.00016620304,0.0011998732,0.0014018145,0.00072601484],"candidate_categories":["metaresearch","sts","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.010817446,0.00020208224,0.00024599218,0.0015174159,0.004315253,0.038211245,0.020617235,0.00012042734,0.00008030182],"category_scores_gemma":[0.02600012,0.00023051637,0.000017839953,0.0017107546,0.0005578555,0.026199372,0.079467766,0.0011685466,0.000037789563],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.000044249442,0.000023075867,0.000036434212,0.00016476019,0.000032611206,0.000022278202,0.0021864101,0.00012307467,0.00019748078,0.9563168,0.0050287484,0.03582403],"study_design_scores_gemma":[0.0010702795,0.00016774469,0.0018861465,0.0005176804,0.00001651375,0.000034765973,0.002866671,0.0069228285,0.00009580771,0.029000401,0.9569517,0.0004694563],"about_ca_topic_score_codex":0.016339164,"about_ca_topic_score_gemma":0.0013748074,"teacher_disagreement_score":0.95192295,"about_ca_system_score_codex":0.00197237,"about_ca_system_score_gemma":0.0010158432,"threshold_uncertainty_score":0.99698097},"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":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":"empirical","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.76416105,0.0015822063,0.00017037506,0.007486605,0.0018539369,0.0030445196,0.00013019665,0.00005727951,0.22151384],"genre_scores_gemma":[0.9388627,0.0019028779,0.000041859344,0.000055006174,0.000090958216,0.0000037784253,0.000032206422,0.000012530059,0.05899809],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.98538876,0.0030823601,0.001218833,0.0010416836,0.008180087,0.0010882912],"domain_scores_gemma":[0.9778947,0.010479948,0.0009405225,0.0018155276,0.008032716,0.0008365524],"candidate_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","bibliometrics","sts","scholarly_communication","open_science"],"category_scores_codex":[0.034248292,0.00032120195,0.0005587932,0.048108373,0.0025172473,0.07960398,0.013145881,0.00050400133,0.0002730589],"category_scores_gemma":[0.03221793,0.00023561661,0.00026169283,0.12425376,0.0028032702,0.07051289,0.012420703,0.001895324,0.000094265975],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0145887425,0.0006695271,0.1319952,0.00048407508,0.0007087655,0.006830854,0.026811624,0.00027788608,0.00041116937,0.18114449,0.5284851,0.10759255],"study_design_scores_gemma":[0.0034131354,0.00027004443,0.08361412,0.0002660428,0.000059751867,0.00004910272,0.044151172,0.0077712033,0.00031162793,0.0038128803,0.8558956,0.0003853631],"about_ca_topic_score_codex":0.07483634,"about_ca_topic_score_gemma":0.001994842,"teacher_disagreement_score":0.32741043,"about_ca_system_score_codex":0.0015331991,"about_ca_system_score_gemma":0.004045351,"threshold_uncertainty_score":0.99991053},"labels":[{"model":"gemma","categories":["metaresearch","bibliometrics","scholarly_communication"],"domain":"evaluation","study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["bibliometrics","scholarly_communication","open_science"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"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":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"codex-gemma-dda1882f352a","genre_candidate":"other","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0013548417,0.000839863,0.0011345736,0.000641165,0.0020004257,0.0061161174,0.004850254,0.0021718056,0.9808909],"genre_scores_gemma":[0.51364625,0.0063354815,0.0027636015,0.0012557551,0.0028058023,0.000001218912,0.055424515,0.073106185,0.34466118],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.9821726,0.0023032979,0.0025402238,0.0047366302,0.005191529,0.0030556952],"domain_scores_gemma":[0.97246486,0.00013710796,0.0020208098,0.005444778,0.018497335,0.0014351067],"candidate_categories":["metaresearch","metaepi_narrow","sts","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["metaepi_narrow","sts","open_science","insufficient_payload"],"category_scores_codex":[0.017370513,0.0014738293,0.001820907,0.0054236157,0.026541106,0.03651002,0.04145694,0.0007781123,0.110561475],"category_scores_gemma":[0.023214111,0.0017355075,0.00038284474,0.014149082,0.006767396,0.0030769345,0.090634905,0.0032955161,0.12062803],"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":[0.0004088453,0.00046471937,0.0000089455425,0.00067466544,0.00053631276,0.000102436796,0.00081419,0.00007933165,0.009974756,0.011354273,0.7322687,0.2433128],"study_design_scores_gemma":[0.0024502086,0.00052284135,0.0017942651,0.0014642468,0.0003681369,0.00074512133,0.0008159984,0.00040955792,0.00088537036,0.000991009,0.98811674,0.0014364802],"about_ca_topic_score_codex":0.0005019638,"about_ca_topic_score_gemma":0.0000036995077,"teacher_disagreement_score":0.63622975,"about_ca_system_score_codex":0.008444999,"about_ca_system_score_gemma":0.0011816178,"threshold_uncertainty_score":0.9998011},"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":"codex-gemma-dda1882f352a","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.48958904,0.00015250954,0.0013314444,0.0032933305,0.010011635,0.0047430573,0.0001394604,0.00016177137,0.49057776],"genre_scores_gemma":[0.87904996,0.00017063737,0.00026991483,0.00006028134,0.0021322498,0.0000062519052,0.000026003348,0.000031111857,0.11825357],"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","domain_scores_codex":[0.99743235,0.0004235484,0.00037600673,0.0007367153,0.00023515103,0.00079624826],"domain_scores_gemma":[0.9964324,0.0012440135,0.00029675156,0.0010545223,0.0006320913,0.00034018708],"candidate_categories":["metaepi_narrow","sts","scholarly_communication","open_science","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0023809166,0.0002499051,0.0005195063,0.0007048679,0.0037741032,0.0023459136,0.008179463,0.0002770416,0.0011381601],"category_scores_gemma":[0.0006605119,0.00026755742,0.00029683072,0.0016790645,0.0005811207,0.005108049,0.0052940403,0.0007427617,0.00023629155],"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":[0.0066250963,0.001639264,0.07036801,0.0001947518,0.001712847,0.00029818268,0.007960869,0.00009255671,0.0012493236,0.4268744,0.473698,0.009286725],"study_design_scores_gemma":[0.0045415,0.00025584872,0.030812746,0.000028866514,0.00004559903,0.0000028998345,0.006000195,0.00037267563,0.0004975029,0.018436687,0.9383609,0.000644576],"about_ca_topic_score_codex":0.028872266,"about_ca_topic_score_gemma":0.00089740683,"teacher_disagreement_score":0.4646629,"about_ca_system_score_codex":0.0013234193,"about_ca_system_score_gemma":0.0006706789,"threshold_uncertainty_score":0.99997765},"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"}]}