{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":2427,"total_is_capped":false,"direct_labels_cover":37,"predictions_cover":2427,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"query_hash":"0ddb50d104d2","filters":{"topic":"Research Data Management Practices"}},"results":[{"id":"W3048186973","doi":"10.5334/dsj-2020-043","title":"The CARE Principles for Indigenous Data Governance","year":2020,"lang":"en","type":"article","venue":"Data Science Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":1206,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Laurentian University","funders":"Rural Development Administration; European Commission","keywords":"Indigenous; Sovereignty; Data governance; Data sharing; Corporate governance; Stewardship (theology); Political science; Public administration; Public relations; Sociology; Law; Business; Medicine; Data quality","retraction":null,"screen_n_in":null,"score":{"opus":0.3437787918939087,"gpt":0.4254753768765994,"spread":0.08169658498269072,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.0068816,0.0001074647,0.0001003795,0.00004539864,0.002599995,0.01663155,0.0692704,0.00001615007,0.000003322821],"category_scores_gemma":[0.00756462,0.00006919435,0.00001808042,0.001027167,0.0003855879,0.1058463,0.02386449,0.0003618239,0.00002741393],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008139827,"about_ca_system_score_gemma":0.001392241,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002073636,"about_ca_topic_score_gemma":0.00005825762,"domain_scores_codex":[0.9963055,0.00007221621,0.0003108673,0.0009715131,0.001663143,0.0006767356],"domain_scores_gemma":[0.9930814,0.0003578861,0.0004000109,0.005540348,0.0002698512,0.0003504913],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00007383252,0.00008483884,0.002294803,0.0000964118,0.00007883446,0.000169792,0.004045895,0.0002448619,0.0008598361,0.2047193,0.09816121,0.6891704],"study_design_scores_gemma":[0.0001636146,0.00008467791,0.002148876,0.00001009439,0.000006029367,0.00003913936,0.0004993922,0.07463297,0.00008463098,0.0001062778,0.9221182,0.0001060448],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.001506938,0.003043496,0.9333755,0.05633418,0.001130232,0.0008137605,0.001950406,0.00008380703,0.0017617],"genre_scores_gemma":[0.1881604,0.03180178,0.7688311,0.006244382,0.003110113,0.00004126283,0.001062811,0.00005429417,0.0006938366],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.823957,"threshold_uncertainty_score":0.9986985,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2037620269","doi":"10.1016/j.cub.2013.11.014","title":"The Availability of Research Data Declines Rapidly with Article Age","year":2013,"lang":"en","type":"article","venue":"Current Biology","topic":"Research Data Management Practices","field":"Computer Science","cited_by":504,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université Laval; Kwantlen Polytechnic University; Women's Health Research Institute; University of British Columbia","funders":"","keywords":"Biology","retraction":null,"screen_n_in":null,"score":{"opus":0.3991685083355045,"gpt":0.495087640138035,"spread":0.09591913180253048,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.004866557,0.0000775462,0.0001160817,0.00008770656,0.000212306,0.000640589,0.006174799,0.0000277665,0.0000297153],"category_scores_gemma":[0.002533594,0.00004102506,0.00001473158,0.0006155953,0.000880581,0.00567347,0.00445513,0.0003032841,0.0001875235],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001504777,"about_ca_system_score_gemma":0.00009279473,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002234836,"about_ca_topic_score_gemma":0.0001381227,"domain_scores_codex":[0.997444,0.0009240634,0.0002554556,0.0005677041,0.0003393029,0.0004694223],"domain_scores_gemma":[0.9938649,0.002041108,0.0001056285,0.003571583,0.0003342467,0.00008251649],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00005419532,0.0004358298,0.06577864,0.0001011237,0.00007515687,0.000005907731,0.0001298814,0.000002432283,0.003870557,0.3075992,0.03097624,0.5909708],"study_design_scores_gemma":[0.0005845145,0.0005547815,0.07932405,0.00003710679,0.000007856302,0.000004206756,0.0001754005,0.01366383,0.001601986,0.02038767,0.8834512,0.0002074517],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7573394,0.007991724,0.188925,0.03634923,0.00133904,0.002975621,0.00007237327,0.0002284826,0.004779148],"genre_scores_gemma":[0.9899062,0.002471835,0.006878156,0.00003953788,0.0001106327,0.00009901939,0.00006011498,0.000006819466,0.0004276833],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8524749,"threshold_uncertainty_score":0.9992023,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2988998130","doi":"10.1162/dint_r_00024","title":"FAIR Principles: Interpretations and Implementation Considerations","year":2019,"lang":"en","type":"article","venue":"Data Intelligence","topic":"Research Data Management Practices","field":"Computer Science","cited_by":459,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Biotechnology and Biological Sciences Research Council; European Regional Development Fund; Horizon 2020 Framework Programme; Innovative Medicines Initiative; National Institutes of Health; National Science Foundation; Ministerio de Economía y Competitividad; Common Fund; Nederlandse Organisatie voor Wetenschappelijk Onderzoek; European Commission; Universidad Politécnica de Madrid","keywords":"Implementation; Interoperability; Computer science; Reusability; USable; Fair use; Reuse; Stakeholder; Convergence (economics); Risk analysis (engineering); Data science; Management science; Software engineering; World Wide Web; Business; Engineering; Political science; Public relations; Software; Law; Economics","retraction":null,"screen_n_in":null,"score":{"opus":0.2135019208723,"gpt":0.4376810240360997,"spread":0.2241791031637997,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.0007842351,0.00008590251,0.00007872999,0.0001283835,0.00011522,0.002064488,0.001949318,0.00001759932,0.0002291715],"category_scores_gemma":[0.0004908955,0.00008303223,0.000009426699,0.0002384644,0.00004974555,0.030371,0.002955171,0.0001136881,0.000360954],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002022336,"about_ca_system_score_gemma":0.00007753687,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000130014,"about_ca_topic_score_gemma":0.0002163334,"domain_scores_codex":[0.9986019,0.0000916307,0.0002488604,0.0005908955,0.0002738547,0.0001928906],"domain_scores_gemma":[0.9972207,0.0004851563,0.00009904593,0.002053975,0.00007327079,0.00006787247],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000001215918,0.00001942568,0.002299073,0.00002128371,0.00002152211,0.000004526263,0.0003904269,0.00005609429,0.0001139016,0.9754397,0.001223473,0.02040932],"study_design_scores_gemma":[0.0002949834,0.0002883842,0.01424445,0.00008083291,0.00003394667,0.00006811807,0.00421137,0.5950091,0.00350402,0.02948134,0.3520832,0.0007002397],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.003090768,0.00007946233,0.9898286,0.002583777,0.0001631397,0.0004797537,0.0001615506,0.00008211775,0.003530796],"genre_scores_gemma":[0.8926328,0.000568666,0.1052161,0.0005619911,0.00001901996,0.00003034075,0.0005144855,0.000006780319,0.0004498043],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9459584,"threshold_uncertainty_score":0.9989715,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1572136682","doi":"10.1038/493159a","title":"Value all research products","year":2013,"lang":"en","type":"article","venue":"Nature","topic":"Research Data Management Practices","field":"Computer Science","cited_by":411,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"OpenAlex","funders":"","keywords":"Altmetrics; Value (mathematics); Data science; Computer science","retraction":null,"screen_n_in":null,"score":{"opus":0.1441714568017237,"gpt":0.442765726256411,"spread":0.2985942694546873,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","research_integrity","insufficient_payload"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.002563203,0.00008005369,0.00007070482,0.0002442825,0.0001327798,0.003138338,0.003930037,0.0002976646,0.00005070258],"category_scores_gemma":[0.002669555,0.00006313909,0.00001736301,0.001136765,0.00005606137,0.02431602,0.001865878,0.003408685,0.001511601],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005133771,"about_ca_system_score_gemma":0.00008394776,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001421421,"about_ca_topic_score_gemma":0.000004144499,"domain_scores_codex":[0.9971055,0.0003679337,0.00009981602,0.0005774269,0.001320754,0.0005286118],"domain_scores_gemma":[0.9973867,0.0002845262,0.00003454765,0.001669991,0.000512174,0.0001120926],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000001600011,0.00004863191,0.00006822976,0.00002704703,0.00001624538,0.00002229755,0.00007627867,0.000002869209,0.001780837,0.5317851,0.4568315,0.00933935],"study_design_scores_gemma":[0.0001015134,0.00005273363,0.005678548,0.000009405438,0.000001180699,0.000004847726,0.00002290387,0.001796437,0.002202383,0.01026785,0.9797618,0.0001004149],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.02692795,0.004093471,0.01621636,0.6678234,0.001735226,0.003323792,0.000006315247,0.0007213066,0.2791522],"genre_scores_gemma":[0.8215324,0.0007491315,0.1159978,0.005890904,0.000653542,0.0002118958,0.00002636268,0.00002843529,0.0549095],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7946045,"threshold_uncertainty_score":0.9992658,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3024722698","doi":"10.1038/s41597-020-0486-7","title":"The TRUST Principles for digital repositories","year":2020,"lang":"en","type":"article","venue":"Scientific Data","topic":"Research Data Management Practices","field":"Computer Science","cited_by":390,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Ocean Networks Canada Society; University of Victoria","funders":"National Institute of Biomedical Imaging and Bioengineering; National Institute of Diabetes and Digestive and Kidney Diseases; Economic and Social Research Council","keywords":"World Wide Web; Computer science; Metadata; Scholarly communication; Data science; Internet privacy; Knowledge management; Publishing; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.2872569636919766,"gpt":0.3757555810503705,"spread":0.08849861735839387,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.002147771,0.00007645638,0.00006558477,0.00003154868,0.0009497717,0.03837242,0.01526573,0.00001304801,0.000001967453],"category_scores_gemma":[0.006170863,0.00005060746,0.00002346914,0.0006003391,0.0002592145,0.04000018,0.01025894,0.00008066493,0.00007849642],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000011219,"about_ca_system_score_gemma":0.0001221455,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000003639181,"about_ca_topic_score_gemma":0.000009160065,"domain_scores_codex":[0.9978414,0.00003923405,0.0002062888,0.0009667238,0.0006246713,0.0003216393],"domain_scores_gemma":[0.9943673,0.0004363721,0.0001152713,0.004846899,0.0001048473,0.0001293331],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0000182429,0.00002992879,0.0001968107,0.00003707585,0.0000348169,0.00001005414,0.0002488848,0.000007382684,0.0001887925,0.4258383,0.5179107,0.05547902],"study_design_scores_gemma":[0.00007565916,0.00002441888,0.0001751642,0.000003098945,0.000003187548,0.000001173154,0.00007910952,0.04619589,0.0002360177,0.0005813912,0.9525578,0.00006707858],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0005563246,0.0004394346,0.9210995,0.06060238,0.003790907,0.0009658859,0.002989911,0.0003062158,0.009249465],"genre_scores_gemma":[0.3362477,0.000683527,0.3351711,0.002322791,0.004216848,0.0004623708,0.02062322,0.000130461,0.300142],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.5859284,"threshold_uncertainty_score":0.9977459,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["metaresearch","open_science"],"domain":"reproducibility","study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"},{"model":"gpt","categories":["scholarly_communication","open_science"],"domain":null,"study_design":"theoretical_or_conceptual","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"high"}],"label_agreement":"split"},{"id":"W2170057885","doi":"10.1016/j.tree.2010.11.006","title":"Data archiving in ecology and evolution: best practices","year":2010,"lang":"en","type":"article","venue":"Trends in Ecology & Evolution","topic":"Research Data Management Practices","field":"Computer Science","cited_by":291,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"CLARITY; Data science; Computer science; Citation; Value (mathematics); World Wide Web; Biology","retraction":null,"screen_n_in":null,"score":{"opus":0.1353790420573214,"gpt":0.4104480124576015,"spread":0.2750689704002801,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.003411046,0.0001633595,0.0002299295,0.001290776,0.00016404,0.0003574563,0.002818979,0.000172455,0.00004313457],"category_scores_gemma":[0.003148348,0.0001745249,0.00001647026,0.001066634,0.0002805951,0.02115717,0.003411037,0.0009549345,0.0000587455],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001975675,"about_ca_system_score_gemma":0.0001687524,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001192658,"about_ca_topic_score_gemma":0.1612739,"domain_scores_codex":[0.997036,0.0006344485,0.0004004163,0.001039072,0.0002498218,0.0006402396],"domain_scores_gemma":[0.9965903,0.001111626,0.0004245694,0.001737083,0.00004201722,0.00009443468],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.00002493881,0.0003549076,0.8702791,0.00001527902,0.0000174134,0.00006715149,0.00009194045,0.00006509096,0.0003993481,0.1109563,0.0007772064,0.01695125],"study_design_scores_gemma":[0.0005342217,0.0001514115,0.9219719,0.000007679913,0.00001013484,0.00006470294,0.00007659324,0.06532863,0.000002686935,0.003009355,0.008694976,0.0001477051],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9637946,0.0001856843,0.01119129,0.01313011,0.001225383,0.0002824604,0.00002716512,0.0001009273,0.01006238],"genre_scores_gemma":[0.9813858,0.0001801348,0.01725057,0.00006339834,0.0001013357,0.00005017704,0.00007026581,0.000008384193,0.0008899245],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1600813,"threshold_uncertainty_score":0.9925334,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2954081892","doi":"10.3389/fmars.2019.00440","title":"Ocean FAIR Data Services","year":2019,"lang":"en","type":"article","venue":"Frontiers in Marine Science","topic":"Research Data Management Practices","field":"Computer Science","cited_by":244,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Fisheries and Oceans Canada","funders":"Natural Environment Research Council; Instituto Nacional de Ciência e Tecnologia - Oceanografia Integrada e Usos Múltiplos da Plataforma Continental e Oceano Adjacente - Centro de Oceanografia Integrada; LifeWatch – Niclas Öberg Foundation; Executive Agency for Small and Medium-sized Enterprises; Conselho Nacional de Desenvolvimento Científico e Tecnológico; National Oceanic and Atmospheric Administration; Sight Research UK; Horizon 2020 Framework Programme; Joint Institute for the Study of the Atmosphere and Ocean; California Institute of Technology","keywords":"Interoperability; Metadata; Computer science; Workflow; Data quality; Data management; Standardization; Data discovery; Data access; Data science; Data curation; World Wide Web; Database; Service (business); Business","retraction":null,"screen_n_in":null,"score":{"opus":0.03631867215974199,"gpt":0.3169767745783257,"spread":0.2806581024185837,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.003912827,0.0001217956,0.0001488438,0.0006055309,0.0001233551,0.002651711,0.02099748,0.00002247937,0.00002684918],"category_scores_gemma":[0.0002346862,0.000110942,0.00001371073,0.002654657,0.000230397,0.07106463,0.0236014,0.0002159959,0.0001270086],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008705237,"about_ca_system_score_gemma":0.0001907515,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000157058,"about_ca_topic_score_gemma":0.00001519032,"domain_scores_codex":[0.996753,0.00007426112,0.0002086915,0.001239914,0.001115707,0.0006083699],"domain_scores_gemma":[0.9953685,0.00005563603,0.0001047409,0.004272366,0.00006768979,0.0001310902],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000009928756,0.00007403827,0.8976062,0.00008073607,0.000008732689,0.00003654797,0.0001367275,0.0001653935,0.0001660871,0.03732006,0.007560811,0.05683477],"study_design_scores_gemma":[0.0006609852,0.0001088769,0.2850074,0.00004642718,0.000004315072,0.000008484531,0.0005983739,0.5191122,0.0002735995,0.01007434,0.1836019,0.0005030762],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.3383431,0.0001513447,0.4617846,0.009387872,0.008753503,0.001795352,0.00003147504,0.0005402844,0.1792125],"genre_scores_gemma":[0.3253089,0.0002320566,0.6687903,0.0005352827,0.00004482439,0.000002171982,0.00003561583,0.000009279312,0.005041604],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.6125988,"threshold_uncertainty_score":0.9983836,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2544237122","doi":"10.1002/asi.23781","title":"Developments in research data management in academic libraries: Towards an understanding of research data service maturity","year":2017,"lang":"en","type":"article","venue":"Journal of the Association for Information Science and Technology","topic":"Research Data Management Practices","field":"Computer Science","cited_by":228,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"RDM; Maturity (psychological); Data management; Data curation; Service (business); Capability Maturity Model; Data management plan; Service delivery framework; Data as a service; Knowledge management; Public relations; Business; Computer science; Data science; Political science; Marketing; Database","retraction":null,"screen_n_in":null,"score":{"opus":0.6200252280254551,"gpt":0.5405688026875997,"spread":0.07945642533785546,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["metaresearch","scholarly_communication","open_science"],"category_scores_codex":[0.06803358,0.00005300273,0.0001324746,0.002812137,0.000814478,0.002376301,0.02052973,0.0001086709,2.900394e-7],"category_scores_gemma":[0.01730596,0.00004062695,0.000006319377,0.003838976,0.0003863393,0.1247002,0.01511551,0.0008249169,0.000001421708],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006760397,"about_ca_system_score_gemma":0.0007814651,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00007911096,"about_ca_topic_score_gemma":0.0001519965,"domain_scores_codex":[0.9960632,0.0001923033,0.0006022178,0.0002382734,0.002497933,0.0004061253],"domain_scores_gemma":[0.9951118,0.0003608016,0.001065851,0.002167942,0.00124575,0.00004790666],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.00003845915,0.00005308373,0.04153537,0.0001152018,0.00002687567,0.000002051529,0.001385444,0.00001116872,0.00008882827,0.9248568,0.001492347,0.03039438],"study_design_scores_gemma":[0.003480372,0.0002787584,0.4660254,0.0006785833,0.0000155625,0.00002047478,0.02749799,0.07241851,0.001763561,0.2980595,0.1294511,0.0003102602],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.4511512,0.0003097227,0.0445619,0.4518813,0.002121032,0.005528682,0.0002743323,0.0001025493,0.04406925],"genre_scores_gemma":[0.9881899,0.0007942202,0.01086177,0.00007756848,0.00001208838,0.000006873842,0.000008521286,0.000001991367,0.0000471233],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.6267973,"threshold_uncertainty_score":0.9986593,"prediction_status":"machine_predicted_unvalidated"},"labels":[{"model":"gemma","categories":["scholarly_communication"],"domain":null,"study_design":"qualitative","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"},{"model":"gpt","categories":["scholarly_communication"],"domain":null,"study_design":"observational","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low"}],"label_agreement":"split"},{"id":"W2093454812","doi":"10.1096/fj.12-218164","title":"Mandated data archiving greatly improves access to research data","year":2013,"lang":"en","type":"article","venue":"The FASEB Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":197,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University; University of British Columbia","funders":"","keywords":"Computer science; Data science; Data access; Population; Data sharing; Odds; Internet privacy; World Wide Web; Database; Medicine","retraction":null,"screen_n_in":null,"score":{"opus":0.4719189794194362,"gpt":0.4834710550941572,"spread":0.011552075674721,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.01430277,0.0001452648,0.0001454937,0.0004580429,0.001055059,0.03095968,0.0629896,0.0000245122,0.0001673298],"category_scores_gemma":[0.003247243,0.00008997316,0.00001989322,0.001151475,0.0001450179,0.09109338,0.06043687,0.001230909,0.0006868774],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005022468,"about_ca_system_score_gemma":0.0002556156,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0009843635,"about_ca_topic_score_gemma":0.0001195069,"domain_scores_codex":[0.9950046,0.001204379,0.0003483303,0.0008235092,0.001710444,0.0009087411],"domain_scores_gemma":[0.9883264,0.001189417,0.0001655717,0.009594803,0.0003001849,0.0004236242],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00004402034,0.000143871,0.001104975,0.00003462874,0.0002166771,0.0002252397,0.0004065718,0.00004104054,0.03756095,0.006433701,0.6249687,0.3288196],"study_design_scores_gemma":[0.001508995,0.0006362975,0.0829215,0.0003711035,0.0000675813,0.0009755013,0.001159192,0.3973005,0.007940752,0.01597837,0.4900619,0.001078275],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1203863,0.0003368352,0.7218584,0.1408232,0.0009269147,0.002070168,0.0002696322,0.0001768762,0.01315165],"genre_scores_gemma":[0.9244533,0.002028449,0.06240194,0.002342712,0.001441101,0.00005670444,0.0004464456,0.00005796834,0.006771362],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.804067,"threshold_uncertainty_score":0.9700463,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2156555857","doi":"10.1890/es14-00402.1","title":"The Tao of open science for ecology","year":2015,"lang":"en","type":"article","venue":"Ecosphere","topic":"Research Data Management Practices","field":"Computer Science","cited_by":195,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"National Science Foundation","keywords":"Mindset; Open science; Transparency (behavior); Openness to experience; Process (computing); Stewardship (theology); Data sharing; Open data; Best practice; Ecology; Computer science; Sociology; Knowledge management; Engineering ethics; Political science; Psychology; Engineering; World Wide Web; Social psychology; Biology; Computer security; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.1825065985399526,"gpt":0.4333599160885168,"spread":0.2508533175485641,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.004488918,0.00004471864,0.00007354454,0.00001876539,0.0002394482,0.002328398,0.01201158,0.00001270008,0.00001322344],"category_scores_gemma":[0.002173902,0.00002977938,0.0000148237,0.0004420406,0.0002075581,0.01612967,0.004888957,0.00005137639,0.00007364243],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004294803,"about_ca_system_score_gemma":0.0004853843,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00008842778,"about_ca_topic_score_gemma":0.0002910131,"domain_scores_codex":[0.9989489,0.0000602015,0.0001275142,0.0002786771,0.0003085448,0.0002761855],"domain_scores_gemma":[0.9981043,0.0004176517,0.0001131379,0.001033127,0.0002404265,0.00009135023],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00001520232,0.00003482621,0.0002610149,0.000006423923,0.000009576031,0.000001716438,0.00006816498,0.00003663209,0.00006898031,0.7563051,0.2220152,0.02117715],"study_design_scores_gemma":[0.0004408188,0.0002671122,0.004090713,0.000004080419,0.000002512132,0.000001995708,0.0002284204,0.01315848,0.001349189,0.02039957,0.9599804,0.00007670889],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.02416452,0.0003524517,0.3094874,0.03109257,0.001469137,0.002319015,0.00001139247,0.00007474817,0.6310288],"genre_scores_gemma":[0.7156928,0.0002342928,0.2518769,0.0005695627,0.00009664307,0.0002836876,0.000003217899,0.00001393162,0.03122897],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7379652,"threshold_uncertainty_score":0.9987073,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2127765089","doi":"10.22230/cjc.2004v29n3a1455","title":"Supporting and Enhancing Scholarship in the Digital Age: The Role of Open-Access Institutional Repositories","year":2004,"lang":"en","type":"article","venue":"Canadian Journal of Communication","topic":"Research Data Management Practices","field":"Computer Science","cited_by":170,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":true,"about_ca":false},"ca_institutions":"University of Toronto","funders":"University of Toronto","keywords":"Scholarship; Audience measurement; Scholarly communication; Publishing; Institution; Institutional repository; Digital scholarship; Political science; Electronic publishing; Public relations; Public domain; Sociology; Library science; World Wide Web; Computer science; The Internet; Geography; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.09435247063864172,"gpt":0.4033541335506127,"spread":0.309001662911971,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.003059462,0.00003913923,0.00006733879,0.0001382704,0.000332721,0.008876058,0.007065397,0.00001428183,8.088753e-7],"category_scores_gemma":[0.001571414,0.00002599602,0.00001418978,0.0003563763,0.0001479676,0.0296228,0.0006652185,0.0002995719,2.999162e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008477306,"about_ca_system_score_gemma":0.0009495905,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006229576,"about_ca_topic_score_gemma":0.03036725,"domain_scores_codex":[0.9990911,0.0002102374,0.0002990106,0.00006270342,0.0002183442,0.0001185438],"domain_scores_gemma":[0.9985468,0.000245307,0.0003588585,0.0006536152,0.0001238424,0.00007154235],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.000008139121,0.00002896241,0.03829569,0.0000101532,0.00002380326,0.00006876868,0.004202294,0.0004577705,0.0002715772,0.9326265,0.00004034489,0.02396596],"study_design_scores_gemma":[0.001016142,0.0001796876,0.6341792,0.0006298079,0.00002285719,0.0006925607,0.00670855,0.0003172604,0.003554338,0.2052025,0.1472478,0.0002492084],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9174841,0.004689975,0.0174085,0.03842658,0.0001408495,0.0004782604,0.000006327607,0.000005028568,0.02136036],"genre_scores_gemma":[0.9978849,0.0001286772,0.001841797,0.0001144462,0.00001445535,0.000002776431,0.000002677919,0.000001633496,0.000008675263],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.727424,"threshold_uncertainty_score":0.9983069,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3003585144","doi":"10.1038/s41562-019-0813-1","title":"The Confidence Database","year":2020,"lang":"en","type":"article","venue":"Nature Human Behaviour","topic":"Research Data Management Practices","field":"Computer Science","cited_by":161,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Victoria","funders":"Economic and Social Research Council; National Institute of Mental Health; Medical Research Council; Agence Nationale de la Recherche; Wellcome Trust; U.S. Department of Health and Human Services","keywords":"Confidence interval; Computer science; Database; Information retrieval; Statistics; Mathematics","retraction":null,"screen_n_in":null,"score":{"opus":0.09731803816968139,"gpt":0.3883518749160733,"spread":0.2910338367463919,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0007510257,0.0001148035,0.00008410158,0.00003784791,0.0005526064,0.0030558,0.005129431,0.00007303913,0.00003220541],"category_scores_gemma":[0.0007827488,0.00008018586,0.00003938928,0.0003687427,0.00008184447,0.01080678,0.001730212,0.001069767,0.0001236723],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001927933,"about_ca_system_score_gemma":0.00004382708,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003145176,"about_ca_topic_score_gemma":0.00003801362,"domain_scores_codex":[0.9981347,0.0001463835,0.0001624514,0.000498781,0.0007362723,0.000321378],"domain_scores_gemma":[0.9980563,0.000247865,0.0001009096,0.00132265,0.00009719079,0.0001750849],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000004955691,0.00002395442,0.001249792,0.000007660717,0.000009700816,0.0001326037,0.00009745381,6.904744e-7,0.001410904,0.942706,0.05282101,0.001535267],"study_design_scores_gemma":[0.0005709387,0.0002004415,0.03057119,0.0000247579,0.00002975285,0.00001533711,0.000169153,0.001228889,0.003256502,0.001846231,0.9616526,0.000434158],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08931471,0.008250925,0.4719695,0.3674562,0.002667818,0.003376823,0.000207378,0.002459756,0.05429684],"genre_scores_gemma":[0.991517,0.0002022706,0.003133581,0.003169254,0.0001685074,0.00002229912,0.0000322914,0.00001066268,0.001744126],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9408598,"threshold_uncertainty_score":0.9979791,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4309721430","doi":"10.1098/rspb.2022.1113","title":"Why don't we share data and code? Perceived barriers and benefits to public archiving practices","year":2022,"lang":"en","type":"article","venue":"Proceedings of the Royal Society B Biological Sciences","topic":"Research Data Management Practices","field":"Computer Science","cited_by":156,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia; Carleton University","funders":"Biological and Environmental Research; Office of Science; National Center for Ecological Analysis and Synthesis; National Academy of Sciences; Svenska Forskningsrådet Formas; Kempestiftelserna; University of New South Wales","keywords":"Code (set theory); Incentive; Open science; Data science; Categorization; Data sharing; Reuse; Open data; Value (mathematics); Computer science; Public relations; Knowledge management; Internet privacy; Business; Political science; World Wide Web; Engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.1869473218736467,"gpt":0.3454665983605413,"spread":0.1585192764868946,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts","scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.004868363,0.0001436709,0.0001707943,0.00005153509,0.001919179,0.002646944,0.008378751,0.00003446007,0.00005014154],"category_scores_gemma":[0.004851971,0.00008556939,0.00004825991,0.0009940648,0.0007331417,0.0075316,0.02533646,0.0003218487,5.80619e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003127128,"about_ca_system_score_gemma":0.00006392669,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001144973,"about_ca_topic_score_gemma":0.00001075523,"domain_scores_codex":[0.9972908,0.00008908704,0.0002107289,0.001058218,0.0008764569,0.0004746841],"domain_scores_gemma":[0.99835,0.000551768,0.0004251962,0.0003355847,0.0000871826,0.0002502836],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00005332586,0.0002413876,0.7457227,0.0002437105,0.0001735107,0.000001710037,0.007212711,0.0001428922,0.005884869,0.1280052,0.0796271,0.03269082],"study_design_scores_gemma":[0.0006028043,0.002386974,0.284869,0.0001364932,0.00004871077,0.00003504691,0.04372605,0.2484856,0.0002267081,0.006246952,0.4121991,0.001036483],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8908747,0.0005820858,0.0004304852,0.1056999,0.0000929621,0.0005000254,0.0001729334,0.0000771277,0.001569834],"genre_scores_gemma":[0.9806125,0.0009845444,0.01586895,0.002307194,0.00003518501,0.00004527058,0.000003421879,0.000003829202,0.0001391772],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4608537,"threshold_uncertainty_score":0.9993802,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2963526428","doi":"10.1108/jd-12-2018-0211","title":"Maturing research data services and the transformation of academic libraries","year":2019,"lang":"en","type":"article","venue":"Journal of Documentation","topic":"Research Data Management Practices","field":"Computer Science","cited_by":132,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"RDM; Originality; Capability Maturity Model; Service (business); Maturity (psychological); Data science; Knowledge management; Computer science; Transformational leadership; Scholarly communication; Public relations; Political science; Business; Sociology; Marketing; Publishing; Qualitative research; Social science","retraction":null,"screen_n_in":null,"score":{"opus":0.1375835139840668,"gpt":0.4348543227040539,"spread":0.2972708087199871,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.00602362,0.00003836683,0.00009130504,0.000218363,0.00006740683,0.001242557,0.002097807,0.00002105809,0.000009680585],"category_scores_gemma":[0.00009737916,0.00002395007,0.00001354718,0.0002887488,0.00006225205,0.08770283,0.0005945235,0.0003147789,0.00000541183],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001320993,"about_ca_system_score_gemma":0.00004445029,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003703867,"about_ca_topic_score_gemma":0.000002644395,"domain_scores_codex":[0.9982756,0.0003654551,0.0003391448,0.0001008462,0.0008153095,0.0001036603],"domain_scores_gemma":[0.9985331,0.0005414932,0.0003646117,0.0004099416,0.00012346,0.00002734227],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000678968,0.00004525837,0.004594244,0.000910119,0.0001775162,0.000004830679,0.01507756,0.0002050955,0.007190028,0.9262936,0.0008711084,0.04395171],"study_design_scores_gemma":[0.03068602,0.00230786,0.1548053,0.002753282,0.0003026788,0.0004024188,0.06757219,0.290056,0.08322262,0.2398051,0.1272104,0.0008762384],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9245747,0.001881223,0.04639523,0.02341645,0.0002570304,0.0006243223,0.000005578471,0.00001126279,0.002834223],"genre_scores_gemma":[0.9902911,0.003977609,0.005513296,0.00006633969,0.00003959004,0.000001306424,0.000007387256,0.000002623117,0.0001007916],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6864885,"threshold_uncertainty_score":0.9997942,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4306292873","doi":"10.1038/s41597-022-01739-y","title":"Implementation of FAIR principles in the IPCC: the WGI AR6 Atlas repository","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Research Data Management Practices","field":"Computer Science","cited_by":132,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Ouranos","funders":"Agencia Estatal de Investigación; Horizon 2020 Framework Programme; Universidad de Cantabria; European Commission; Engineering and Physical Sciences Research Council; U.S. Department of Energy","keywords":"Atlas (anatomy); Computer science; Scrutiny; Data science; Underpinning; Open source; World Wide Web; Information retrieval; Engineering; Software; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.2052956095585589,"gpt":0.4042555140695923,"spread":0.1989599045110334,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.01234975,0.00006620706,0.00006652682,0.0001836417,0.0008341637,0.00273598,0.01712907,0.000006644027,0.00005775915],"category_scores_gemma":[0.0002324062,0.00004122381,0.00002086235,0.001593614,0.0001924618,0.01022348,0.01284557,0.0002062162,0.00001495877],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004116907,"about_ca_system_score_gemma":0.0001880781,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005634693,"about_ca_topic_score_gemma":0.0005265749,"domain_scores_codex":[0.9964911,0.0008271927,0.0003119047,0.0006923985,0.001421596,0.0002558084],"domain_scores_gemma":[0.9935728,0.0003401077,0.000236306,0.005786446,0.00004106186,0.00002327025],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00001563363,0.0004393321,0.01203267,0.00007654819,0.00006845448,0.0001020988,0.009574858,0.0004872331,0.003895187,0.3797972,0.5527745,0.04073631],"study_design_scores_gemma":[0.0001586132,0.00003491616,0.03034921,0.000003815295,0.000005867401,0.00001114328,0.008650947,0.008877522,0.0005082379,0.0005037805,0.9508216,0.00007428559],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6605628,0.001767158,0.1436136,0.1109161,0.01487995,0.009405917,0.007352585,0.0003931157,0.05110883],"genre_scores_gemma":[0.990975,0.0000259823,0.003197361,0.0001995301,0.00004636464,0.000123411,0.001190038,0.000006315073,0.004236021],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3980471,"threshold_uncertainty_score":0.9982993,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2093426783","doi":"10.1371/journal.pbio.1001779","title":"Troubleshooting Public Data Archiving: Suggestions to Increase Participation","year":2014,"lang":"en","type":"article","venue":"PLoS Biology","topic":"Research Data Management Practices","field":"Computer Science","cited_by":130,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Australian Research Council; Natural Sciences and Engineering Research Council of Canada; Australian National University","keywords":"Troubleshooting; Open data; Data sharing; Data science; Dissemination; Information Dissemination; Internet privacy; World Wide Web; Computer science; Telecommunications","retraction":null,"screen_n_in":null,"score":{"opus":0.2915186612585918,"gpt":0.4105975123136387,"spread":0.119078851055047,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.001830791,0.00008950161,0.0001188877,0.0002397935,0.0001667652,0.0008578234,0.004068547,0.00003046237,0.00001637735],"category_scores_gemma":[0.01121653,0.00008003332,0.00001291185,0.0004277672,0.00004743508,0.008106567,0.004285051,0.0001426222,0.000227685],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002010085,"about_ca_system_score_gemma":0.00005043276,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000152501,"about_ca_topic_score_gemma":0.0002072983,"domain_scores_codex":[0.9978632,0.0006606871,0.0002033891,0.0006501953,0.0001732333,0.0004492524],"domain_scores_gemma":[0.9964379,0.0009241343,0.00008955992,0.002258711,0.00006646982,0.0002232078],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000006697957,0.000204199,0.0190243,0.00001543115,0.0000396155,0.000002504337,0.0000626959,0.00001975447,0.003435422,0.9189256,0.001994203,0.05626952],"study_design_scores_gemma":[0.0005004809,0.0004828543,0.05244799,0.00003019562,0.00003077469,0.000006059569,0.00002586998,0.434846,0.0004923117,0.00940489,0.5012928,0.0004398833],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06613673,0.00001078257,0.9011511,0.02497179,0.0001049389,0.0002418227,0.00002354356,0.0001911352,0.007168161],"genre_scores_gemma":[0.9492698,0.0000168222,0.04944615,0.0006794414,0.0001159908,0.00005376996,0.0002584773,0.000006149504,0.0001534051],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9095207,"threshold_uncertainty_score":0.9971124,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2117542660","doi":"10.1016/j.tree.2015.07.006","title":"Archiving Primary Data: Solutions for Long-Term Studies","year":2015,"lang":"en","type":"review","venue":"Trends in Ecology & Evolution","topic":"Research Data Management Practices","field":"Computer Science","cited_by":129,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Guelph; University of British Columbia; University of Alberta; Université Laval; Université du Québec à Montréal; Université de Sherbrooke; Ministry of Environment","funders":"Agence Nationale de la Recherche; Natural Environment Research Council; Sight Research UK","keywords":"Term (time); Principal (computer security); Data sharing; Apprehension; Data science; Computer science; Data curation; Open data; World Wide Web; Psychology; Medicine; Alternative medicine; Computer security","retraction":null,"screen_n_in":null,"score":{"opus":0.5277370542661391,"gpt":0.5182983800153955,"spread":0.009438674250743628,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.005797026,0.0003987595,0.001379771,0.001771097,0.0003083089,0.0003953166,0.006048614,0.0002319094,0.000004602438],"category_scores_gemma":[0.001401567,0.0003645896,0.0001690509,0.001281432,0.0002504607,0.01364588,0.009406678,0.0006238691,0.00005626251],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001647466,"about_ca_system_score_gemma":0.0007354606,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001555735,"about_ca_topic_score_gemma":0.001327459,"domain_scores_codex":[0.9952015,0.001029035,0.0008324061,0.00146829,0.0004037338,0.001064997],"domain_scores_gemma":[0.9944762,0.001611176,0.0006432731,0.003019467,0.0001393393,0.0001105494],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000004094918,0.0001785884,0.0003503956,0.003672876,0.0003378618,0.00002003984,0.00003551626,0.0000111717,3.082154e-8,0.01710345,0.01784886,0.9604371],"study_design_scores_gemma":[0.0004307061,0.0001888083,0.01609838,0.001456208,0.0004775542,0.00004080331,0.00001293567,0.001372721,4.717617e-9,0.001058864,0.9784101,0.0004529001],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.000002229578,0.9183139,0.07798803,0.0004656202,0.001185678,0.0008603953,0.0001852021,0.000135224,0.0008637373],"genre_scores_gemma":[0.00001759712,0.9848936,0.009381891,0.00002276145,0.0002500911,0.000722455,0.00253285,0.00002793665,0.002150795],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.9605613,"threshold_uncertainty_score":0.9998806,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2147717883","doi":"10.1177/0165551511412705","title":"A conceptual framework for managing very diverse data for complex, interdisciplinary science","year":2011,"lang":"en","type":"article","venue":"Journal of Information Science","topic":"Research Data Management Practices","field":"Computer Science","cited_by":127,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Aboriginal Affairs Northern Dev Canada; University of Waterloo","funders":"Sixth Framework Programme; Norges Forskningsråd; European Commission; Association of Research Libraries; National Science Foundation","keywords":"Data science; Computer science; Adaptation (eye); Dimension (graph theory); Diversity (politics); Data management; Conceptual framework; Scale (ratio); Knowledge management; Sociology; Data mining; Social science; Geography","retraction":null,"screen_n_in":null,"score":{"opus":0.328305162865112,"gpt":0.4488184725383007,"spread":0.1205133096731887,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.01128896,0.0001022246,0.0001564551,0.001112701,0.0009190767,0.002992001,0.01331308,0.00002292163,0.000009939642],"category_scores_gemma":[0.004827922,0.00008264939,0.00005203715,0.001745695,0.00136811,0.2586565,0.006604997,0.0001843186,0.00001264838],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001393354,"about_ca_system_score_gemma":0.0008197164,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000003719784,"about_ca_topic_score_gemma":8.198658e-7,"domain_scores_codex":[0.9971529,0.00002584161,0.0006319692,0.0002741787,0.001450003,0.000465134],"domain_scores_gemma":[0.9956592,0.0004526569,0.001014698,0.00126924,0.001381095,0.000223102],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00009918568,0.00004111677,0.00007903342,0.00003350655,0.00001280472,0.000002803519,0.005581321,0.00006551837,0.0002358137,0.9578583,0.00227073,0.03371989],"study_design_scores_gemma":[0.002526373,0.002102356,0.01228197,0.0003820089,0.0000561006,0.0001994727,0.0351969,0.7143938,0.002509564,0.1306745,0.09892161,0.0007552773],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.002156645,0.00001027947,0.9922838,0.001160684,0.0008293118,0.0003712488,0.00003514858,0.00002141014,0.003131495],"genre_scores_gemma":[0.2853339,0.00003466212,0.7141077,0.0004335933,0.00006206489,0.000007221541,0.000005647826,0.000002373436,0.00001280659],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.8271837,"threshold_uncertainty_score":0.998043,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2515291001","doi":"10.1177/0270467616668760","title":"How Do Scientists Define Openness? Exploring the Relationship Between Open Science Policies and Research Practice","year":2016,"lang":"en","type":"article","venue":"Bulletin of Science Technology & Society","topic":"Research Data Management Practices","field":"Computer Science","cited_by":121,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Victoria","funders":"Economic and Social Research Council","keywords":"Openness to experience; Open science; Science policy; Government (linguistics); Political science; Research policy; Public relations; Best practice; Engineering ethics; Sociology; Psychology; Public administration; Engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.3020403008169142,"gpt":0.4386357982622262,"spread":0.1365954974453119,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","sts","scholarly_communication","open_science"],"category_scores_codex":[0.04363168,0.0001781192,0.0002316451,0.001171246,0.004699969,0.009029695,0.02579698,0.00008487247,0.000005867632],"category_scores_gemma":[0.04923072,0.0001085561,0.00004411837,0.01395786,0.03776485,0.03017032,0.03228496,0.0006493577,0.00003687358],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002709323,"about_ca_system_score_gemma":0.000865276,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002457856,"about_ca_topic_score_gemma":0.000005704285,"domain_scores_codex":[0.9936678,0.0002969492,0.0003461379,0.001545935,0.002869417,0.001273723],"domain_scores_gemma":[0.988823,0.005934826,0.0003853998,0.003011581,0.00162946,0.0002157206],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00000423303,0.0000472723,0.02724683,0.00001604486,0.00001055306,0.00000331569,0.001032384,3.766822e-7,0.007157296,0.9461898,0.003073511,0.01521836],"study_design_scores_gemma":[0.001212259,0.0005629291,0.2966384,0.0004313556,0.00003147016,0.0001051006,0.03407964,0.00006456078,0.03053248,0.0549828,0.5806715,0.000687444],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.4401687,0.0003221582,0.0157162,0.5382999,0.0001776086,0.0009916878,0.000007674726,0.0001589768,0.004157052],"genre_scores_gemma":[0.9349493,0.0006799573,0.06272803,0.000157169,0.00002902214,0.0001435727,2.333477e-7,0.0000088327,0.001303888],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.891207,"threshold_uncertainty_score":0.9965958,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3008436047","doi":"10.1371/journal.pone.0229182","title":"The views, perspectives, and experiences of academic researchers with data sharing and reuse: A meta-synthesis","year":2020,"lang":"en","type":"article","venue":"PLoS ONE","topic":"Research Data Management Practices","field":"Computer Science","cited_by":116,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Carleton University; Ontario Council of University Libraries; University of Toronto","funders":"","keywords":"Data sharing; Incentive; Data curation; Computer science; Reuse; Data quality; Data management; Data science; Open data; Thematic analysis; Knowledge management; World Wide Web; Qualitative research; Medicine; Business; Database; Sociology; Engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.696846400623449,"gpt":0.4203261462132086,"spread":0.2765202544102405,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.002240267,0.00008789195,0.0002011385,0.0000501008,0.0001835486,0.0009134718,0.005717515,0.00002088675,0.00001265132],"category_scores_gemma":[0.006367038,0.00005171595,0.000009678759,0.0004168408,0.0003544892,0.01008421,0.007516144,0.0002922922,0.000001696124],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000009302085,"about_ca_system_score_gemma":0.00004164531,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004233193,"about_ca_topic_score_gemma":0.00001048855,"domain_scores_codex":[0.997919,0.0002227049,0.0001532334,0.0006922579,0.0007842318,0.0002285707],"domain_scores_gemma":[0.9969982,0.000875076,0.0001096568,0.001810711,0.00006659472,0.0001398003],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"qualitative","study_design_scores_codex":[0.001414648,0.001776789,0.01778392,0.003071533,0.02723242,0.0002179387,0.6076261,0.000006952985,0.02058561,0.2584311,0.01153504,0.05031792],"study_design_scores_gemma":[0.00232833,0.003014729,0.007319228,0.001125296,0.004270654,0.00002535667,0.5555016,0.3085067,0.05570684,0.003107056,0.05702895,0.002065338],"study_design_candidate":"qualitative","study_design_consensus":"qualitative","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7632408,0.05629367,0.01251954,0.1639888,0.00001158379,0.001531044,0.00004150551,0.0001597018,0.002213349],"genre_scores_gemma":[0.8664714,0.04755714,0.08480951,0.0002190864,0.00006072899,0.0005038346,0.000003276622,0.00002062446,0.0003543726],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3084997,"threshold_uncertainty_score":0.999662,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2179000687","doi":"10.1371/journal.pcbi.1005097","title":"Ten Simple Rules for Digital Data Storage","year":2016,"lang":"en","type":"editorial","venue":"PLoS Computational Biology","topic":"Research Data Management Practices","field":"Computer Science","cited_by":105,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal; Western University","funders":"Natural Sciences and Engineering Research Council of Canada; Energy Biosciences Institute; Washington State University; Life Sciences Research Foundation; Gordon and Betty Moore Foundation; U.S. Environmental Protection Agency; National Science Foundation","keywords":"Petabyte; Pace; Computer science; Data sharing; Data science; Data curation; Data collection; Computer data storage; Information repository; Open data; Digital data; Database; World Wide Web; Data mining; Big data; Geography; Transmission (telecommunications); Telecommunications","retraction":null,"screen_n_in":null,"score":{"opus":0.1112147531237753,"gpt":0.3966651724734826,"spread":0.2854504193497073,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.0007931145,0.0002950951,0.0003895577,0.0002918806,0.0001831457,0.00211796,0.008938952,0.0003360936,0.00001956776],"category_scores_gemma":[0.007725737,0.0002302522,0.00007097566,0.0001485292,0.0001607698,0.01343856,0.006049826,0.0003363485,0.0002775156],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009328287,"about_ca_system_score_gemma":0.0006113497,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001076244,"about_ca_topic_score_gemma":0.000003726562,"domain_scores_codex":[0.9966436,0.0001799544,0.0004263554,0.00147521,0.0007482464,0.0005266116],"domain_scores_gemma":[0.9853175,0.01173287,0.0004085644,0.001890933,0.0005306194,0.0001195719],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00002505838,0.00008663139,0.00001578509,0.00007144951,0.0002396639,0.000007773135,0.000005337189,0.00003590395,0.00000495728,0.05150296,0.9388916,0.009112822],"study_design_scores_gemma":[0.0004336739,0.0001629239,0.00001696676,0.0000243525,0.00001766445,9.559407e-7,0.000001662999,0.02357441,0.000001075171,0.052682,0.922811,0.0002732678],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"editorial","genre_scores_codex":[0.000008072181,0.0001233142,0.770673,0.001558635,0.2030906,0.0005176387,0.02321798,0.0001610652,0.0006496689],"genre_scores_gemma":[0.0003924657,0.0002240219,0.09412514,0.000168144,0.728475,0.0002068195,0.1742451,0.00006756116,0.002095748],"genre_candidate":"editorial","genre_consensus":null,"teacher_disagreement_score":0.6765479,"threshold_uncertainty_score":0.9989179,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2239105819","doi":"10.7191/jeslib.2015.1085","title":"Research Data Services in Academic Libraries: Data Intensive Roles for the Future?","year":2015,"lang":"en","type":"article","venue":"Journal of eScience Librarianship","topic":"Research Data Management Practices","field":"Computer Science","cited_by":103,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"National Science Foundation","keywords":"Academic library; Service (business); Stratified sampling; Sample (material); Survey data collection; Library science; Public relations; Political science; Business; Psychology; Computer science; Marketing; Statistics; Mathematics","retraction":null,"screen_n_in":null,"score":{"opus":0.5760048811854739,"gpt":0.4825300829947619,"spread":0.09347479819071203,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["metaresearch","scholarly_communication","open_science"],"category_scores_codex":[0.03288246,0.0001633374,0.0002695722,0.0007998267,0.0003261284,0.007304121,0.05750553,0.0001087274,0.000004544506],"category_scores_gemma":[0.008808945,0.0001014058,0.00003462449,0.003066883,0.0004581302,0.2098778,0.01711431,0.001683567,0.0000136011],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00005210643,"about_ca_system_score_gemma":0.001960943,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000513273,"about_ca_topic_score_gemma":0.00009838688,"domain_scores_codex":[0.994323,0.0008996286,0.0007499778,0.0008404728,0.002403749,0.0007831351],"domain_scores_gemma":[0.9872795,0.005116418,0.0006447195,0.005649698,0.0008785792,0.0004310631],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0005075656,0.0001622376,0.01121819,0.0001504512,0.00009813113,0.0002432734,0.005372099,0.0003279046,0.0001813905,0.6633213,0.2694723,0.04894518],"study_design_scores_gemma":[0.001133779,0.0003988389,0.008809464,0.0002603005,0.00002133225,0.0001146057,0.03877836,0.1772248,0.0001450524,0.04446603,0.7283874,0.0002601153],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.007509307,0.02480567,0.4918541,0.4678672,0.003882453,0.001592403,0.0002147274,0.00008063054,0.002193461],"genre_scores_gemma":[0.5372964,0.01616155,0.4190141,0.01422865,0.01077608,0.00007432807,0.000266259,0.0001171296,0.00206557],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.6188553,"threshold_uncertainty_score":0.9995403,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1724212071","doi":"","title":"The nine flavours of open access scholarly publishing.","year":2004,"lang":"en","type":"article","venue":"PubMed","topic":"Research Data Management Practices","field":"Computer Science","cited_by":101,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Medicine; Open access publishing; Publishing; Scholarly communication; MEDLINE; World Wide Web; Library science; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.2412217605689317,"gpt":0.4019658656179179,"spread":0.1607441050489861,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.005979112,0.00007926321,0.0001020691,0.0001437571,0.0002478324,0.1878898,0.02975554,0.00002822084,0.000003809691],"category_scores_gemma":[0.006904059,0.00005386945,0.00002903584,0.001126809,0.00008228408,0.467559,0.01495694,0.0003012093,0.000009210322],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006946234,"about_ca_system_score_gemma":0.000125735,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001036837,"about_ca_topic_score_gemma":0.000175729,"domain_scores_codex":[0.9979764,0.0001598915,0.0002228269,0.0003642715,0.0008030942,0.0004735371],"domain_scores_gemma":[0.9974996,0.0002995566,0.0002165422,0.001582494,0.0002527305,0.0001490235],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.00001691551,0.00007366687,0.001728737,0.00001056723,0.00005085808,0.00001257666,0.00004000011,0.00004525221,0.000007175675,0.6237978,0.01371588,0.3605006],"study_design_scores_gemma":[0.0009774751,0.00002566171,0.5366974,0.00001046536,0.000008085844,0.000003760039,0.00004745474,0.00008741723,0.001241026,0.02525124,0.4354871,0.000162915],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.01944564,0.001112266,0.1735933,0.4544176,0.001304137,0.004840211,0.00001912536,0.0002561065,0.3450117],"genre_scores_gemma":[0.9796163,0.0007971794,0.00714009,0.001325995,0.0001149773,0.001688402,0.000009407116,0.00001613184,0.009291519],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9601707,"threshold_uncertainty_score":0.9930099,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2943122689","doi":"10.1016/j.tree.2019.04.005","title":"Ecological Data Should Not Be So Hard to Find and Reuse","year":2019,"lang":"en","type":"article","venue":"Trends in Ecology & Evolution","topic":"Research Data Management Practices","field":"Computer Science","cited_by":89,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University; McGill University; Espace pour la vie; Université de Sherbrooke; Université de Montréal; Cégep Marie-Victorin","funders":"","keywords":"Interoperability; Reuse; Domain (mathematical analysis); Data science; Computer science; Transformation (genetics); Ecology; World Wide Web; Biology; Mathematics","retraction":null,"screen_n_in":null,"score":{"opus":0.2314376691550181,"gpt":0.408916438861722,"spread":0.1774787697067039,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001972165,0.000125316,0.0001974524,0.0005261223,0.0001028526,0.000420869,0.003700724,0.0001355227,0.0001755818],"category_scores_gemma":[0.00124565,0.0001177329,0.00001654048,0.0006480024,0.00009319494,0.008898151,0.006979758,0.000312576,0.0002315371],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000171461,"about_ca_system_score_gemma":0.0000619396,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001115145,"about_ca_topic_score_gemma":0.001377428,"domain_scores_codex":[0.9977211,0.00031353,0.0002439484,0.0009499293,0.0002933588,0.0004780866],"domain_scores_gemma":[0.996752,0.0004528364,0.00008551127,0.00257312,0.00003056452,0.0001059356],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.000218484,0.0008518702,0.623575,0.00003965041,0.00009044282,0.0001242624,0.0004497752,0.0008103364,0.002165781,0.1864848,0.1310512,0.05413852],"study_design_scores_gemma":[0.0004166756,0.0003205066,0.920889,0.000003604792,0.000006154914,0.000007353416,0.00001960138,0.02433086,0.00001213642,0.0004064079,0.05346532,0.0001223472],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9277917,0.00004195373,0.006229297,0.06267818,0.0005012666,0.0003028303,0.00004022432,0.00009540806,0.002319126],"genre_scores_gemma":[0.9734311,0.00008546687,0.02006246,0.001295529,0.00003768211,0.00003065737,0.00008702945,0.000006285472,0.00496377],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2973141,"threshold_uncertainty_score":0.8699769,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1605878099","doi":"10.1111/j.1755-263x.2012.00259.x","title":"Willing or unwilling to share primary biodiversity data: results and implications of an international survey","year":2012,"lang":"en","type":"article","venue":"Conservation Letters","topic":"Research Data Management Practices","field":"Computer Science","cited_by":80,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"","keywords":"Data sharing; Standardization; Biodiversity; Data quality; Publishing; Business; Survey data collection; Computer science; Environmental resource management; Internet privacy; Marketing; Political science; Ecology; Medicine","retraction":null,"screen_n_in":null,"score":{"opus":0.3041531773281683,"gpt":0.3848607261312832,"spread":0.08070754880311487,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.002259227,0.0000651009,0.00008291851,0.0001681574,0.0001229675,0.00046978,0.001793655,0.00001543771,0.000004020066],"category_scores_gemma":[0.001120817,0.00006274279,0.000007484319,0.0003823039,0.00003731744,0.02788154,0.00134718,0.00006410669,0.000007897943],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004392263,"about_ca_system_score_gemma":0.00004500317,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000509881,"about_ca_topic_score_gemma":0.0001337276,"domain_scores_codex":[0.9988373,0.0001754629,0.0002211562,0.0003437064,0.0002537213,0.0001686885],"domain_scores_gemma":[0.9981563,0.0004256259,0.0001815736,0.001002091,0.0001213014,0.0001130761],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0001785453,0.0001743639,0.8793918,0.00006471871,0.0001168596,0.000003255918,0.001121648,0.0001547666,0.004704757,0.003825956,0.09191964,0.01834366],"study_design_scores_gemma":[0.0002057594,0.00001851707,0.9572164,0.00001398424,0.000002978448,0.000001756736,0.00002385881,0.001760793,0.00006327908,0.000003975266,0.04060908,0.00007964275],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.4564701,0.00002345727,0.4246425,0.116817,0.0002119755,0.0003043168,0.00137183,0.00006016112,0.00009859651],"genre_scores_gemma":[0.6107793,0.0003786305,0.3251955,0.05217503,0.0001786784,0.0000219521,0.01110775,0.00001357114,0.0001495904],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1543092,"threshold_uncertainty_score":0.985715,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2095114014","doi":"10.1177/0340035212473089","title":"Academic librarians and research data services: preparation and attitudes","year":2013,"lang":"en","type":"article","venue":"IFLA Journal","topic":"Research Data Management Practices","field":"Computer Science","cited_by":78,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"National Endowment for the Humanities; National Science Foundation","keywords":"Metadata; Preparedness; Service (business); Public relations; Data curation; Library science; Scholarly communication; Business; Knowledge management; Political science; World Wide Web; Computer science; Marketing; Publishing","retraction":null,"screen_n_in":null,"score":{"opus":0.2727682969496663,"gpt":0.4705235793203441,"spread":0.1977552823706778,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.00372381,0.00007251534,0.00007921413,0.0002265607,0.0003910526,0.009979502,0.003186583,0.00005045391,0.00003419374],"category_scores_gemma":[0.0002777094,0.00005827427,0.000006281202,0.0002819162,0.00008804152,0.09265433,0.004233859,0.0007710475,0.00006238358],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001533933,"about_ca_system_score_gemma":0.00008134456,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000111603,"about_ca_topic_score_gemma":0.00002226943,"domain_scores_codex":[0.9979756,0.0004111179,0.0001969795,0.000406295,0.0006860059,0.0003240149],"domain_scores_gemma":[0.9982927,0.0003715985,0.00009528067,0.0008968731,0.0001238368,0.0002196724],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001146886,0.0003149049,0.1419089,0.0008629463,0.0004608286,0.0003328186,0.008955541,0.00004895795,0.01304159,0.2492179,0.2062447,0.3784963],"study_design_scores_gemma":[0.001280079,0.0005488348,0.3645467,0.0003737917,0.0000236034,0.00112559,0.000941578,0.1785361,0.0003264874,0.03958856,0.4121893,0.0005193669],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7743282,0.008948624,0.1384529,0.06525798,0.0004317489,0.001342822,0.00001941028,0.0001722421,0.01104605],"genre_scores_gemma":[0.9301426,0.02210769,0.04455916,0.0005503059,0.000498595,0.00002055834,0.00002053754,0.00001579088,0.002084792],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3779769,"threshold_uncertainty_score":0.9910482,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3093459717","doi":"10.3390/nano10102033","title":"Metadata Stewardship in Nanosafety Research: Community-Driven Organisation of Metadata Schemas to Support FAIR Nanoscience Data","year":2020,"lang":"en","type":"article","venue":"Nanomaterials","topic":"Research Data Management Practices","field":"Computer Science","cited_by":74,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"High Energy Physics; Institute of High Energy Physics; Center for the Environmental Implications of NanoTechnology; University of California, San Diego; Universiteit Maastricht; Hanyang University; Rijksinstituut voor Volksgezondheid en Milieu; European Commission; National Cancer Institute; National Institutes of Health; University of Alberta","keywords":"Metadata; Computer science; Interoperability; Workflow; Data quality; Data curation; Data science; Metadata management; Stewardship (theology); Data governance; Data management; Schema (genetic algorithms); World Wide Web; Database; Engineering; Information retrieval; Metric (unit)","retraction":null,"screen_n_in":null,"score":{"opus":0.623165160250243,"gpt":0.471601839170566,"spread":0.1515633210796771,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.01992585,0.0002407363,0.0005765284,0.0005448369,0.000331994,0.003419103,0.01947796,0.00009205427,0.00009246069],"category_scores_gemma":[0.01200986,0.0002249962,0.00003235682,0.002696125,0.0002349772,0.06558072,0.02604819,0.0004120145,0.0001792795],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001002537,"about_ca_system_score_gemma":0.0006426059,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001123164,"about_ca_topic_score_gemma":0.0003327458,"domain_scores_codex":[0.9905834,0.004340414,0.0009990816,0.001224765,0.002055174,0.0007972305],"domain_scores_gemma":[0.9911574,0.001182512,0.000354595,0.006663225,0.0002765478,0.0003657931],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0001364513,0.0003347538,0.0009525463,0.000357263,0.00007052934,0.00007667836,0.002467397,0.00002507894,0.9032159,0.08309174,0.006792277,0.002479368],"study_design_scores_gemma":[0.00129176,0.001419222,0.008995967,0.0001315998,0.00003858639,0.00001412225,0.001681684,0.001396397,0.73825,0.0005920437,0.2454764,0.0007121887],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7223527,0.0001059119,0.2352531,0.03116485,0.0009542905,0.004014221,0.002984263,0.0004015227,0.002769053],"genre_scores_gemma":[0.9299687,0.0001467897,0.06822887,0.0005660841,0.00006418812,0.00003917491,0.0007124284,0.00002414412,0.0002496263],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.2386841,"threshold_uncertainty_score":0.9976155,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3196831919","doi":"10.1111/cobi.13835","title":"Closing the knowledge‐action gap in conservation with open science","year":2021,"lang":"en","type":"article","venue":"Conservation Biology","topic":"Research Data Management Practices","field":"Computer Science","cited_by":74,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa; Nature Conservancy of Canada; Canadian Parks and Wilderness Society; Carleton University","funders":"H2020 Marie Skłodowska-Curie Actions; Natural Sciences and Engineering Research Council of Canada; Horizon 2020 Framework Programme; Environment and Climate Change Canada; European Commission","keywords":"Closing (real estate); Action (physics); Conservation science; Political science; Ecology; Biology; Biodiversity; Physics","retraction":null,"screen_n_in":null,"score":{"opus":0.3036386708784558,"gpt":0.4516840456361518,"spread":0.148045374757696,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.002847936,0.00008375625,0.0001059862,0.0001829884,0.0003558681,0.002044501,0.002720304,0.00003788565,0.00001434197],"category_scores_gemma":[0.001455113,0.0000597092,0.00001015311,0.002352328,0.0003792592,0.01417755,0.00175298,0.0001700503,0.00002618366],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001060682,"about_ca_system_score_gemma":0.0009943959,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005982285,"about_ca_topic_score_gemma":0.0027341,"domain_scores_codex":[0.9982845,0.0005078895,0.0002164805,0.0005442452,0.0001788847,0.0002680344],"domain_scores_gemma":[0.9978296,0.0006067788,0.0001563992,0.0008685152,0.0004986866,0.00004005191],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.00002418554,0.00007709696,0.0849485,0.00001281102,0.00001087046,0.0000112079,0.0002217034,0.00001417975,0.01443288,0.8862991,0.0009788102,0.01296864],"study_design_scores_gemma":[0.001361763,0.0002053576,0.5265577,0.00008561056,0.000009996742,0.0000664812,0.0007195806,0.05877664,0.01014291,0.01560432,0.3861274,0.0003422317],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3126135,0.0003050727,0.4988485,0.1607881,0.0005916234,0.0009730365,0.000003701329,0.0001031856,0.02577332],"genre_scores_gemma":[0.9707653,0.0002820806,0.02127087,0.006077148,0.00004611093,0.0001131559,0.00004949346,0.000007400561,0.001388435],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8706948,"threshold_uncertainty_score":0.9996107,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4376140397","doi":"10.1002/ece3.9961","title":"Reproducibility in ecology and evolution: Minimum standards for data and code","year":2023,"lang":"en","type":"article","venue":"Ecology and Evolution","topic":"Research Data Management Practices","field":"Computer Science","cited_by":69,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University; University of British Columbia; York University","funders":"","keywords":"Commit; Computer science; Data science; Citation; Open science; Simple (philosophy); Code (set theory); Open data; Ecology; World Wide Web; Biology; Database; Programming language; Epistemology; Mathematics","retraction":null,"screen_n_in":null,"score":{"opus":0.1016794395175727,"gpt":0.3895820031292132,"spread":0.2879025636116405,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.009806084,0.00008546516,0.0001612997,0.0002184742,0.0002451936,0.000164873,0.0005550907,0.0001003623,0.000003198444],"category_scores_gemma":[0.005314224,0.00008638905,0.000007784691,0.0003485758,0.0002312427,0.005282576,0.002010361,0.0001347733,0.000004488174],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001266661,"about_ca_system_score_gemma":0.0001636446,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000619533,"about_ca_topic_score_gemma":0.003874653,"domain_scores_codex":[0.9975591,0.0002967988,0.0002147312,0.001452771,0.0001360914,0.0003405248],"domain_scores_gemma":[0.9975972,0.000657578,0.00007823195,0.001525441,0.00007837482,0.00006322216],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0001500858,0.00008048055,0.8864599,0.000160556,0.00003439323,0.00002013929,0.0001881754,0.00002264501,0.0001453056,0.08890407,0.02057157,0.003262673],"study_design_scores_gemma":[0.0005710675,0.0002142327,0.8804646,0.000005062634,0.000008215909,0.0000127275,0.00008265592,0.08536927,0.000003389414,0.01766995,0.01551875,0.00008004723],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.929816,0.0006569316,0.0513491,0.01640953,0.0004501352,0.0007310191,0.0002564954,0.000100278,0.0002305084],"genre_scores_gemma":[0.9953154,0.0007670351,0.003265324,0.00007020734,0.00004981809,0.00005942802,0.00007770723,0.000004014377,0.0003910416],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.08534662,"threshold_uncertainty_score":0.6362007,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2891515454","doi":"10.1093/biosci/biy097","title":"Data-Intensive Ecological Research Is Catalyzed by Open Science and Team Science","year":2018,"lang":"en","type":"article","venue":"BioScience","topic":"Research Data Management Practices","field":"Computer Science","cited_by":69,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Global Lake Ecological Observatory Network; U.S. Department of Agriculture; National Institute of Food and Agriculture; Center for Hierarchical Manufacturing, National Science Foundation; National Science Foundation","keywords":"Open science; Citizen science; Sociology; Engineering ethics; Ecology; Computer science; Data science; Engineering; Biology; Mathematics","retraction":null,"screen_n_in":null,"score":{"opus":0.3927260235470146,"gpt":0.5131003182401698,"spread":0.1203742946931552,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts","scholarly_communication","open_science"],"consensus_categories":["metaresearch","sts","scholarly_communication","open_science"],"category_scores_codex":[0.03933838,0.0001879727,0.0002036244,0.0008885725,0.004173272,0.02253016,0.06483028,0.00004822292,0.00003667732],"category_scores_gemma":[0.02792864,0.0001462084,0.00001085365,0.01172876,0.02713388,0.09557435,0.09145036,0.0003863026,0.0002970122],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002942256,"about_ca_system_score_gemma":0.002956514,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005620451,"about_ca_topic_score_gemma":0.00006167498,"domain_scores_codex":[0.989381,0.0002400683,0.0003037759,0.003538602,0.004849086,0.001687496],"domain_scores_gemma":[0.988779,0.0005704648,0.0001590098,0.005535755,0.004218948,0.0007368131],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00003168095,0.0003427459,0.001211844,0.00001936477,0.000007835294,0.00003857839,0.0009924471,1.151769e-7,0.5688369,0.1462237,0.2614947,0.02080006],"study_design_scores_gemma":[0.000820846,0.001519373,0.02195143,0.00006513846,0.000007968527,0.00009769807,0.001741134,0.07761497,0.2271452,0.003465384,0.6646307,0.0009401664],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.612812,0.0005978395,0.07539754,0.1313433,0.003470372,0.007281938,0.0007039891,0.0006925277,0.1677005],"genre_scores_gemma":[0.9649482,0.0002644534,0.02992257,0.003197412,0.00007575923,0.00003763715,0.000006047563,0.000007981851,0.001539899],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.403136,"threshold_uncertainty_score":0.9971232,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3004422628","doi":"10.1111/jvs.12864","title":"Managing data locally to answer questions globally: The role of collaborative science in ecology","year":2020,"lang":"en","type":"article","venue":"Journal of Vegetation Science","topic":"Research Data Management Practices","field":"Computer Science","cited_by":67,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université Laval; Université de Sherbrooke; Natural Resources Canada; Canadian Forest Service","funders":"","keywords":"Grassroots; Acknowledgement; Scope (computer science); Stewardship (theology); Citizen science; Scale (ratio); Quality (philosophy); Incentive; Data quality; Business; Ecology; Computer science; Political science; Service (business); Marketing; Geography","retraction":null,"screen_n_in":null,"score":{"opus":0.07106950312918711,"gpt":0.3957373691196359,"spread":0.3246678659904488,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.007852842,0.00006784702,0.0001336077,0.0005793218,0.0002218646,0.001399474,0.01104288,0.00001198326,0.000002041559],"category_scores_gemma":[0.005431374,0.00004880221,0.00001381027,0.007216575,0.0007234713,0.03393184,0.002443521,0.000188587,0.00001088873],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001107058,"about_ca_system_score_gemma":0.00282921,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001937171,"about_ca_topic_score_gemma":0.00007286632,"domain_scores_codex":[0.9970253,0.0001738196,0.0004408125,0.0003919559,0.001703086,0.000264999],"domain_scores_gemma":[0.9970345,0.0002464034,0.000480213,0.000752412,0.001275575,0.0002109315],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00009008043,0.0003208736,0.02474224,0.00005260929,0.00002463325,0.00007532067,0.008213266,0.06338798,0.05977119,0.706142,0.0006786268,0.1365012],"study_design_scores_gemma":[0.0006686139,0.001168095,0.2691655,0.0002328803,0.00002271684,0.00005703857,0.004748188,0.6742853,0.007132026,0.0127939,0.02942013,0.0003056517],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0676311,0.0006606705,0.8375276,0.08527172,0.0006612891,0.0006062113,0.000007350423,0.00001963238,0.007614485],"genre_scores_gemma":[0.9362245,0.0001417655,0.06221201,0.001381084,0.00002856994,0.000002088527,2.308279e-7,0.000001764264,0.000007989499],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8685934,"threshold_uncertainty_score":0.9996372,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2074846102","doi":"10.1038/ejhg.2013.131","title":"Data sharing in large research consortia: experiences and recommendations from ENGAGE","year":2013,"lang":"en","type":"article","venue":"European Journal of Human Genetics","topic":"Research Data Management Practices","field":"Computer Science","cited_by":66,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"Norwegian Institute of Public Health; Karolinska Institutet; University of Oxford; Vrije Universiteit Amsterdam; Wellcome Trust; Helsingin Yliopisto","keywords":"Data sharing; Genetic data; Survey data collection; Data science; Knowledge management; Business; Public relations; Computer science; Political science; Medicine; Environmental health; Alternative medicine","retraction":null,"screen_n_in":null,"score":{"opus":0.3909849470253792,"gpt":0.4525716711664441,"spread":0.06158672414106486,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.009384478,0.00008013518,0.0001190289,0.0003458215,0.0002781699,0.002780299,0.005709731,0.00001108503,0.0001623505],"category_scores_gemma":[0.0007777676,0.00007163564,0.0000133389,0.0002810548,0.0001052698,0.009343696,0.005756874,0.0006346407,0.00004466459],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001928284,"about_ca_system_score_gemma":0.00003359602,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002688715,"about_ca_topic_score_gemma":0.00004826441,"domain_scores_codex":[0.9966048,0.001660986,0.000507014,0.0003893342,0.0005231606,0.0003146995],"domain_scores_gemma":[0.9977733,0.0003846701,0.000229037,0.001294662,0.0001749447,0.0001433499],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.000059751,0.002486732,0.1840775,0.0001660664,0.0006374759,0.005425516,0.1374055,0.0001209698,0.03284118,0.08012071,0.244233,0.3124256],"study_design_scores_gemma":[0.003799939,0.001559718,0.5457882,0.0005112392,0.0000298936,0.00009608168,0.05963806,0.03039948,0.001238496,0.006169909,0.3499253,0.0008437343],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9652985,0.001290023,0.02586727,0.002596617,0.0001562945,0.0001818786,0.00001558627,0.00001272122,0.004581115],"genre_scores_gemma":[0.9683675,0.00156762,0.0296549,0.00008155683,0.0001238689,0.000002272641,0.0000138653,0.00001149483,0.000176888],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3617107,"threshold_uncertainty_score":0.9996698,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2617597628","doi":"10.1371/journal.pone.0178261","title":"Research data management in academic institutions: A scoping review","year":2017,"lang":"en","type":"review","venue":"PLoS ONE","topic":"Research Data Management Practices","field":"Computer Science","cited_by":64,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"St. Michael's Hospital; Carleton University; University of Toronto","funders":"","keywords":"Data management; Data quality; Data collection; Data sharing; Data curation; Data science; Stakeholder; Psychological intervention; Data element; Research design; Knowledge management; Metadata; Computer science; Medicine; Political science; Engineering; Public relations; World Wide Web; Database; Sociology; Operations management; Social science; Metric (unit); Alternative medicine","retraction":null,"screen_n_in":null,"score":{"opus":0.9400748228732019,"gpt":0.6471055609402162,"spread":0.2929692619329857,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","scholarly_communication","open_science","research_integrity","insufficient_payload"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.02015797,0.0003556329,0.00170158,0.001498296,0.0004367517,0.002814791,0.03597315,0.0001987055,0.00002351943],"category_scores_gemma":[0.005829882,0.0003203492,0.00008873404,0.001992885,0.000212801,0.03054821,0.03550202,0.002867382,0.001226007],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003437317,"about_ca_system_score_gemma":0.001311679,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000536731,"about_ca_topic_score_gemma":0.00005518724,"domain_scores_codex":[0.9910957,0.001914535,0.001159359,0.002042575,0.002822537,0.0009652493],"domain_scores_gemma":[0.9846984,0.0008594667,0.0006712484,0.01343602,0.0001489242,0.000185947],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[4.783509e-7,0.000244701,4.28259e-7,0.369239,0.0002447912,0.0001833685,0.000002526247,2.953029e-8,1.274046e-8,0.02365426,0.001630025,0.6048004],"study_design_scores_gemma":[0.00005214295,0.00001003219,4.821011e-7,0.4870763,0.0001936514,0.000002276391,0.000001082768,0.00003518412,4.307276e-8,0.00005847196,0.5124093,0.0001610181],"study_design_candidate":"systematic_review","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[1.669869e-8,0.9691039,0.00136759,0.001468729,0.0000678731,0.007716419,0.00004282278,0.00008639904,0.02014624],"genre_scores_gemma":[3.823917e-8,0.9803587,0.01541881,0.0001084217,0.0001193304,0.001537367,0.0003795165,0.00002720107,0.002050678],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.6046394,"threshold_uncertainty_score":0.9999248,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4322505217","doi":"10.1017/aap.2022.40","title":"Will It Ever Be FAIR?","year":2023,"lang":"en","type":"article","venue":"Advances in Archaeological Practice","topic":"Research Data Management Practices","field":"Computer Science","cited_by":64,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia, Okanagan Campus; University of British Columbia","funders":"Arizona State University","keywords":"Stewardship (theology); Interoperability; Reuse; Discipline; Engineering ethics; Political science; Sociology; Business; Computer science; Law; Engineering; World Wide Web","retraction":null,"screen_n_in":null,"score":{"opus":0.09616126529380842,"gpt":0.4240506872703638,"spread":0.3278894219765554,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.003751536,0.0001903999,0.0002171501,0.0003408666,0.0001532374,0.0003626888,0.003327386,0.00007637517,0.00007087891],"category_scores_gemma":[0.01877758,0.0001570351,0.00005117495,0.002232567,0.000377452,0.116832,0.004621472,0.0006886293,0.0005547117],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006810871,"about_ca_system_score_gemma":0.00004819578,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004919367,"about_ca_topic_score_gemma":0.00008433243,"domain_scores_codex":[0.9960563,0.0008952513,0.000376783,0.0009226216,0.0009581904,0.0007908193],"domain_scores_gemma":[0.9911223,0.007251363,0.0002087219,0.001159831,0.0001045386,0.0001532495],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00008047396,0.0001581825,0.00137716,0.00003062123,0.00001821579,0.001573577,0.000440845,0.001245985,0.00003369972,0.9439575,0.002480202,0.04860356],"study_design_scores_gemma":[0.000299539,0.0001760525,0.00242068,0.00001484202,0.000004314383,0.0000330878,0.0004154603,0.006121433,0.00002286016,0.04535472,0.9449291,0.0002079222],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.001906133,0.001495837,0.5579078,0.3096512,0.0007236171,0.0007055136,0.000007851268,0.0008852992,0.1267168],"genre_scores_gemma":[0.5720155,0.1005187,0.2959594,0.02169679,0.0002437911,0.0003623832,0.00004883792,0.0000372456,0.009117335],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9424489,"threshold_uncertainty_score":0.9894876,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2058232347","doi":"10.1016/j.acalib.2014.06.011","title":"Data Sharing and Discovery: What Librarians Need to Know","year":2014,"lang":"en","type":"article","venue":"The Journal of Academic Librarianship","topic":"Research Data Management Practices","field":"Computer Science","cited_by":63,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Calgary","funders":"","keywords":"Data sharing; Scholarly communication; Work (physics); Data management; Computer science; Bridge (graph theory); Knowledge management; Data science; World Wide Web; Public relations; Political science; Publishing; Engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.1681255910920915,"gpt":0.3566202742845169,"spread":0.1884946831924254,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.008379388,0.0001696393,0.0002496847,0.0003015467,0.0001855922,0.009729943,0.0181538,0.00009311194,0.000006630666],"category_scores_gemma":[0.002193214,0.0001091813,0.00003279727,0.0006904204,0.00008698679,0.2556486,0.009282369,0.001238197,0.00001964692],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001799213,"about_ca_system_score_gemma":0.0001312864,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000009299724,"about_ca_topic_score_gemma":0.000002004666,"domain_scores_codex":[0.9972473,0.0006397978,0.0005504958,0.000417636,0.000725119,0.0004196882],"domain_scores_gemma":[0.9952857,0.001504694,0.0004663617,0.002376341,0.0000382664,0.0003285769],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0002459758,0.00004535614,0.007676648,0.00007869252,0.0001797414,0.00003063527,0.00195996,0.0002479335,0.001614363,0.8516715,0.08559506,0.0506541],"study_design_scores_gemma":[0.001479675,0.0006236426,0.01454066,0.0009885154,0.000157304,0.00046105,0.002417481,0.04646063,0.001078219,0.07480623,0.8563249,0.0006617201],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02066506,0.003138588,0.7242929,0.2482145,0.00075953,0.0003140824,0.00001121402,0.00006666133,0.002537405],"genre_scores_gemma":[0.8578446,0.04805346,0.06380869,0.02080047,0.003861216,0.000007834427,0.00002659553,0.0001156352,0.005481513],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8371795,"threshold_uncertainty_score":0.9987304,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2111744339","doi":"10.1186/gm316","title":"Open science versus commercialization: a modern research conflict?","year":2012,"lang":"en","type":"article","venue":"Genome Medicine","topic":"Research Data Management Practices","field":"Computer Science","cited_by":61,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":true},"ca_institutions":"McGill University; University of Alberta","funders":"Economic and Social Research Council; University of Edinburgh; Stem Cell Network","keywords":"Commercialization; Open science; Science policy; Open research; Open innovation; Exploratory research; Public relations; Dissemination; Political science; Engineering ethics; Business; Knowledge management; Sociology; Computer science; Marketing; Engineering; Social science; World Wide Web; Public administration","retraction":null,"screen_n_in":null,"score":{"opus":0.5862541650389435,"gpt":0.5421073726736438,"spread":0.04414679236529961,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.01819687,0.0001141314,0.0001828997,0.0006266842,0.0007590538,0.001891783,0.0139973,0.00002948038,0.0001704289],"category_scores_gemma":[0.002846444,0.0000920036,0.00001214165,0.003102469,0.001021505,0.03170933,0.01127414,0.0003073777,0.00027214],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001957714,"about_ca_system_score_gemma":0.000285962,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000515631,"about_ca_topic_score_gemma":0.00001609081,"domain_scores_codex":[0.9956524,0.0003601091,0.0002458338,0.0005714496,0.002165259,0.001004971],"domain_scores_gemma":[0.9963444,0.0005344872,0.00009022656,0.002051648,0.0004922813,0.0004869605],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00008575017,0.0001572967,0.0005914146,0.00003329623,0.00003225357,0.00001897834,0.006587237,0.00001423811,0.004511119,0.9537349,0.01096608,0.02326747],"study_design_scores_gemma":[0.001627087,0.0003860059,0.01026153,0.0000278297,0.000008192996,0.000007626275,0.0005710954,0.003610494,0.0001574423,0.001121938,0.9820433,0.0001774874],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.00574698,0.004066182,0.5286623,0.05585419,0.002125847,0.001798092,0.000005483295,0.0001717742,0.4015692],"genre_scores_gemma":[0.982419,0.001272929,0.009948576,0.001133039,0.0008870547,0.0001005485,0.00002669202,0.00001748172,0.004194733],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.976672,"threshold_uncertainty_score":0.9991444,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2794631061","doi":"10.1108/s0065-28302018000044b010","title":"Archival Records and Training in the Age of Big Data","year":2018,"lang":"en","type":"book-chapter","venue":"","topic":"Research Data Management Practices","field":"Computer Science","cited_by":60,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Cyberinfrastructure; Archival science; Analytics; Computer science; Data science; Digital humanities; Big data; World Wide Web; Library science; Data mining","retraction":null,"screen_n_in":null,"score":{"opus":0.4102439094970225,"gpt":0.3841935006342329,"spread":0.02605040886278959,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["open_science"],"consensus_categories":[],"category_scores_codex":[0.002840619,0.0001299016,0.000181248,0.0002446605,0.00003763167,0.000878303,0.006953285,0.00003836358,0.00003702487],"category_scores_gemma":[0.0002605827,0.0000856429,0.00001804786,0.00006228959,0.0002031854,0.005016809,0.004950133,0.0003053885,0.0000158564],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000006238844,"about_ca_system_score_gemma":0.00008422874,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001317708,"about_ca_topic_score_gemma":0.0006318319,"domain_scores_codex":[0.9983802,0.00009555829,0.000242043,0.0005776912,0.0005162965,0.000188155],"domain_scores_gemma":[0.9962038,0.0006221876,0.0001410002,0.002983767,0.00001958888,0.00002963385],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000004436372,0.000006109755,0.000005271557,0.00003487417,0.00002543832,0.00007991812,0.0006604985,4.902143e-8,0.000002727613,0.7514772,0.005761807,0.2419417],"study_design_scores_gemma":[0.0001063917,0.0000999637,0.0002661352,0.00007957857,0.000008821536,0.0000101899,0.00008716821,0.0008034828,9.786864e-7,0.05138846,0.9470186,0.0001302317],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"other","genre_gemma":"other","genre_scores_codex":[0.0000157744,0.00005148687,0.07323571,0.001090418,0.0001446387,0.0002200296,0.00003748057,0.00001659628,0.9251879],"genre_scores_gemma":[0.003387421,0.005158465,0.08378876,0.001036708,0.0007900618,0.00001181431,0.0004755465,0.00004101561,0.9053102],"genre_candidate":"other","genre_consensus":"other","teacher_disagreement_score":0.9412568,"threshold_uncertainty_score":0.9984196,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1970555408","doi":"10.1087/095315108x254476","title":"Data, disciplines, and scholarly publishing","year":2007,"lang":"en","type":"article","venue":"Learned Publishing","topic":"Research Data Management Practices","field":"Computer Science","cited_by":60,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"BC Studies","funders":"","keywords":"Scholarly communication; Publishing; Publication; Scholarship; Incentive; Value (mathematics); Computer science; Digital scholarship; Data science; Product (mathematics); World Wide Web; Library science; Political science","retraction":null,"screen_n_in":null,"score":{"opus":0.2012887828853295,"gpt":0.389130009738767,"spread":0.1878412268534375,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.02824921,0.000183811,0.0001721449,0.0006050181,0.0005032025,0.3952757,0.01256779,0.0001480488,0.00001433324],"category_scores_gemma":[0.03705353,0.0001712007,0.00002452386,0.001588532,0.00007811251,0.8248829,0.01643035,0.001641565,0.00002539701],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000676823,"about_ca_system_score_gemma":0.000118934,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0006146767,"about_ca_topic_score_gemma":0.0006030645,"domain_scores_codex":[0.9956216,0.0002952676,0.0004116636,0.001327355,0.001466236,0.0008778362],"domain_scores_gemma":[0.9940715,0.001374225,0.0002648887,0.003532413,0.0003199984,0.0004369638],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00001301515,0.00005448915,0.009344877,0.00003752778,0.00004306373,0.0001372798,0.0005570581,0.000008264162,0.0002642696,0.5586334,0.03374321,0.3971635],"study_design_scores_gemma":[0.0005801617,0.00003118873,0.03381495,0.00004479888,0.00001097593,0.00003356497,0.0007424818,0.008458844,0.00005648804,0.007463438,0.9483863,0.000376843],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.008790635,0.0008164142,0.8251446,0.1017647,0.0005708333,0.0002686754,0.00003397033,0.0005057919,0.06210441],"genre_scores_gemma":[0.6372004,0.0007988013,0.3421172,0.006436897,0.001689236,0.00001731551,0.001024279,0.00008773088,0.01062811],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.914643,"threshold_uncertainty_score":0.9927747,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2154639572","doi":"10.1371/journal.pbio.1001634","title":"Spatially Explicit Data: Stewardship and Ethical Challenges in Science","year":2013,"lang":"en","type":"article","venue":"PLoS Biology","topic":"Research Data Management Practices","field":"Computer Science","cited_by":59,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Stewardship (theology); Scope (computer science); Data sharing; Relation (database); Data science; Engineering ethics; Data management; Replication (statistics); Data curation; Ethical issues; Biology; Knowledge management; Computer science; Political science; Database; Engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.3438626461642547,"gpt":0.4044572490824292,"spread":0.06059460291817453,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002491343,0.00008731151,0.0001361501,0.0002524111,0.00008068029,0.0007239322,0.004486378,0.0000964022,0.0000146447],"category_scores_gemma":[0.002402383,0.00007014068,0.000005562021,0.0003414041,0.0003409163,0.009807931,0.005823846,0.0003923869,0.00008241715],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002073325,"about_ca_system_score_gemma":0.00009527711,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000309056,"about_ca_topic_score_gemma":0.0002071772,"domain_scores_codex":[0.997961,0.0003029886,0.0001662579,0.0008805394,0.0002666213,0.000422531],"domain_scores_gemma":[0.9976091,0.0005803813,0.00005878312,0.001569647,0.0000714683,0.000110671],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.000004470831,0.00008668234,0.005152153,0.00004068518,0.00001315537,0.0000141505,0.0003239219,7.163081e-7,0.009079066,0.9098445,0.000153252,0.07528723],"study_design_scores_gemma":[0.002278433,0.001973649,0.5835863,0.0001796175,0.00002212448,0.00004676614,0.0009243914,0.2046649,0.005635254,0.1232226,0.07586146,0.001604552],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5993197,0.003586709,0.05554999,0.3161001,0.0005290319,0.001825562,0.000026442,0.000335869,0.02272661],"genre_scores_gemma":[0.9824918,0.003895867,0.01294092,0.0005438908,0.00004294924,0.00004491791,0.000007563723,0.000003734069,0.00002837662],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7866219,"threshold_uncertainty_score":0.8336881,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3020737886","doi":"10.1057/s41267-020-00323-z","title":"A new approach to data access and research transparency (DART)","year":2020,"lang":"en","type":"article","venue":"Journal of International Business Studies","topic":"Research Data Management Practices","field":"Computer Science","cited_by":57,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Western University","funders":"","keywords":"Transparency (behavior); Confidentiality; Intellectual property; International business; Business; Public relations; Knowledge management; Computer science; Data science; Internet privacy; Economics; Political science; Computer security; Management; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.6291798837322439,"gpt":0.5274048100589995,"spread":0.1017750736732443,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.002064562,0.00009224141,0.0002011545,0.000372146,0.0001130713,0.002218141,0.007339932,0.00001582484,0.00001153925],"category_scores_gemma":[0.005886938,0.00007007815,0.00001931502,0.001107732,0.00005699036,0.02595068,0.005671079,0.0002697803,0.00001179324],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004350078,"about_ca_system_score_gemma":0.0001895587,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004042035,"about_ca_topic_score_gemma":0.000008153182,"domain_scores_codex":[0.9974949,0.0001221534,0.0003701454,0.0003844762,0.001439194,0.0001890813],"domain_scores_gemma":[0.9970629,0.0003904325,0.0001827599,0.0004507948,0.001723273,0.0001898263],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0004436283,0.0003408027,0.006397691,0.0004245164,0.00173695,0.0003208752,0.005907125,0.000884079,0.0004528205,0.08541264,0.8238357,0.07384323],"study_design_scores_gemma":[0.001998126,0.0003140639,0.09120127,0.0004041267,0.00006082136,0.000181245,0.001918575,0.01533883,0.0001683512,0.00472951,0.8832468,0.0004382805],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0009845989,0.001468989,0.7876223,0.202152,0.0003928299,0.0001559836,0.00001337399,0.00001396235,0.007196011],"genre_scores_gemma":[0.5393091,0.0429614,0.4091433,0.003670141,0.003037236,0.00001842042,0.00002759257,0.00003209138,0.00180071],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.5383245,"threshold_uncertainty_score":0.9988177,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4385617775","doi":"10.1016/j.cub.2023.05.036","title":"Open science","year":2023,"lang":"en","type":"article","venue":"Current Biology","topic":"Research Data Management Practices","field":"Computer Science","cited_by":56,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Carleton University","funders":"","keywords":"Open science; Replication (statistics); Incentive; Data science; Biology; Publishing; Engineering ethics; Data sharing; Computer science; Code (set theory); Political science; Engineering","retraction":null,"screen_n_in":null,"score":{"opus":0.3595992090866648,"gpt":0.5213519444646127,"spread":0.1617527353779479,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication","open_science","insufficient_payload"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.002454505,0.00005848968,0.00007491323,0.0003293383,0.0002083818,0.002137579,0.01456452,0.00001395594,0.00001603884],"category_scores_gemma":[0.0008520156,0.00004721383,0.00001312606,0.002170044,0.0002609407,0.01570917,0.01755487,0.0001078361,0.001299208],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002619916,"about_ca_system_score_gemma":0.0001650773,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002181693,"about_ca_topic_score_gemma":0.000002990966,"domain_scores_codex":[0.9986106,0.00009536702,0.0001045119,0.0005757949,0.000185366,0.0004283299],"domain_scores_gemma":[0.9986388,0.0001330245,0.00005154412,0.001022927,0.00006228063,0.00009142937],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000001000683,0.00001749934,0.001245957,0.000004361366,0.000002320438,0.000002927565,0.00002194828,0.000001155331,0.0009646861,0.81102,0.006932045,0.1797861],"study_design_scores_gemma":[0.0001351205,0.00005306156,0.01011922,0.000005801956,8.586906e-7,0.000001478405,0.00001073919,0.004525945,0.0003976188,0.013007,0.9716481,0.00009503806],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06074411,0.001272226,0.7183657,0.04802655,0.01404348,0.002407352,0.00004561036,0.001982774,0.1531122],"genre_scores_gemma":[0.9752876,0.004829179,0.01418546,0.0006149802,0.0002918527,0.0002388187,0.0001141272,0.00001527552,0.004422721],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9647161,"threshold_uncertainty_score":0.9994784,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2754812078","doi":"10.11646/zootaxa.4322.1.1","title":"Collection management and study of microscope slides: Storage, profiling, deterioration, restoration procedures, and general recommendations","year":2017,"lang":"en","type":"article","venue":"Zootaxa","topic":"Research Data Management Practices","field":"Computer Science","cited_by":56,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"Universität Salzburg; Universitetet i Oslo; West Virginia University; Naturalis Biodiversity Center; Fundação Oswaldo Cruz; U.S. Department of Agriculture","keywords":"Data science; Microscope slide; Profiling (computer programming); Biology; Computer science; Microscope; Pathology; Medicine","retraction":null,"screen_n_in":null,"score":{"opus":0.0633898256145856,"gpt":0.3617825416376064,"spread":0.2983927160230208,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0007100377,0.00009699692,0.0001084416,0.0001808156,0.0007760016,0.002490211,0.000620831,0.00002436139,0.000002489746],"category_scores_gemma":[0.0002694039,0.00009514271,0.000008442578,0.0001518336,0.00005457471,0.01012546,0.0007887369,0.00007676933,0.00000174185],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003546422,"about_ca_system_score_gemma":0.0000385589,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002844963,"about_ca_topic_score_gemma":0.0008841606,"domain_scores_codex":[0.9988336,0.0001383365,0.0002314004,0.0004092853,0.0002484882,0.0001388703],"domain_scores_gemma":[0.9986115,0.0000300569,0.0003725623,0.0008199606,0.0001141615,0.00005176879],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0005284522,0.004522975,0.3686727,0.002350171,0.001056805,0.0001641831,0.01418688,0.000213016,0.01373444,0.1940483,0.05251962,0.3480024],"study_design_scores_gemma":[0.007787228,0.004276783,0.7583058,0.0003327111,0.0002342105,0.00004150725,0.005640568,0.04303596,0.006602406,0.001527124,0.170934,0.00128165],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9498364,0.0000980286,0.04569474,0.001481985,0.0002050473,0.001464329,0.000004849489,0.00004771069,0.001166955],"genre_scores_gemma":[0.9360378,0.0005046481,0.06044954,0.00003233501,0.00003115133,0.0001330744,0.00001246797,0.000008121977,0.002790837],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3896331,"threshold_uncertainty_score":0.9985453,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3112394909","doi":"10.23962/10539/30360","title":"Indigenous Peoples, Data Sovereignty and Self-Determination: Current Realities and Imperatives","year":2020,"lang":"en","type":"article","venue":"The African Journal of Information and Communication (AJIC)","topic":"Research Data Management Practices","field":"Computer Science","cited_by":52,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":true},"ca_institutions":"Centre for International Governance Innovation; University of Ottawa","funders":"Social Sciences and Humanities Research Council of Canada; University of Cape Town; American University in Cairo; International Development Research Centre; University of Johannesburg; University of Ottawa","keywords":"Indigenous; Sovereignty; Political science; State (computer science); Corporate governance; Indigenous rights; Movement (music); Political economy; Sociology; Law; Human rights; Politics; Economics; Management","retraction":null,"screen_n_in":null,"score":{"opus":0.08466041331795772,"gpt":0.3323703453015533,"spread":0.2477099319835956,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":["scholarly_communication"],"category_scores_codex":[0.00146622,0.00007529589,0.0001111437,0.0000928667,0.000340529,0.001844827,0.001902711,0.00001459684,0.000004188952],"category_scores_gemma":[0.0003758268,0.00005223454,0.00001053163,0.0001924238,0.0001044087,0.02669471,0.001701236,0.0002232476,0.00000177606],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001513005,"about_ca_system_score_gemma":0.0001279864,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001346815,"about_ca_topic_score_gemma":0.000006888701,"domain_scores_codex":[0.9988418,0.0003238659,0.00037238,0.00007515209,0.0002935151,0.00009330886],"domain_scores_gemma":[0.9981021,0.0003588757,0.000550887,0.000700585,0.0001874515,0.0001000732],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00005675026,0.00006172138,0.0004710238,0.0001974394,0.00008776721,0.000001283543,0.1669725,0.00001169349,0.000009257823,0.5044166,0.00186097,0.325853],"study_design_scores_gemma":[0.003379852,0.001315958,0.04069963,0.000303777,0.0001716503,0.0006030265,0.1457731,0.2130943,0.0001128409,0.02292083,0.5708222,0.0008028996],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.201139,0.03484501,0.4786018,0.2577272,0.0002417074,0.002030012,0.0002420101,0.0002745548,0.02489861],"genre_scores_gemma":[0.9259248,0.05965663,0.01389273,0.0004606018,0.00002346858,0.000003501656,0.00002665922,0.000002379835,0.000009224455],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7247857,"threshold_uncertainty_score":0.9991913,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2998616672","doi":"10.1139/facets-2019-0012","title":"Envisioning the scientific paper of the future","year":2020,"lang":"en","type":"article","venue":"FACETS","topic":"Research Data Management Practices","field":"Computer Science","cited_by":52,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":true,"about_ca":false},"ca_institutions":"St. Stephen's University; University of Toronto; Carleton University; Okanagan University College; University of British Columbia, Okanagan Campus; University of British Columbia; Canadian Wood Council","funders":"Natural Sciences and Engineering Research Council of Canada; Liber Ero Foundation; Canada Research Chairs","keywords":"Altmetrics; Preprint; Computer science; Data science; Trustworthiness; Scientific progress; Publication; Impact factor; Scientific discovery; Scientific literature; Internet privacy; World Wide Web; Engineering ethics; Political science; Epistemology; Psychology; Law; Engineering; Biology; Cognitive science","retraction":null,"screen_n_in":null,"score":{"opus":0.0637761545101611,"gpt":0.3111888045033636,"spread":0.2474126499932024,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0006868991,0.0000493225,0.00004900178,0.00001815901,0.0003025694,0.001349163,0.003684776,0.00001389693,0.00004099608],"category_scores_gemma":[0.0003085425,0.00002345271,0.00003593384,0.0006185106,0.0001010609,0.005493095,0.001982214,0.0001586853,0.00007554402],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000005627724,"about_ca_system_score_gemma":0.00003998464,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000005235425,"about_ca_topic_score_gemma":0.000005696802,"domain_scores_codex":[0.9987848,0.000149619,0.00009607871,0.00024926,0.0005703422,0.0001498916],"domain_scores_gemma":[0.9987454,0.00009921379,0.00007562443,0.001001183,0.00003725217,0.00004135318],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0000118884,0.00007512875,0.002469723,0.00006575231,0.00007147057,0.00001162558,0.006690879,0.0006699183,0.03561654,0.5840718,0.2686588,0.1015866],"study_design_scores_gemma":[0.00006639086,0.00001429744,0.01083862,0.000006762242,0.000002414059,6.182343e-7,0.0001891653,0.004426948,0.001991624,0.0002181192,0.9822052,0.00003989394],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.09336568,0.001541056,0.08967777,0.7921715,0.002599914,0.001035747,0.00002138562,0.0001570835,0.01942985],"genre_scores_gemma":[0.9944675,0.00004796734,0.001949198,0.002014551,0.00009791046,0.000005008303,0.000001303983,0.00000352876,0.00141304],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9011018,"threshold_uncertainty_score":0.9996876,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4283394500","doi":"10.1186/s40793-022-00425-1","title":"Metadata harmonization–Standards are the key for a better usage of omics data for integrative microbiome analysis","year":2022,"lang":"en","type":"letter","venue":"Environmental Microbiome","topic":"Research Data Management Practices","field":"Computer Science","cited_by":52,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"Horizon 2020 Framework Programme","keywords":"Metadata; Microbiome; Computer science; Interoperability; Harmonization; Data science; Ontology; World Wide Web; Bioinformatics; Biology","retraction":null,"screen_n_in":null,"score":{"opus":0.06388592106245132,"gpt":0.3156390226104956,"spread":0.2517531015480443,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.00274833,0.0005778128,0.0008499245,0.0009001835,0.0005370363,0.002081188,0.01398057,0.0002306692,0.0002290252],"category_scores_gemma":[0.0002600623,0.0004475769,0.0004678629,0.001279831,0.0004657257,0.01051502,0.009971969,0.0009039044,0.000009057168],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000645262,"about_ca_system_score_gemma":0.0001220915,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005101721,"about_ca_topic_score_gemma":0.00002135995,"domain_scores_codex":[0.9951259,0.0005212625,0.0008084947,0.001950284,0.0008812679,0.0007128216],"domain_scores_gemma":[0.9914822,0.001096364,0.001285214,0.006020335,0.00005696294,0.00005894939],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00005758014,0.0001591038,0.00005953362,0.0002267718,0.004502302,0.00002850369,0.0002337061,0.00002772506,0.01571187,0.0001448047,0.9781024,0.00074566],"study_design_scores_gemma":[0.0005482289,0.0001357103,0.00005789941,0.00002016158,0.00139659,0.000008666981,0.0002047633,0.001911951,0.002216278,0.00006319632,0.9929608,0.0004758079],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"dataset","genre_scores_codex":[0.000152141,0.0005691637,0.6124277,0.1788698,0.0003021093,0.002298678,0.2053294,0.00003251462,0.00001853409],"genre_scores_gemma":[0.001436904,0.002002647,0.232779,0.2802555,0.001155932,0.001747789,0.457448,0.000338194,0.02283599],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.3796487,"threshold_uncertainty_score":0.9997976,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2520322156","doi":"10.1111/ropr.12188","title":"Open Data for Science, Policy, and the Public Good","year":2016,"lang":"en","type":"article","venue":"Review of Policy Research","topic":"Research Data Management Practices","field":"Computer Science","cited_by":49,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Centre for Social Innovation","funders":"","keywords":"Open data; Transparency (behavior); Open government; Accountability; Open science; Public policy; Political science; Democracy; Public administration; Government (linguistics); Science policy; Open research; Space (punctuation); Public relations; Public participation; Politics; Computer science; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.5723920137215006,"gpt":0.6100907361421334,"spread":0.03769872242063277,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["metaresearch","scholarly_communication","open_science"],"category_scores_codex":[0.06830912,0.0001097198,0.0003302153,0.000915542,0.0005708306,0.005607627,0.03989347,0.00002212901,0.00001074337],"category_scores_gemma":[0.08634466,0.00005109324,0.00003954159,0.005010204,0.002595444,0.04530368,0.0560265,0.0001730762,0.00003380368],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001340527,"about_ca_system_score_gemma":0.004146072,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001158235,"about_ca_topic_score_gemma":0.00004197735,"domain_scores_codex":[0.9944806,0.001259178,0.0004101316,0.0008948426,0.001966157,0.0009890856],"domain_scores_gemma":[0.9786094,0.008718778,0.000317899,0.01044559,0.001432711,0.000475581],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000007940091,0.00002373533,0.00001072327,0.0008608294,0.00001398572,5.392221e-7,0.000008522537,1.574074e-9,0.00007414034,0.7933471,0.01401014,0.1916423],"study_design_scores_gemma":[0.0008053576,0.0000832083,0.0001156906,0.001127138,0.00000411377,0.00000858943,0.00001056574,0.0001867448,0.00006923258,0.01972296,0.977791,0.00007546307],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"review","genre_scores_codex":[0.00002322794,0.01923223,0.02644162,0.8760934,0.00004139622,0.004458804,0.0001829606,0.00002434733,0.07350197],"genre_scores_gemma":[0.02748884,0.9365776,0.01634707,0.004141587,0.0007081377,0.0007766336,0.00002205182,0.00002987,0.01390819],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.9637808,"threshold_uncertainty_score":0.9954246,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3195755369","doi":"10.1136/bmjopen-2021-049228","title":"Status, use and impact of sharing individual participant data from clinical trials: a scoping review","year":2021,"lang":"en","type":"review","venue":"BMJ Open","topic":"Research Data Management Practices","field":"Computer Science","cited_by":49,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Ottawa Hospital","funders":"Horizon 2020 Framework Programme; Agence Nationale de la Recherche; European Commission","keywords":"Medicine; Data sharing; Citation; Clinical trial; MEDLINE; Data extraction; Data collection; Citation index; Data science; Alternative medicine; World Wide Web; Computer science; Social science","retraction":null,"screen_n_in":null,"score":{"opus":0.9685302552935026,"gpt":0.7475361438016244,"spread":0.2209941114918782,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","scholarly_communication","open_science"],"consensus_categories":["metaresearch","scholarly_communication","open_science"],"category_scores_codex":[0.07680736,0.0003583541,0.005277998,0.0001400724,0.00007488466,0.01207362,0.01602955,0.0001304468,0.0001334648],"category_scores_gemma":[0.09461315,0.0002491684,0.0004088819,0.000699071,0.0000742017,0.03318053,0.07956816,0.0005959713,0.00002848268],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003017907,"about_ca_system_score_gemma":0.003386365,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001980971,"about_ca_topic_score_gemma":0.0001430006,"domain_scores_codex":[0.9819465,0.01070022,0.003846156,0.00197985,0.0009579741,0.0005693394],"domain_scores_gemma":[0.9676876,0.01690272,0.004201251,0.01062523,0.0001403291,0.0004428676],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"systematic_review","study_design_scores_codex":[0.000009025678,0.00009301402,0.000224283,0.05574306,0.002238259,0.0001118817,0.00001755708,1.91502e-7,9.947524e-9,0.0004986055,0.008544802,0.9325193],"study_design_scores_gemma":[0.0003012403,0.00007403776,0.0001491299,0.5768619,0.002402463,0.000007408709,0.000004722562,0.000259764,1.743413e-8,0.00001414364,0.4196338,0.0002914388],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.000001164719,0.9785205,0.00214509,0.00009935739,0.0001420207,0.01741027,0.001592713,0.00001692473,0.00007200249],"genre_scores_gemma":[1.152995e-7,0.9773985,0.01919521,0.00009816935,0.0001125391,0.0006635172,0.002448538,0.00002324731,0.00006015379],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.9322278,"threshold_uncertainty_score":0.9999961,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3110795479","doi":"10.2196/22505","title":"Initiatives, Concepts, and Implementation Practices of FAIR (Findable, Accessible, Interoperable, and Reusable) Data Principles in Health Data Stewardship Practice: Protocol for a Scoping Review","year":2020,"lang":"en","type":"review","venue":"JMIR Research Protocols","topic":"Research Data Management Practices","field":"Computer Science","cited_by":48,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"","funders":"Universität Greifswald; Deutsche Forschungsgemeinschaft","keywords":"Interoperability; Stewardship (theology); Protocol (science); Health data; Knowledge management; Computer science; Data science; Health care; Process management; World Wide Web; Political science; Medicine; Business; Alternative medicine","retraction":null,"screen_n_in":null,"score":{"opus":0.8781572789485774,"gpt":0.7617979020894372,"spread":0.1163593768591402,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","metaepi_narrow","scholarly_communication","open_science"],"consensus_categories":["metaresearch","scholarly_communication","open_science"],"category_scores_codex":[0.04972202,0.0006934291,0.002618142,0.0009605128,0.0004543131,0.005550957,0.01530685,0.0002085483,0.00003590427],"category_scores_gemma":[0.03514064,0.0005634802,0.00006996132,0.003078445,0.0004288669,0.07839058,0.0359917,0.002150533,0.000007113009],"about_ca_system_candidate":true,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003004895,"about_ca_system_score_gemma":0.009692278,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004307283,"about_ca_topic_score_gemma":0.0004620487,"domain_scores_codex":[0.9752818,0.01327915,0.003563608,0.003847116,0.00257681,0.001451506],"domain_scores_gemma":[0.9754764,0.009290415,0.006783158,0.007091409,0.00086889,0.0004897484],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"systematic_review","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0001319217,0.0002790163,0.00002530398,0.5532889,0.0001066126,0.00003046838,0.0001023342,7.831105e-8,1.658528e-7,0.003186634,0.008830209,0.4340184],"study_design_scores_gemma":[0.001184096,0.0009869229,0.000007684367,0.3841111,0.00001851971,0.0000159773,0.000144245,0.0002481955,0.000001025085,0.00002884359,0.6129791,0.0002742627],"study_design_candidate":"systematic_review","study_design_consensus":null,"genre_codex":"protocol","genre_gemma":"protocol","genre_scores_codex":[2.424431e-9,0.2994454,0.002147012,0.004471588,0.000005254225,0.6929425,0.000637644,0.00004636546,0.0003042338],"genre_scores_gemma":[6.795507e-9,0.4748611,0.02023113,0.0001309535,0.00005036623,0.5040587,0.0005887477,0.00004122695,0.00003784054],"genre_candidate":"protocol","genre_consensus":"protocol","teacher_disagreement_score":0.6041489,"threshold_uncertainty_score":0.9996817,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2048623744","doi":"10.1177/0162243912437221","title":"Making an Issue out of a Standard","year":2012,"lang":"en","type":"article","venue":"Science Technology & Human Values","topic":"Research Data Management Practices","field":"Computer Science","cited_by":48,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université du Québec à Montréal","funders":"","keywords":"Institutionalisation; Sociology; Standardization; Storytelling; Narrative; Engineering ethics; Work (physics); Data sharing; Ethnography; Public relations; Political science; Law","retraction":null,"screen_n_in":null,"score":{"opus":0.1666250991205808,"gpt":0.4629193681976315,"spread":0.2962942690770507,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["sts","scholarly_communication","open_science"],"consensus_categories":[],"category_scores_codex":[0.004771335,0.0001279992,0.0001861837,0.001676264,0.0007551373,0.0006802083,0.007167598,0.00007017021,0.00003738232],"category_scores_gemma":[0.0006985626,0.0001144401,0.00002716623,0.002346974,0.0032494,0.03547128,0.002724373,0.0002424365,0.00006247369],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000855413,"about_ca_system_score_gemma":0.0001148926,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000009982571,"about_ca_topic_score_gemma":0.000008038152,"domain_scores_codex":[0.9972933,0.00008513492,0.000258694,0.0005772319,0.0009812933,0.0008043292],"domain_scores_gemma":[0.9973714,0.00004711431,0.0002151695,0.002050092,0.0002062938,0.000109947],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.000002451787,0.00008927369,0.005958959,0.00001332989,0.000008179954,0.000004105792,0.0009869573,0.00000292292,0.04911802,0.9234888,0.0002282731,0.02009875],"study_design_scores_gemma":[0.001191291,0.00333091,0.05357029,0.0002583712,0.00006399001,0.00005843658,0.009252152,0.00395089,0.4079702,0.3004755,0.2182394,0.001638575],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7478268,0.0005506012,0.2117571,0.004007565,0.0009640736,0.0005976043,0.000007654818,0.001374844,0.03291378],"genre_scores_gemma":[0.9399506,0.00002331265,0.05940594,0.00006726309,0.00005071914,0.00001387898,6.507829e-7,0.000006297511,0.0004813769],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6230133,"threshold_uncertainty_score":0.9994632,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4309046817","doi":"10.1038/s41597-022-01815-3","title":"Modeling community standards for metadata as templates makes data FAIR","year":2022,"lang":"en","type":"article","venue":"Scientific Data","topic":"Research Data Management Practices","field":"Computer Science","cited_by":46,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"","funders":"Common Fund; U.S. National Library of Medicine; National Science Foundation; Wellcome; Canadian Institutes of Health Research; Austrian Science Fund; Wellcome Trust; National Institute of General Medical Sciences; National Cancer Institute; Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Institutes of Health; U.S. Department of Health and Human Services; NIH Office of the Director","keywords":"Metadata; Template; Computer science; Geospatial metadata; Meta Data Services; Information retrieval; World Wide Web; Data science; Data element; Programming language","retraction":null,"screen_n_in":null,"score":{"opus":0.4877564196382878,"gpt":0.4680528498693035,"spread":0.0197035697689843,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","sts","scholarly_communication","open_science"],"consensus_categories":["scholarly_communication","open_science"],"category_scores_codex":[0.04114348,0.0001889104,0.0002245314,0.0003696023,0.004446315,0.01843311,0.06754351,0.00002213197,0.0001333084],"category_scores_gemma":[0.005257171,0.0001843636,0.00003293527,0.001370461,0.0002040726,0.09434966,0.1283229,0.0006238261,0.00003013908],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001325495,"about_ca_system_score_gemma":0.0008475233,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001451117,"about_ca_topic_score_gemma":0.0008167194,"domain_scores_codex":[0.9933884,0.0009660429,0.0004238989,0.001975572,0.002578682,0.0006674499],"domain_scores_gemma":[0.9673089,0.0007093287,0.0001747032,0.03136464,0.0002723713,0.0001700287],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00007264615,0.0004976807,0.00002935288,0.0001751936,0.0002149929,0.00003237092,0.0005315779,0.003462096,0.0004763448,0.07662398,0.8901795,0.02770429],"study_design_scores_gemma":[0.000183401,0.00004031999,0.000002706376,0.000005881872,0.00001852394,0.00000685752,0.0007483257,0.4794274,0.00003606458,0.002658882,0.5167375,0.0001342104],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"methods","genre_gemma":"dataset","genre_scores_codex":[0.001362548,0.0003769122,0.8549292,0.002855775,0.001519195,0.0007873322,0.1368914,0.0002082188,0.001069434],"genre_scores_gemma":[0.2878486,0.0002187943,0.3066527,0.0008876339,0.0001928414,0.0003191733,0.3794333,0.00008968855,0.02435727],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.5482765,"threshold_uncertainty_score":0.9968498,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}